CN101059909A - All-round computer vision-based electronic parking guidance system - Google Patents

Abstract

Translated fromChinese

一种基于全方位计算机视觉的电子泊车诱导系统，该电子泊车诱导系统包括微处理器、用于监测停车场内停放车辆情况的全方位视觉传感器、用于与外界通信的通信模块，所述的全方位视觉传感器安装在待监测的停车场的上部；微处理器检测每一个车位是否被占用，为停车提供动态的场内诱导以及场外诱导信息。本发明的车位检测方法具有检测范围广、安装维护无干扰、维护方便低耗、检测参数丰富、可视性、检测可靠性、准确度高、统计计算方便、算法实现简单、具有良好的先进性、可扩展性、可持续发展性，给停车场配备一双智能化的慧眼。

An electronic parking guidance system based on all-round computer vision, the electronic parking guidance system includes a microprocessor, an all-round visual sensor for monitoring the situation of parked vehicles in a parking lot, and a communication module for communicating with the outside world, the The omnidirectional vision sensor mentioned above is installed on the upper part of the parking lot to be monitored; the microprocessor detects whether each parking space is occupied, and provides dynamic on-site guidance and off-site guidance information for parking. The parking space detection method of the present invention has the advantages of wide detection range, no interference in installation and maintenance, convenient maintenance and low consumption, rich detection parameters, visibility, detection reliability, high accuracy, convenient statistical calculation, simple algorithm implementation, and good advancement , scalability, and sustainable development, equip the parking lot with a pair of intelligent eyes.

Description

Translated fromChinese

基于全方位计算机视觉的电子泊车诱导系统Electronic parking guidance system based on omnidirectional computer vision

(一)技术领域(1) Technical field

本发明属于全方位计算机视觉传感器技术、图像识别技术、数据库技术和网络通信技术在停车诱导系统方面的应用，尤其是一种电子泊车诱导系统。The invention belongs to the application of omnidirectional computer vision sensor technology, image recognition technology, database technology and network communication technology in parking guidance system, especially an electronic parking guidance system.

(二)背景技术(2) Background technology

城市道路交通由动态交通和静态交通两部分所组成，静态交通是指车辆为完成不同的出行目的或保管而产生的在不同区域、不同停放场所的停放状态。静态交通与动态交通一样，是城市交通中不可分割的组成部分，动态交通以静态交通为起点，静态交通是动态交通的延续。动态交通与静态交通既相互促进又相互制约，需要协调发展共同构成城市交通系统。Urban road traffic is composed of dynamic traffic and static traffic. Static traffic refers to the parking status of vehicles in different areas and different parking places for different travel purposes or storage. Static traffic, like dynamic traffic, is an integral part of urban traffic. Dynamic traffic starts with static traffic, and static traffic is the continuation of dynamic traffic. Dynamic traffic and static traffic both promote and restrict each other, and need coordinated development to form an urban traffic system.

目前的城市道路交通管理时，人们往往只重视对城市动态交通的疏导与控制，忽视了对车辆停放等静态交通的规划、建设与管理，忽视了对车辆停放的诱导，使得城市交通拥挤、堵塞、事故频发等老大难问题更显突出，人们费尽心机采取很多治理措施却力不从心，收效甚微。导致这一问题的一个主要症结是城市停车场库的建设不足、管理不善以及缺乏先进的泊车诱导系统，造成了为了寻找车位而盲目流动的时间，使得违章停车和占用道路等现象时有发生。In the current urban road traffic management, people often only pay attention to the guidance and control of urban dynamic traffic, ignoring the planning, construction and management of static traffic such as vehicle parking, and ignoring the induction of vehicle parking, which makes urban traffic congestion and congestion The long-standing and difficult problems such as frequent accidents and frequent accidents have become more prominent. People have tried their best to take many control measures, but they are unable to do what they want, with little effect. One of the main crux of this problem is the insufficient construction of urban parking garages, poor management, and the lack of advanced parking guidance systems, resulting in the blind flow of time in order to find parking spaces, making illegal parking and road occupation occur from time to time .

国内外调查数据表明，驾驶员由于不了解停车位，随意、任意、没有目的地去找停车位，给道路交通增加了额外的负担，城市路面车流中大约有12^15％左右的车辆是正在寻找停车泊位的车辆。另有国外报道说，在巴黎为了在市区找停车位耗费的汽油，占整个行车汽油的40％，增加了车辆尾气的排放，大幅度增加了汽车尾气造成的环境污染。Survey data at home and abroad show that because drivers do not know about parking spaces, they search for parking spaces at will, randomly, and without destination, which adds an extra burden to road traffic. About 12^15% of the vehicles in the urban road traffic flow are on the road. Vehicles looking for parking spaces. Another foreign report said that in Paris, the gasoline consumed in order to find a parking space in the urban area accounts for 40% of the entire driving gasoline, which increases the emission of vehicle exhaust and greatly increases the environmental pollution caused by automobile exhaust.

目前我国一方面停车场数量不能适应汽车保有量的增长，再加上驾驶员盲目寻找停车场，导致了违章占路停车现象增多，从而大大降低了道路通行能力，容易引发交通拥堵和交通事故，这种矛盾正日显突出。At present, on the one hand, the number of parking lots in our country cannot adapt to the growth of car ownership. Coupled with the fact that drivers are blindly looking for parking lots, the phenomenon of illegal parking on the road has increased, which has greatly reduced road traffic capacity and easily caused traffic congestion and traffic accidents. This contradiction is becoming increasingly prominent.

停车场库是城市静态交通的最重要条件，这方面的建设与管理的不足，直接影响到动态交通的正常运行，而动态交通的不畅，同时又反过来对静态交通的管理产生影响。产生了交通困难的恶性循环，使得城市道路交通的安全、畅通、有序无法得到保障，正严重地阻碍着城市及其经济的发展。The parking garage is the most important condition for urban static traffic. The lack of construction and management in this area directly affects the normal operation of dynamic traffic, and the poor dynamic traffic also affects the management of static traffic. A vicious cycle of traffic difficulties has been produced, making the safety, smoothness and order of urban road traffic unable to be guaranteed, which is seriously hindering the development of cities and their economies.

目前我国同时存在着城市机动车拥有量和停车设施间比例相差悬殊和停车场利用率低的问题，一方面表现为停车场内车位闲置，资源浪费；另一方面，大量外地车辆和部分本地车辆由于驾驶员不了解停车场停车泊位状况，因此花费驾驶员很长的时间寻找有停车泊位的停车场。这不仅增加了城市道路负荷，严重影响道路的动态交通，大幅度增加了汽车尾气造成的环境污染。国外的实际情况表明，只有通过有效的停车诱导信息系统才会改善汽车盲目寻找停车场的局面，进而减少交通事故和降低空气污染。先进的停车诱导和信息系统关键理论和实施技术研究是智能交通的重要研究内容，也是国际上正在深入研究的前沿课题之一，是我国城市交通亟待解决的关键问题。At present, there are also problems in our country such as the disparity between the number of urban motor vehicles and the proportion of parking facilities and the low utilization rate of parking lots. On the one hand, the parking spaces in the parking lot are idle and resources are wasted; Because the driver does not understand the parking space situation in the parking lot, it takes a long time for the driver to look for a parking lot with a parking space. This not only increases the urban road load, seriously affects the dynamic traffic of the road, but also greatly increases the environmental pollution caused by automobile exhaust. The actual situation in foreign countries shows that only through an effective parking guidance information system can the situation of cars blindly looking for parking lots be improved, thereby reducing traffic accidents and reducing air pollution. Research on the key theory and implementation technology of advanced parking guidance and information system is an important research content of intelligent transportation, and it is also one of the frontier topics being studied in depth in the world.

要解决停车问题的途径主要可以从以下三个方面着手：The way to solve the parking problem can mainly start from the following three aspects:

(1)合理规划和发展静态交通基础设施；(1) Rationally plan and develop static traffic infrastructure;

(2)采用先进的管理手段对停车进行管理和控制；(2) Adopt advanced management methods to manage and control parking;

(3)实施智能交通运输系统。(3) Implement intelligent transportation system.

静态交通基础设施主要包括社会停车场、市区外围出入口的公共停车场、配建停车场以及对停车进行科学管理所需配备的检测器、可变信息显示装置、网络设备等。Static traffic infrastructure mainly includes social parking lots, public parking lots at the entrances and exits of urban areas, auxiliary parking lots, and detectors, variable information display devices, and network equipment required for scientific management of parking.

先进的停车诱导信息系统要利用地区统一的可变电子标志牌为驾驶员提供不同的动态停车诱导信息，包括停车场位置、停车库位置、路边停车位置、驾驶员预先选择的停车泊位以及最佳行驶路线等信息。利用这些信息，引导驾驶员避开拥挤并快速找到较为理想的停车泊位。The advanced parking guidance information system should use the regional unified variable electronic signs to provide drivers with different dynamic parking guidance information, including parking lot location, parking garage location, roadside parking location, driver's pre-selected parking space and the most Best driving route and other information. Use this information to guide drivers to avoid congestion and quickly find an ideal parking space.

因此先进的停车诱导信息系统必须由停车信息采集、信息处理、信息传输以及信息发布等四部分组成，其各部分的作用如下：1)信息采集系统，系统通过远程监视装置、传感装置，采集对象区域内各停车场相关信息，主要包括了停车场的车位使用状况等信息；2)信息处理系统，系统将采集到的停车场使用状况以及周边道路信息加工处理成向驾驶员提供的适当形式的信息，如停车场的满空(剩余车位情况)、集散道路是否拥堵等。另外，信息处理系统还担负着存储停车场信息、加工处理停车场使用情况的变化模式等任务。这些功能将为未来提供停车需求状况预报、停车位预约等服务奠定基础；3)信息传输系统，信息传输的基本任务是保证从信息采集系统到信息处理系统再到信息发布系统的畅通。其常用的形式有光传输网、电话交换网以及光接入网等形式；4)信息发布系统，系统的任务是将信息处理系统处理过的信息，以适当的方式向外界分若干个层次发布出来。通常是由控制中心，随时将各个停车场的使用状况在可变信息显示板上以视觉的方式或通过广播以听觉的方式向驾驶员提供，也可以利用互联网、移动电话以及车载导航装置等方式发布。目前最为基础、最为常用的发布形式为设置于路侧的诱导信息板。Therefore, an advanced parking guidance information system must be composed of four parts: parking information collection, information processing, information transmission, and information release. The functions of each part are as follows: 1) Information collection system. The relevant information of each parking lot in the target area mainly includes information such as the parking space usage status of the parking lot; 2) the information processing system, which processes the collected parking lot usage status and surrounding road information into an appropriate form for the driver information, such as the fullness of the parking lot (remaining parking spaces), whether the collection and distribution roads are congested, etc. In addition, the information processing system is also responsible for storing parking lot information, processing and processing the changing patterns of parking lot usage. These functions will lay the foundation for providing services such as parking demand forecast and parking space reservation in the future; 3) Information transmission system, the basic task of information transmission is to ensure the smooth flow from the information collection system to the information processing system and then to the information release system. Its commonly used forms include optical transmission network, telephone switching network, and optical access network; 4) Information release system, the task of the system is to release the information processed by the information processing system to the outside world in several levels in an appropriate way come out. Usually, the control center will provide the driver with the usage status of each parking lot on the variable information display board visually or audibly through the broadcast at any time, and can also use the Internet, mobile phones, and car navigation devices. release. At present, the most basic and commonly used form of release is the guidance information board set up on the roadside.

在本发明做出以前要检测停车场内的车位使用状况是通过埋设在车位前的地感线圈通过电磁感应方式获得该车位是否被占用的信息，这种方式虽然能比较好地检测出车位的使用状况，但是存在着当地感线圈故障时，需要封道挖开路面维修，增加了维修人员的维护工作量，增加了维护成本，同时对容量较大的停车场要对每一个车位埋设地感线圈一次性投资比较大，且车位越多也会带来检测的复杂性以及通信和计算的压力。Before the present invention is made, to detect the parking space usage status in the parking lot is to obtain the information of whether the parking space is occupied by the ground induction coil buried in front of the parking space through electromagnetic induction, although this method can better detect the parking space. However, when there is a failure of the local induction coil, the road needs to be closed and dug for maintenance, which increases the maintenance workload of the maintenance personnel and increases the maintenance cost. The one-time investment of the coil is relatively large, and the more parking spaces will bring the complexity of detection and the pressure of communication and calculation.

(三)发明内容(3) Contents of the invention

为了克服已有的泊车诱导系统的成本高、维护费用高、可靠性差的不足，本发明提供一种成本低、维护费用少、可靠性好的基于全方位计算机视觉的电子泊车诱导系统。In order to overcome the disadvantages of high cost, high maintenance cost and poor reliability of the existing parking guidance system, the present invention provides an electronic parking guidance system based on omnidirectional computer vision with low cost, low maintenance cost and high reliability.

本发明解决其技术问题所采用的技术方案是：The technical solution adopted by the present invention to solve its technical problems is:

一种基于全方位计算机视觉的电子泊车诱导系统，该电子泊车诱导系统包括微处理器、用于监测停车场内停放车辆情况的全方位视觉传感器、用于与外界通信的通信模块，所述的全方位视觉传感器安装在待监测的停车场的上部；An electronic parking guidance system based on all-round computer vision, the electronic parking guidance system includes a microprocessor, an all-round visual sensor for monitoring the situation of parked vehicles in a parking lot, and a communication module for communicating with the outside world, the The omnidirectional vision sensor described above is installed on the top of the parking lot to be monitored;

所述的全方位视觉传感器包括用于反射监控领域中物体的外凸反射镜面、透明圆柱体、摄像头，所述的外凸反射镜面朝下，所述的透明圆柱体支撑外凸反射镜面，黑色圆锥体固定在折反射镜面外凸部的中心，用于拍摄外凸反射镜面上成像体的摄像头位于透明圆柱体的内部，摄像头位于外凸反射镜面的虚焦点上；The omni-directional vision sensor includes a convex reflective mirror used to reflect objects in the monitoring field, a transparent cylinder, and a camera. The convex reflective mirror faces downward, and the transparent cylinder supports the convex reflective mirror, black The cone is fixed at the center of the convex portion of the catadioptric mirror, the camera for shooting the imaging body on the convex mirror is located inside the transparent cylinder, and the camera is located on the virtual focal point of the convex mirror;

所述的微处理器包括：Described microprocessor comprises:

图像数据读取模块，用于读取从视觉传感器传过来的停车场内车位的图像信息；The image data reading module is used to read the image information of the parking spaces in the parking lot transmitted from the visual sensor;

图像数据文件存储模块，用于将读进来的图像信息通过文件方式保存在存储单元中；The image data file storage module is used to save the read image information in the storage unit by means of a file;

虚拟车位框设置模块，用于将读取的整个停车场的图像信息，按照车位分布情况设置与实际车位一一对应的虚拟车位框，并保存基准参考图像；The virtual parking space frame setting module is used to set the virtual parking space frame corresponding to the actual parking space one-to-one according to the parking space distribution of the image information of the entire parking lot read, and save the benchmark reference image;

传感器标定模块，用于对全方位视觉传感器的参数进行标定，建立空间的实物图像与所获得的视频图像的线性对应关系；The sensor calibration module is used to calibrate the parameters of the omnidirectional visual sensor, and establish the linear correspondence between the physical image of the space and the obtained video image;

虚拟车框检测模块，用于对各个虚拟车位框，将所获得的当前帧现场视频图像与基准参考图像进行差值运算，图像相减的计算公式如式(17)表示：The virtual car frame detection module is used for each virtual parking space frame, and the obtained current frame live video image and the benchmark reference image are carried out difference calculation, and the calculation formula of image subtraction is expressed as formula (17):

f_d(X，t₀，t_i)＝f(X，t_i)-f(X，t₀)                (17)f_d (X, t₀ , t_i )=f(X, t_i )-f(X, t₀ ) (17)

上式中，f_d(X，t₀，t_i)是实时拍摄到图像与基准参考图像间进行图像相减的结果；f(X，t_i)是实时拍摄到图像；f(X，t₀)是基准参考图像；In the above formula, f_d (X, t₀ , t_i ) is the result of image subtraction between the image captured in real time and the benchmark reference image; f(X, t_i ) is the image captured in real time; f(X, t₀ ) is the benchmark reference image;

如f_d(X，t₀，t_i)≥阈值时，判定为可疑有车事件；If f_d (X, t₀ , t_i )≥threshold, it is judged as a suspicious car-occupancy event;

如f_d(X，t₀，t_i)＜阈值时，判定为无可疑有车事件；If f_d (X, t₀ , t_i )<threshold value, it is determined that there is no suspicious vehicle occurrence;

连通区域计算模块，用于在判定有可疑有车事件后，对当前图像进行标记，像素灰度为0的小区表示此小区无可疑有车，像素灰度为1则表示此小区有可疑有车，计算当前图像中的像素是否与当前像素周围相邻的某一个点的像素相等，如灰度相等判断为具有连通性，将所有具有连通性的像素作为一个连通区域；The connected area calculation module is used to mark the current image after determining that there is a suspicious car event. A cell with a pixel grayscale of 0 means that there is no suspicious car in this cell, and a pixel grayscale of 1 means that there is a suspicious car in this cell , to calculate whether the pixel in the current image is equal to the pixel of a certain point adjacent to the current pixel, if the gray level is equal, it is judged to have connectivity, and all the connected pixels are regarded as a connected region;

车辆判断模块，用于根据得到的连通区域，统计求出连通区域面积Si，并将连通区域面积与预设的阈值做比较：The vehicle judging module is used to statistically calculate the area Si of the connected area according to the obtained connected area, and compare the area of the connected area with a preset threshold:

如连通区域面积Si大于阈值S_阈，判定为在该车位上有车；If the area Si of the connected area is greater than the threshold S_threshold , it is determined that there is a car in the parking space;

如连通区域面积Si小于阈值S_阈，判定为在该车位上没有车；If the area Si of the connected area is smaller than the threshold S_threshold , it is determined that there is no car in the parking space;

车位信息发布模块，用于根据各个虚拟车位框中的车辆判断，得到停车场内的车位占用信息，通过通信模块发布车位占用信息。The parking space information publishing module is used to obtain the parking space occupancy information in the parking lot according to the judgment of the vehicles in each virtual parking space frame, and release the parking space occupancy information through the communication module.

进一步，所述的微处理器还包括：Further, the microprocessor also includes:

车位信息更新模块，用于根据当前监测的车位信息，跟上一次统计的车位信息做比较，如果车位占用信息有变化，更新发布车位占用信息。The parking space information update module is used to compare the current monitored parking space information with the last statistical parking space information, and update and release the parking space occupancy information if there is a change in the parking space occupancy information.

再进一步，所述的微处理器还包括：Further, the microprocessor also includes:

车位预约处理模块，用于管理人员根据预约车位情况预先设置车位预约情况，并将预约信息输入到车位信息发布模块，综合判断车位占用信息。The parking space reservation processing module is used for management personnel to pre-set the parking space reservation situation according to the reserved parking space situation, and input the reservation information into the parking space information release module to comprehensively judge the parking space occupancy information.

更进一步，所述的微处理器还包括：Furthermore, the microprocessor also includes:

颜色空间转化处理模块，用于将图像数据读取模块采集的图像从RGB色彩空间转化到HSI色彩空间，转换的计算式为(18)：The color space conversion processing module is used to convert the image collected by the image data reading module from the RGB color space to the HSI color space, and the conversion calculation formula is (18):

$\mathrm{<span><span>I</span>I</span>}<span><span>=</span>=</span>\frac{\mathrm{<span><span>R</span>R</span>}<span><span>+</span>+</span>\mathrm{<span><span>G</span>G</span>}<span><span>+</span>+</span>\mathrm{<span><span>B</span><figure-callout\; label="B"\; filenames="A20061005047100271.png,A20061005047100301.png"\; state="\{\{state\}\}">B</figure-callout></span>}}{\mathrm{<span><span>3</span>3</span>}}$

$\mathrm{<span><span>H</span>h</span>}<span><span>=</span>=</span>\frac{\mathrm{<span><span>1</span>1</span>}}{\mathrm{<span><span>360</span>360</span>}}<span><span>[</span>[</span>\mathrm{<span><span>90</span>90</span>}<span><span>-</span>-</span>\mathrm{<span><span>Arc</span>Arc</span>}\mathrm{<span><span>tan</span>the\; tan</span>}<span><span>(</span>(</span>\frac{\mathrm{<span><span>F</span>f</span>}}{\sqrt{\mathrm{<span><span>3</span>3</span>}}}<span><span>)</span>)</span><span><span>+</span>+</span><span><span>\{</span>\{</span>\mathrm{<span><span>0</span>0</span>}<span><span>,</span>,</span>\mathrm{<span><span>G</span>G</span>}<span><span>></span>></span>\mathrm{<span><span>B</span>B</span>}<span><span>;</span>;</span>\mathrm{<span><span>180</span>180</span>}<span><span>,</span>,</span>\mathrm{<span><span>G</span>G</span>}<span><span><</span><</span>\mathrm{<span><span>B</span>B</span>}<span><span>\}</span>\}</span><span><span>]</span>]</span>$

$\mathrm{<span><span>S</span>S</span>}<span><span>=</span>=</span>\mathrm{<span><span>1</span>1</span>}<span><span>-</span>-</span><span><span>\{</span>\{</span>\frac{\mathrm{<span><span>min</span>min</span>}<span><span>(</span>(</span>\mathrm{<span><span>R</span>R</span>}<span><span>,</span>,</span>\mathrm{<span><span>G</span>G</span>}<span><span>,</span>,</span>\mathrm{<span><span>B</span>B</span>}<span><span>)</span>)</span>}{\mathrm{<span><span>I</span>I</span>}}<span><span>\}</span>\}</span><span><span>-</span>-</span><span><span>-</span>-</span><span><span>-</span>-</span><span><span>(</span>(</span>\mathrm{<span><span>18</span>18</span>}<span><span>)</span>)</span>$

其中，$F=\frac{2R-G-B}{G-B}$in,$f=\frac{2R-G-B}{G-B}$

上式中，H为HSI色彩空间的色调，S为HSI色彩空间的饱和度，I为HSI色彩空间的亮度，R为RGB色彩空间的红色；G为RGB色彩空间的绿色；B为RGB色彩空间的蓝色；In the above formula, H is the hue of the HSI color space, S is the saturation of the HSI color space, I is the brightness of the HSI color space, R is the red of the RGB color space; G is the green of the RGB color space; B is the RGB color space the blue;

颜色空间转化处理模块的输入端连接所述的虚拟车框检测模块。The input end of the color space conversion processing module is connected to the virtual vehicle frame detection module.

或者是，所述的微处理器还包括：Or, the microprocessor also includes:

颜色空间转化处理模块，用于将图像数据读取模块采集的图像从RGB色彩空间转化到(Cr，Cb)空间颜色模型，转换的计算式为(19)：The color space conversion processing module is used to convert the image collected by the image data reading module from the RGB color space to the (Cr, Cb) space color model, and the calculation formula for conversion is (19):

Y＝0.29990*R+0.5870*G+0.1140*BY＝0.29990*R+0.5870*G+0.1140*B

Cr＝0.5000*R-0.4187*G-0.0813*B+128Cr＝0.5000*R-0.4187*G-0.0813*B+128

Cb＝-0.1787*R-0.3313*G+0.5000*B+128      (19)Cb＝-0.1787*R-0.3313*G+0.5000*B+128 (19)

上式中，Y代表(Cr，Cb)空间颜色模型的亮度，Cr、Cb是(Cr，Cb)空间颜色模型的两个彩色分量，表示色差；R表示RGB色彩空间的红色；G表示RGB色彩空间的绿色；B表示RGB色彩空间的蓝色。In the above formula, Y represents the brightness of the (Cr, Cb) space color model, Cr and Cb are the two color components of the (Cr, Cb) space color model, representing the color difference; R represents the red color of the RGB color space; G represents the RGB color The green of the space; B represents the blue of the RGB color space.

本发明的工作原理是：图像处理与计算机视觉是一个不断发展的新技术，原则上采用计算机视觉进行观测有四个目的，即预处理、最底层的特征提取、中级特征的辩识以及通过图像对高级情景的解释。一般来说，计算机视觉包括主要特征、图像处理以及图像理解。图像是人类视觉的延伸。通过机器视觉，可以立即准确地把握停车场内的车位使用状况。图像检测快速性的基础是视觉所接受的信息以光为传播媒介；而图像信息的丰富和直观，是其它目前各种探测技术均不能提供如此丰富和直观的信息。The working principle of the present invention is: image processing and computer vision are a new technology that is constantly developing. In principle, there are four purposes for using computer vision to observe, namely, preprocessing, feature extraction at the lowest level, identification of intermediate features, and identification of intermediate features through images. Explanation of advanced scenarios. In general, computer vision includes main features, image processing, and image understanding. Images are an extension of human vision. Through machine vision, it is possible to immediately and accurately grasp the usage status of parking spaces in the parking lot. The basis of the rapidity of image detection is that the information received by vision uses light as the medium of transmission; and the richness and intuition of image information is that other current detection technologies cannot provide such rich and intuitive information.

近年发展起来的全方位视觉传感器ODVS(OmniDirectional Vision Sensors)为实时获取场景的全景图像提供了一种新的解决方案。ODVS的特点是视野广(360度)，能把一个半球视野中的信息压缩成一幅图像，一幅图像的信息量更大；获取一个场景图像时，ODVS在场景中的安放位置更加自由；监视环境时ODVS不用瞄准目标；检测和跟踪监视范围内的运动物体时算法更加简单；可以获得场景的实时图像。因此基于ODVS的全方位视觉系统近几年迅速发展，正成为计算机视觉研究中的重要领域，IEEE从2000年开始举办每年一次的全方位视觉的专门研讨会(IEEE workshop on Omni-directional vision)。由于在停车场中场内车位检测需要覆盖所有的车位，因此利用全方位视觉传感器可以随时检测每一个车位，只要将全方位视觉传感器安装在停车场顶部的中间就非常容易地把握停车场内的车位使用状况，目前还没有检索到将全方位视觉传感器运用到电子泊车诱导系统技术领域的论文与专利。OmniDirectional Vision Sensors (ODVS), developed in recent years, provides a new solution for obtaining panoramic images of scenes in real time. ODVS is characterized by a wide field of view (360 degrees), which can compress the information in a hemispheric field of view into one image, and the amount of information in one image is larger; when acquiring a scene image, the placement position of ODVS in the scene is more free; monitoring ODVS does not need to aim at the target in the environment; the algorithm is simpler when detecting and tracking moving objects within the monitoring range; real-time images of the scene can be obtained. Therefore, the omni-directional vision system based on ODVS has developed rapidly in recent years and is becoming an important field in computer vision research. Since 2000, IEEE has held an annual special seminar on omni-directional vision (IEEE workshop on Omni-directional vision). Since the detection of parking spaces in the parking lot needs to cover all the parking spaces, each parking space can be detected at any time by using the omnidirectional vision sensor. As long as the omnidirectional vision sensor is installed in the middle of the top of the parking lot, it is very easy to grasp the parking space in the parking lot. Parking space usage status, no papers or patents have been retrieved that apply omnidirectional vision sensors to the technical field of electronic parking guidance systems.

因此，采用全方位视觉传感器ODVS并利用数字图像处理技术，结合停车场的车位分布以及停放车辆的一些特征，检测每一个车位是否被占用，为停车提供场内诱导以及场外诱导信息；在以车位检测为重点的同时又能监控停车场中安全，给停车场配备一双智能化的慧眼。Therefore, the omnidirectional visual sensor ODVS is used and digital image processing technology is used to detect whether each parking space is occupied by combining the parking space distribution of the parking lot and some characteristics of the parked vehicles, and to provide parking guidance and off-site guidance information; in the following While focusing on parking space detection, it can also monitor the safety of the parking lot, and equip the parking lot with a pair of intelligent eyes.

ODVS摄像装置的光学部分的制造技术方案，ODVS摄像装置主要由垂直向下的折反射镜和面向上的摄像头所构成。具体构成是由聚光透镜以及CCD构成的摄像单元固定在由透明树脂或者玻璃制的圆筒体的下部，圆筒体的上部固定有一个向下的大曲率的折反射镜，在折反射镜和聚光透镜之间有一根直径逐渐变小的圆锥状体，该圆锥状体固定在折反射镜的中部，圆锥状体的目的是为了防止过剩的光射入而导致在圆筒体内部的光饱和现象。图2是表示本发明的全方位的视觉传感器的光学系统的原理图。The manufacturing technical scheme of the optical part of the ODVS camera device, the ODVS camera device is mainly composed of a vertically downward refracting mirror and an upward-facing camera. The specific structure is that the camera unit consisting of a condenser lens and a CCD is fixed on the lower part of a cylindrical body made of transparent resin or glass, and a downward large-curvature catadioptric mirror is fixed on the upper part of the cylindrical body. There is a conical body with gradually smaller diameter between it and the condenser lens, which is fixed in the middle of the catadioptric mirror. The purpose of the conical body is to prevent excessive light from entering and causing the light saturation phenomenon. Fig. 2 is a schematic diagram showing the optical system of the omnidirectional vision sensor of the present invention.

折反射全景成像系统能用针孔成像模型进行成像分析，但要获得透视全景图像必须对采集的实景图像逆投影，必须满足实时性的要求。The catadioptric panoramic imaging system can use the pinhole imaging model for imaging analysis, but in order to obtain the perspective panoramic image, it must back-project the collected real-scene images, which must meet the real-time requirements.

停车场的场景中物点的水平坐标与相应像点的坐标成线性关系就能确保水平场景无畸变，作为电子泊车诱导系统的全方位视觉装置安装在停车场顶部，监视着整个停车场中水平方向上的车位情况，因此在设计全方位视觉装置的折反射镜面时要保证在水平方向上的不变形。The horizontal coordinates of the object points in the scene of the parking lot are in a linear relationship with the coordinates of the corresponding image points to ensure that the horizontal scene has no distortion. As an electronic parking guidance system, the omnidirectional visual device is installed on the top of the parking lot to monitor the entire parking lot. The parking space in the horizontal direction, so when designing the catadioptric mirror surface of the omnidirectional visual device, it is necessary to ensure that it does not deform in the horizontal direction.

设计中首先选用CCD(CMOS)器件和成像透镜构成摄像头，在对摄像头内部参数标定的基础上初步估算系统外形尺寸，然后根据高度方向的视场确定反射镜面形参数。In the design, a CCD (CMOS) device and an imaging lens are first selected to form a camera, and the overall dimensions of the system are initially estimated on the basis of calibration of the internal parameters of the camera, and then the surface parameters of the mirror are determined according to the field of view in the height direction.

如图1所示，摄像头的投影中心C在道路水平场景上方距离水平场景h处，反射镜的顶点在投影中心上方，距离投影中心zo处。本发明中以摄像头投影中心为坐标原点建立坐标系，反射镜的面形用z(X)函数表示。在像平面内距离像中心点ρ的像素q接受了来自水平场景O点(距离Z轴d)，在反射镜M点反射的光线。水平场景无畸变要求场景物点的水平坐标与相应像点的坐标成线性关系；As shown in Figure 1, the projection center C of the camera is above the horizontal scene of the road at a distance h from the horizontal scene, and the apex of the mirror is above the projection center and at a distance zo from the projection center. In the present invention, the coordinate system is established with the camera projection center as the coordinate origin, and the surface shape of the mirror is represented by the z(X) function. The pixel q at a distance from the image center point ρ in the image plane receives the light from the horizontal scene O point (distance Z axis d) and reflected at the mirror M point. No distortion in the horizontal scene requires that the horizontal coordinates of the scene object points have a linear relationship with the coordinates of the corresponding image points;

d(ρ)＝αρ                              (1)d(ρ)=αρ (1)

式(1)中ρ是与反射镜的面形中心点的距离，α为成像系统的放大率。In formula (1), ρ is the distance from the center point of the surface shape of the mirror, and α is the magnification of the imaging system.

设反射镜在M点的法线与Z轴的夹角为γ，入射光线与Z轴的夹角为Φ，反射光线与Z轴的夹角为θ。则Let the angle between the normal of the mirror at point M and the Z axis be γ, the angle between the incident light and the Z axis be Φ, and the angle between the reflected light and the Z axis be θ. but

$\mathrm{<span><span>tg</span>tg</span>}<span><span>(</span>(</span>\mathrm{<span><span>x</span>x</span>}<span><span>)</span>)</span><span><span>=</span>=</span>\frac{\mathrm{<span><span>d</span>d</span>}<span><span>(</span>(</span>\mathrm{<span><span>x</span>x</span>}<span><span>)</span>)</span><span><span>-</span>-</span>\mathrm{<span><span>x</span>x</span>}}{\mathrm{<span><span>z</span>z</span>}<span><span>(</span>(</span>\mathrm{<span><span>x</span>x</span>}<span><span>)</span>)</span><span><span>-</span>-</span>\mathrm{<span><span>h</span>h</span>}}<span><span>-</span>-</span><span><span>-</span>-</span><span><span>-</span>-</span><span><span>(</span>(</span>\mathrm{<span><span>2</span>2</span>}<span><span>)</span>)</span>$

$\mathrm{<span><span>tg\&gamma;</span>tg\&gamma;</span>}<span><span>=</span>=</span>\frac{\mathrm{<span><span>dz</span>dz</span>}<span><span>(</span>(</span>\mathrm{<span><span>x</span>x</span>}<span><span>)</span>)</span>}{\mathrm{<span><span>dx</span>dx</span>}}<span><span>-</span>-</span><span><span>-</span>-</span><span><span>-</span>-</span><span><span>(</span>(</span>\mathrm{<span><span>3</span>3</span>}<span><span>)</span>)</span>$

$\mathrm{<span><span>tg</span>tg</span>}<span><span>(</span>(</span>\mathrm{<span><span>2</span>2</span>}\mathrm{<span><span>\&gamma;</span>\&gamma;</span>}<span><span>)</span>)</span><span><span>=</span>=</span>\frac{\mathrm{<span><span>2</span>2</span>}\frac{\mathrm{<span><span>dz</span>dz</span>}<span><span>(</span>(</span>\mathrm{<span><span>x</span>x</span>}<span><span>)</span>)</span>}{\mathrm{<span><span>dx</span>dx</span>}}}{\mathrm{<span><span>1</span>1</span>}<span><span>-</span>-</span>\frac{{\mathrm{<span><span>d</span>d</span>}}^{\mathrm{<span><span>2</span>2</span>}}\mathrm{<span><span>z</span>z</span>}<span><span>(</span>(</span>\mathrm{<span><span>z</span>z</span>}<span><span>)</span>)</span>}{{\mathrm{<span><span>dx</span>dx</span>}}^{\mathrm{<span><span>2</span>2</span>}}}}<span><span>-</span>-</span><span><span>-</span>-</span><span><span>-</span>-</span><span><span>(</span>(</span>\mathrm{<span><span>4</span>4</span>}<span><span>)</span>)</span>$

由反射定律by the law of reflection

2γ＝φ-θ2γ＝φ-θ

∴$\mathrm{tg}\left(2\mathrm{\&gamma;}\right)=\mathrm{tg}(\mathrm{\&phi;}-\mathrm{\&theta;})=\frac{\mathrm{tg\&phi;}-\mathrm{tg\&theta;}}{1+\mathrm{tg\&phi;tg\&theta;}}---\left(6\right)$∴$\mathrm{tg}\left(2\mathrm{\&gamma;}\right)=\mathrm{tg}(\mathrm{\&phi;}-\mathrm{\&theta;})=\frac{\mathrm{tg\&phi;}-\mathrm{tg\&theta;}}{1+\mathrm{tg\&phi;tg\&theta;}}---\left(6\right)$

由式(2)、(4)、(5)和(6)得到微分方程(7)From formulas (2), (4), (5) and (6) get differential equation (7)

$\frac{{\mathrm{<span><span>d</span>d</span>}}^{\mathrm{<span><span>2</span>2</span>}}\mathrm{<span><span>z</span>z</span>}<span><span>(</span>(</span>\mathrm{<span><span>z</span>z</span>}<span><span>)</span>)</span>}{{\mathrm{<span><span>dx</span><figure-callout\; label="dx"\; filenames="A20061005047100271.png,A20061005047100301.png"\; state="\{\{state\}\}">dx</figure-callout></span>}}^{\mathrm{<span><span>2</span>2</span>}}}<span><span>+</span>+</span>\mathrm{<span><span>2</span>2</span>}\mathrm{<span><span>k</span>k</span>}\frac{\mathrm{<span><span>dz</span>dz</span>}<span><span>(</span>(</span>\mathrm{<span><span>x</span>x</span>}<span><span>)</span>)</span>}{\mathrm{<span><span>dx</span>dx</span>}}<span><span>-</span>-</span>\mathrm{<span><span>1</span>1</span>}<span><span>=</span>=</span>\mathrm{<span><span>0</span>0</span>}<span><span>-</span>-</span><span><span>-</span>-</span><span><span>-</span>-</span><span><span>(</span>(</span>\mathrm{<span><span>7</span>7</span>}<span><span>)</span>)</span>$

式中；$k=\frac{z\left(x\right)[z\left(x\right)-h]+x[d\left(x\right)-x]}{z\left(x\right)[d\left(x\right)-x]+x[z\left(x\right)-h]}---\left(8\right)$in the formula;$k=\frac{z\left(x\right)[z\left(x\right)-h]+x[d\left(x\right)-x]}{z\left(x\right)[d\left(x\right)-x]+x[z\left(x\right)-h]}---\left(8\right)$

由式(7)得到微分方程(9)Differential equation (9) is obtained from equation (7)

$\frac{\mathrm{<span><span>dz</span>dz</span>}<span><span>(</span>(</span>\mathrm{<span><span>x</span>x</span>}<span><span>)</span>)</span>}{\mathrm{<span><span>dx</span>dx</span>}}<span><span>+</span>+</span>\mathrm{<span><span>k</span>k</span>}<span><span>-</span>-</span>\sqrt{{\mathrm{<span><span>k</span><figure-callout\; label="k"\; filenames="A20061005047100271.png,A20061005047100301.png"\; state="\{\{state\}\}">k</figure-callout></span>}}^{\mathrm{<span><span>2</span>2</span>}}<span><span>+</span>+</span>\mathrm{<span><span>1</span>1</span>}}<span><span>=</span>=</span>\mathrm{<span><span>0</span>0</span>}<span><span>-</span>-</span><span><span>-</span>-</span><span><span>-</span>-</span><span><span>(</span>(</span>\mathrm{<span><span>9</span>9</span>}<span><span>)</span>)</span>$

由式(1)、(5)得到式(10)From formula (1), (5) to get formula (10)

$\mathrm{<span><span>d</span>d</span>}<span><span>(</span>(</span>\mathrm{<span><span>x</span>x</span>}<span><span>)</span>)</span><span><span>=</span>=</span>\frac{\mathrm{<span><span>afx</span>afx</span>}}{\mathrm{<span><span>z</span>z</span>}<span><span>(</span>(</span>\mathrm{<span><span>x</span>x</span>}<span><span>)</span>)</span>}<span><span>-</span>-</span><span><span>-</span>-</span><span><span>-</span>-</span><span><span>(</span>(</span>\mathrm{<span><span>10</span>10</span>}<span><span>)</span>)</span>$

由式(8)、(9)、(10)和初始条件，解微分方程可以得到反射镜面形的数字解。系统外形尺寸主要指反射镜离摄像头的距离Ho和反射镜的口径D。折反射全景系统设计时根据应用要求选择合适的摄像头，标定出Rmin，透镜的焦距f确定反射镜离摄像头的距离Ho，由(1)式计算出反射镜的口径Do。According to equations (8), (9), (10) and initial conditions, the numerical solution of the mirror surface shape can be obtained by solving the differential equation. The overall dimensions of the system mainly refer to the distance Ho between the mirror and the camera and the aperture D of the mirror. When designing the catadioptric panoramic system, select the appropriate camera according to the application requirements, calibrate Rmin, the focal length f of the lens determines the distance Ho between the mirror and the camera, and calculate the aperture Do of the mirror by formula (1).

系统参数的确定：Determination of system parameters:

根据应用所要求的高度方向的视场确定系统参数af。由式(1)、(2)和(5)得到式(11)，这里作了一些简化，将z(x)≈z₀，主要考虑对于镜面的高度变化相对于镜面与摄像头的位置变化比较小；Determine the system parameter af according to the field of view in the height direction required by the application. Formula (11) is obtained from formulas (1), (2) and (5). Some simplifications are made here, and z(x)≈z₀ is mainly considered for the height change of the mirror surface relative to the position change of the mirror surface and the camera. Small;

$\mathrm{<span><span>tg\&phi;</span>tg\&phi;</span>}<span><span>=</span>=</span>\frac{<span><span>(</span>(</span>\mathrm{<span><span>af</span>af</span>}<span><span>-</span>-</span>{\mathrm{<span><span>z</span>z</span>}}_{\mathrm{<span><span>0</span>0</span>}}<span><span>)</span>)</span>\frac{\mathrm{<span><span>\&rho;</span>\&rho;</span>}}{\mathrm{<span><span>f</span>f</span>}}}{{\mathrm{<span><span>z</span>z</span>}}_{\mathrm{<span><span>0</span>0</span>}}<span><span>-</span>-</span>\mathrm{<span><span>h</span>h</span>}}<span><span>-</span>-</span><span><span>-</span>-</span><span><span>-</span>-</span><span><span>(</span>(</span>\mathrm{<span><span>11</span>11</span>}<span><span>)</span>)</span>$

在像平面以像中心点为圆心的最大圆周处ρ＝Rmin

则可以得到式(12)；ρ=Rmin at the maximum circumference of the image plane with the image center point as the center

Then formula (12) can be obtained;

$\frac{\mathrm{<span><span>\&rho;</span>\&rho;</span>}}{\mathrm{<span><span>f</span>f</span>}}<span><span>=</span>=</span>\frac{<span><span>(</span>(</span>{\mathrm{<span><span>z</span>z</span>}}_{\mathrm{<span><span>0</span>0</span>}}<span><span>-</span>-</span>\mathrm{<span><span>h</span>h</span>}<span><span>)</span>)</span>\mathrm{<span><span>tg</span>tg</span>}{\mathrm{<span><span>\&phi;</span>\&phi;</span>}}_{\mathrm{<span><span>max</span>max</span>}}}{{\mathrm{<span><span>\&omega;</span>\&omega;</span>}}_{\mathrm{<span><span>max</span>max</span>}}}<span><span>+</span>+</span>{\mathrm{<span><span>z</span>z</span>}}_{\mathrm{<span><span>0</span>0</span>}}<span><span>-</span>-</span><span><span>-</span>-</span><span><span>-</span>-</span><span><span>(</span>(</span>\mathrm{<span><span>12</span>12</span>}<span><span>)</span>)</span>$

成像模拟采用与实际光线相反的方向进行。设光源在摄像头投影中心，在像平面内等间距的选取像素点，通过这些像素点的光线，经反射镜反射后与水平面相交，若交点是等间距的，则说明反射镜具有水平场景无畸变的性质。成像模拟一方面可以评价反射镜的成像性质，另一方面可以准确地计算出反射镜的口径和厚度。Imaging simulations are performed using the opposite direction of actual light rays. Assume that the light source is at the projection center of the camera, and pixels are selected at equal intervals in the image plane. The light passing through these pixels intersects the horizontal plane after being reflected by the mirror. If the intersection points are equidistant, it means that the mirror has a horizontal scene without distortion. nature. On the one hand, the imaging simulation can evaluate the imaging properties of the mirror, and on the other hand, it can accurately calculate the aperture and thickness of the mirror.

成像变换涉及不同坐标系之间的变换。在摄像机的成像系统中，涉及到的有以下4个坐标系；(1)现实世界坐标系XYZ；(2)以摄像机为中心制定的坐标系x^y^z^；(3)像平面坐标系，在摄像机内所形成的像平面坐标系x^*y^*o^*；(4)计算机图像坐标系，计算机内部数字图像所用的坐标系MN，以像素为单位。Imaging transformations involve transformations between different coordinate systems. In the imaging system of the camera, the following four coordinate systems are involved; (1) the real world coordinate system XYZ; (2) the coordinate system x^y^z^ formulated with the camera as the center; (3) the image plane coordinates system, the image plane coordinate system x^* y^* o^* formed in the camera; (4) computer image coordinate system, the coordinate system MN used by the digital image inside the computer, with pixels as the unit.

根据以上几个坐标系不同的转换关系，就可以得到所需要的全方位摄像机成像模型，换算出二维图像到三维场景的对应关系。本发明中采用折反射全方位成像系统的近似透视成像分析方法将摄像机内所形成的像平面坐标二维图像换算到三维场景的对应关系，图3为一般的透视成像模型，d为物高，ρ为像高，t为物距，F为像距(等效焦距)。可以得到式(13)According to the different transformation relations of the above several coordinate systems, the required omni-directional camera imaging model can be obtained, and the corresponding relationship from the two-dimensional image to the three-dimensional scene can be converted. In the present invention, the approximate perspective imaging analysis method of the catadioptric omnidirectional imaging system is used to convert the two-dimensional image of the image plane coordinates formed in the camera into the corresponding relationship of the three-dimensional scene. Fig. 3 is a general perspective imaging model, and d is the height of the object. ρ is the image height, t is the object distance, and F is the image distance (equivalent focal length). Can get formula (13)

$\mathrm{<span><span>d</span>d</span>}<span><span>=</span>=</span>\frac{\mathrm{<span><span>t</span>t</span>}}{\mathrm{<span><span>F</span>f</span>}}\mathrm{<span><span>\&rho;</span>\&rho;</span>}<span><span>-</span>-</span><span><span>-</span>-</span><span><span>-</span>-</span><span><span>(</span>(</span>\mathrm{<span><span>13</span>13</span>}<span><span>)</span>)</span>$

在上述水平场景无变形的折反射全方位成像系统的设计时，要求场景物点的水平坐标与相应像点的坐标成线性关系，如式(1)表示；比较式(13)，(1)，可以看出水平场景无变形的折反射全方位成像系统对水平场景的成像为透视成像。因此就水平场景成像而言，可以将水平场景无变形的折反射全方位成像系统视为透视相机，α为成像系统的放大率。设该虚拟透视相机的投影中心为C点(见附图3)，其等效焦距为F。比较式(13)，(1)式可以得到式(14)；In the design of the catadioptric omni-directional imaging system without deformation in the above-mentioned horizontal scene, the horizontal coordinates of the scene object points are required to be in a linear relationship with the coordinates of the corresponding image points, as shown in formula (1); compare formula (13), (1) , it can be seen that the imaging of the horizontal scene by the catadioptric omnidirectional imaging system without deformation of the horizontal scene is perspective imaging. Therefore, in terms of horizontal scene imaging, the catadioptric omnidirectional imaging system without deformation of the horizontal scene can be regarded as a perspective camera, and α is the magnification of the imaging system. Assume that the projection center of the virtual perspective camera is point C (see accompanying drawing 3), and its equivalent focal length is F. Comparing formula (13), formula (1) can get formula (14);

$\mathrm{<span><span>\&alpha;</span>\&alpha;</span>}<span><span>=</span>=</span>\frac{\mathrm{<span><span>t</span>t</span>}}{\mathrm{<span><span>F</span>f</span>}}<span><span>;</span>;</span>\mathrm{<span><span>t</span>t</span>}<span><span>=</span>=</span>\mathrm{<span><span>h</span>h</span>}<span><span>-</span>-</span><span><span>-</span>-</span><span><span>-</span>-</span><span><span>(</span>(</span>\mathrm{<span><span>14</span>14</span>}<span><span>)</span>)</span>$

由式(12)、(14)得到式(15)From formula (12), (14) to get formula (15)

$\mathrm{<span><span>F</span>f</span>}<span><span>=</span>=</span>\frac{\mathrm{<span><span>fh</span>fh</span>}{\mathrm{<span><span>\&omega;</span>\&omega;</span>}}_{\mathrm{<span><span>max</span>max</span>}}}{<span><span>(</span>(</span>{\mathrm{<span><span>z</span>z</span>}}_{\mathrm{<span><span>0</span>0</span>}}<span><span>-</span>-</span>\mathrm{<span><span>h</span>h</span>}<span><span>)</span>)</span>\mathrm{<span><span>tg</span>tg</span>}{\mathrm{<span><span>\&phi;</span>\&phi;</span>}}_{\mathrm{<span><span>max</span>max</span>}}<span><span>+</span>+</span>{\mathrm{<span><span>z</span>z</span>}}_{\mathrm{<span><span>0</span>0</span>}}{\mathrm{<span><span>\&omega;</span>\&omega;</span>}}_{\mathrm{<span><span>max</span>max</span>}\mathrm{<span><span>0</span>0</span>}}}<span><span>-</span>-</span><span><span>-</span>-</span><span><span>-</span>-</span><span><span>(</span>(</span>\mathrm{<span><span>15</span>15</span>}<span><span>)</span>)</span>$

根据上述全方位摄像机成像模型进行系统成像模拟，由摄像头投影中心发出的经过像素平面内等间距像素点的光线族反射后，在距离投影中心5m的停车场的水平面上的交点基本上是等间距的，如附图4所示。因此根据上述设计原理本专利中将停车场水平面的坐标与相应全方位像点的坐标之间的关系简化为线性关系，也就是说通过反射镜面的设计将现实世界坐标系XYZ到像平面坐标系的转化可以用放大率α为比例的线形关系。下面是从像平面坐标系到计算机内部数字图像所用的坐标系的转化，计算机中使用的图像坐标单位是存储器中离散像素的个数，所以对实际像平面的坐标还需取整转换才能映射到计算机的成像平面，其变换表达式为由式(16)给出；According to the above-mentioned omnidirectional camera imaging model for system imaging simulation, the intersection points on the horizontal plane of the parking lot 5m away from the projection center are basically equidistant after the light rays emitted by the projection center of the camera are reflected by the equidistant pixel points in the pixel plane , as shown in Figure 4. Therefore, according to the above-mentioned design principle, the relationship between the coordinates of the horizontal plane of the parking lot and the coordinates of the corresponding omnidirectional image points is simplified into a linear relationship in this patent, that is to say, the coordinate system of the real world XYZ is converted to the coordinate system of the image plane through the design of the mirror surface The conversion can use the magnification α as a proportional linear relationship. The following is the transformation from the coordinate system of the image plane to the coordinate system used by the digital image inside the computer. The image coordinate unit used in the computer is the number of discrete pixels in the memory, so the coordinates of the actual image plane need to be rounded and converted to map to The imaging plane of the computer, its transformation expression is given by formula (16);

$\mathrm{<span><span>M</span>m</span>}<span><span>=</span>=</span>{\mathrm{<span><span>O</span>o</span>}}_{\mathrm{<span><span>m</span>m</span>}}<span><span>-</span>-</span>\frac{{\mathrm{<span><span>x</span>x</span>}}^{<span><span>*</span>*</span>}}{{\mathrm{<span><span>S</span>S</span>}}_{\mathrm{<span><span>x</span>x</span>}}}<span><span>;</span>;</span>\mathrm{<span><span>N</span>N</span>}<span><span>=</span>=</span>{\mathrm{<span><span>O</span>o</span>}}_{\mathrm{<span><span>n</span>no</span>}}<span><span>-</span>-</span>\frac{{\mathrm{<span><span>y</span>the\; y</span>}}^{<span><span>*</span>*</span>}}{{\mathrm{<span><span>S</span>S</span>}}_{\mathrm{<span><span>y</span>the\; y</span>}}}<span><span>;</span>;</span><span><span>-</span>-</span><span><span>-</span>-</span><span><span>-</span>-</span><span><span>(</span>(</span>\mathrm{<span><span>16</span>16</span>}<span><span>)</span>)</span>$

式中：Om、On分别为象平面的原点在计算机图像平面上所映射的点像素所在的行数和列数；Sx、Sy分别为在x和y方向上的尺度因子。Sx、Sy的确定是通过在摄像头与反射镜面之间距离Z处放置标定板，对摄像机进行标定得到Sx、Sy的数值，单位是(pixel)；Om、On。的确定是根据所选择的摄像头分辨率像素，单位是(pixel)。In the formula: Om, On are the number of rows and columns of the pixel mapped by the origin of the image plane on the computer image plane; Sx, Sy are the scale factors in the x and y directions, respectively. The determination of Sx and Sy is by placing a calibration plate at the distance Z between the camera and the mirror surface, and the camera is calibrated to obtain the values of Sx and Sy, and the unit is (pixel); Om, On. The determination of is based on the selected camera resolution pixels, and the unit is (pixel).

停车场中停放中的车辆处在相对静止状态属于静止对象，而出入停车场的车辆以及在停车场内走动中的人都属于运动对象，在获取这两种前景对象时都可以采用背景减算法图像处理方法，但是两者在背景模型的建立与更新策略是有所不同的，前者基本上要求图像背景尽可能不更新，以免将停放在停车场内的车辆作为背景；而后者则要求不断地更新图像背景，以便通过背景消减得到前景点集。本发明主要关注的是前者，因此采用的是尽可能不更新图像背景的策略。Vehicles parked in the parking lot in a relatively static state are static objects, while vehicles entering and leaving the parking lot and people walking in the parking lot are moving objects. The background subtraction algorithm can be used to obtain these two foreground objects Image processing methods, but the establishment and update strategies of the background model are different between the two. The former basically requires that the image background should not be updated as much as possible, so as not to use the vehicles parked in the parking lot as the background; while the latter requires constant Update the image background so that the set of foreground points is obtained by background subtraction. The present invention mainly focuses on the former, so it adopts a strategy of not updating the image background as much as possible.

所述的背景减算法也称为差分方法，是一种常用于检测图像变化和运动物体的图像处理方法。根据三维空间与图像像素的对应性关系把有光源点存在的那些像素部分检测出来，首先要有一个比较稳定的基准参考图像，并将该基准参考图像存储在计算机的存储器里，并通过上述的背景自适应法对基准参考图像进行动态更新，通过实时拍摄到图像与该基准参考图像间进行图像相减，相减的结果发生变化的区域亮度增强，图像相减的计算公式如式(17)表示，The background subtraction algorithm is also called the difference method, which is an image processing method commonly used to detect image changes and moving objects. According to the corresponding relationship between the three-dimensional space and the image pixels, to detect those pixels with light source points, first of all, there must be a relatively stable reference reference image, and the reference reference image is stored in the memory of the computer, and through the above-mentioned The background self-adaptive method dynamically updates the benchmark reference image, and performs image subtraction between the image captured in real time and the benchmark reference image, and the brightness of the region where the result of the subtraction changes is enhanced. The calculation formula of the image subtraction is as shown in formula (17) express,

f_d(X，t₀，t_i)＝f(X，t_i)-f(X，t₀)                (17)f_d (X, t₀ , t_i )=f(X, t_i )-f(X, t₀ ) (17)

式中f_d(X，t₀，t_i)是实时拍摄到图像与基准参考图像间进行图像相减的结果；f(X，t_i)是实时拍摄到图像，f(X，t₀)是基准参考图像。In the formula, f_d (X, t₀ , t_i ) is the result of image subtraction between the image captured in real time and the benchmark reference image; f(X, t_i ) is the image captured in real time, and f(X, t₀ ) is the benchmark reference image.

考虑到停车场内的地面的颜色都接近于灰白色，与停放的车辆的颜色是有明显区别的，因此可以利用颜色模型来进行图像相减计算。一幅彩色图像每个像素点的颜色通常由红绿蓝三刺激值加权合成，其它的彩色基如强度、色调t饱和度HSI基等可以由红绿蓝RGB值线性或非线性变换获得。为了获取停车场内图像上的停放车辆区域和背景区域在不同的色彩空间里和不同的光照的颜色特征值的区别，本专利中采用了HSI空间颜色模型。Considering that the color of the ground in the parking lot is close to off-white, which is obviously different from the color of the parked vehicles, the color model can be used for image subtraction calculation. The color of each pixel in a color image is usually weighted and synthesized by red, green and blue tristimulus values, and other color bases such as intensity, hue, saturation, HSI base, etc. can be obtained by linear or nonlinear transformation of red, green, and blue RGB values. In order to obtain the difference between the parked vehicle area and the background area on the image in the parking lot in different color spaces and different lighting color feature values, the HSI space color model is adopted in this patent.

HSI色彩空间是基于人类对色彩的感觉，HSI模型描述颜色有如下三个基本特征：1、色调H，在0到360度的标准色轮上，色调是按位置度量的。在通常的使用中，色调是由颜色名称标识的，比如红、橙或绿色；2、饱和度S，是指颜色的强度或纯度。饱和度表示色相中彩色成分所占的比例，用从0％(灰色)到100％(完全饱和)的百分比来度量。在标准色轮上，从中心向边缘饱和度是递增的；3、亮度I，是颜色的相对明暗程度，通常用从0％(黑)到100％(白)的百分比来度量。The HSI color space is based on human perception of color. The HSI model describes the following three basic characteristics of color: 1. Hue H. On the standard color wheel from 0 to 360 degrees, the hue is measured by position. In common use, hue is identified by a color name, such as red, orange, or green; 2. Saturation, S, refers to the intensity or purity of a color. Saturation represents the proportion of the color component in the hue, measured as a percentage from 0% (gray) to 100% (fully saturated). On the standard color wheel, the saturation increases from the center to the edge; 3. Brightness I is the relative lightness and darkness of the color, usually measured by the percentage from 0% (black) to 100% (white).

通常把色调和饱和度通称为色度，用来表示颜色的类别与深浅程度。由于人的视觉对亮度的敏感程度远强于对颜色浓淡的敏感程度，为了便于色彩处理和识别，人的视觉系统经常采用HSI色彩空间，它比RGB色彩空间更符合人的视觉特性。在图像处理和计算机视觉中大量算法都可在HSI色彩空间中方便地使用，它们可以分开处理而且是相互独立的。因此，在HSI色彩空间可以大大简化图像分析和处理的工作量。注意到HSI模型有两个重要的事实，首先是I分量与颜色无关，主要受光源强弱影响，其次H与S分量与人感受色彩的方式紧密相连。停车场中的光线会有强弱变化，而停车场的地面颜色与停放车辆的颜色的H与S分量是独立的，不会由于光线的强弱变化而变化。所以利用HSI色彩空间中的颜色的H与S分量进行实时拍摄到图像与基准参考图像的图像相减计算就能非常容易地得到停车场内的每个车位是否被占用的信息。HSI色彩空间与红绿蓝RGB色彩空间的转化关系由式(18)表示，Hue and saturation are usually referred to as chroma, which is used to indicate the category and depth of color. Since human vision is much more sensitive to brightness than to color shades, in order to facilitate color processing and recognition, the human visual system often uses the HSI color space, which is more in line with human visual characteristics than the RGB color space. A large number of algorithms in image processing and computer vision can be conveniently used in the HSI color space, and they can be processed separately and independently of each other. Therefore, the workload of image analysis and processing can be greatly simplified in the HSI color space. Note that there are two important facts in the HSI model. First, the I component has nothing to do with color, but is mainly affected by the intensity of the light source. Second, the H and S components are closely related to the way people perceive colors. The light in the parking lot will change in intensity, but the ground color of the parking lot and the H and S components of the color of the parked vehicle are independent, and will not change due to the intensity of light. Therefore, by using the H and S components of the color in the HSI color space to subtract the images captured in real time and the benchmark reference image, it is very easy to obtain information about whether each parking space in the parking lot is occupied. The conversion relationship between the HSI color space and the red, green and blue RGB color space is expressed by formula (18),

$\mathrm{<span><span>I</span>I</span>}<span><span>=</span>=</span>\frac{\mathrm{<span><span>R</span>R</span>}<span><span>+</span>+</span>\mathrm{<span><span>G</span>G</span>}<span><span>+</span>+</span>\mathrm{<span><span>B</span><figure-callout\; label="B"\; filenames="A20061005047100271.png,A20061005047100301.png"\; state="\{\{state\}\}">B</figure-callout></span>}}{\mathrm{<span><span>3</span>3</span>}}$

$\mathrm{<span><span>H</span>h</span>}<span><span>=</span>=</span>\frac{\mathrm{<span><span>1</span>1</span>}}{\mathrm{<span><span>360</span>360</span>}}<span><span>[</span>[</span>\mathrm{<span><span>90</span>90</span>}<span><span>-</span>-</span>\mathrm{<span><span>Arc</span>Arc</span>}\mathrm{<span><span>tan</span>the\; tan</span>}<span><span>(</span>(</span>\frac{\mathrm{<span><span>F</span>f</span>}}{\sqrt{\mathrm{<span><span>3</span>3</span>}}}<span><span>)</span>)</span><span><span>+</span>+</span><span><span>\{</span>\{</span>\mathrm{<span><span>0</span>0</span>}<span><span>,</span>,</span>\mathrm{<span><span>G</span>G</span>}<span><span>></span>></span>\mathrm{<span><span>B</span>B</span>}<span><span>;</span>;</span>\mathrm{<span><span>180</span>180</span>}<span><span>,</span>,</span>\mathrm{<span><span>G</span>G</span>}<span><span><</span><</span>\mathrm{<span><span>B</span>B</span>}<span><span>\}</span>\}</span><span><span>]</span>]</span>$

$\mathrm{<span><span>S</span>S</span>}<span><span>=</span>=</span>\mathrm{<span><span>1</span>1</span>}<span><span>-</span>-</span><span><span>\{</span>\{</span>\frac{\mathrm{<span><span>min</span>min</span>}<span><span>(</span>(</span>\mathrm{<span><span>R</span>R</span>}<span><span>,</span>,</span>\mathrm{<span><span>G</span>G</span>}<span><span>,</span>,</span>\mathrm{<span><span>B</span>B</span>}<span><span>)</span>)</span>}{\mathrm{<span><span>I</span>I</span>}}<span><span>\}</span>\}</span><span><span>-</span>-</span><span><span>-</span>-</span><span><span>-</span>-</span><span><span>(</span>(</span>\mathrm{<span><span>18</span>18</span>}<span><span>)</span>)</span>$

其中，$F=\frac{2R-G-B}{G-B}.$in,$f=\frac{2R-G-B}{G-B}.$

像素间的连通性是确定区域的一个重要概念。在确定停车场车位是否被占用时可以利用检查虚拟车位框区域中是否存在有连通区域方法。具体的做法是：在二维图像中，假设目标像素周围有m(m＜＝8)个相邻的像素，如果该像素灰度与这m个像素中某一个点A的灰度相等，那么称该像素与点A具有连通性。常用的连通性有4连通和8连通。4连通一般选取目标像素的上、下、左、右四个点。8连通则选取目标像素在二维空间中所有的相邻像素。将所有具有连通性的像素作为一个区域则构成了一个连通区域。The connectivity between pixels is an important concept to determine the region. When determining whether the parking space in the parking lot is occupied, the method of checking whether there is a connected area in the virtual parking space frame area can be used. The specific method is: in a two-dimensional image, assuming that there are m (m<=8) adjacent pixels around the target pixel, if the gray level of this pixel is equal to the gray level of a certain point A in these m pixels, then The pixel is said to be connected to point A. The commonly used connectivity is 4-connectivity and 8-connectivity. 4 connectivity generally selects the upper, lower, left and right four points of the target pixel. 8-connectivity selects all adjacent pixels of the target pixel in two-dimensional space. Taking all connected pixels as a region constitutes a connected region.

所述的连通区域计算主要解决在图像处理过程中，一幅二值图像，其背景和目标分别具有灰度值0和1。我们将像素为0的小区表示此小区无物体存在，若为1则表示此小区有物体存在。由于有些车辆的玻璃与地面的颜色比较接近，所以可以采用连通成分标记法进行缺陷区域的合并。连通标记算法可以找到图像中的所有连通成分，并对同一连通成分中的所有点分配同一标记。图5为连通标记原理图。下面是连通区域算法，The calculation of the connected regions mainly solves the problem that in a binary image, the background and the target have gray values of 0 and 1, respectively, during image processing. We use a cell with a pixel of 0 to indicate that there is no object in this cell, and a cell with a pixel of 1 to indicate that there is an object in this cell. Since the glass of some vehicles is close to the color of the ground, the connected component labeling method can be used to merge defect areas. Connected labeling algorithms can find all connected components in an image and assign the same label to all points in the same connected component. Fig. 5 is a schematic diagram of connectivity marking. The following is the connected region algorithm,

1)从左到右、从上到下扫描图像；1) Scan the image from left to right and top to bottom;

2)如果像素点为1，则：2) If the pixel is 1, then:

·如果上面点和左面点有一个标记，则复制这一标记。• If there is a mark for the upper point and the left point, copy this mark.

·如果两点有相同的标记，复制这一标记。• If two points have the same label, copy this label.

·如果两点有不同的标记，则复制上点的标记且将两个标记输入等价表中作为等价标记。• If two points have different labels, copy the label of the previous point and enter both labels into the equivalence table as equivalence labels.

·否则给这个象素点分配新的标记并将这一标记输入等价表。• Otherwise assign a new label to this pixel and enter this label into the equivalence table.

3)如果需考虑更多的点则回到第2步。3) Return to step 2 if more points need to be considered.

4)在等价表的每一等价集中找到最低的标记。4) Find the lowest token in each equivalence set of the equivalence table.

5)扫描图像，用等价表中的最低标记取代每一标记。5) Scan the image, replacing each token with the lowest token in the equivalence table.

然后根据每个虚拟车位框中的连通区域统计求出其面积Si，当统计求出其面积Si超过阈值1时就认为是在该车位上有车，小于阈值1时就认为是噪声或者是丢弃物(纸张、塑料薄膜等)。Then calculate its area Si according to the statistics of the connected regions in each virtual parking space frame. When the statistically calculated area Si exceeds thethreshold value 1, it is considered that there is a car in the parking space, and when it is less than thethreshold value 1, it is considered as noise or discarded. objects (paper, plastic film, etc.).

本发明的有益效果主要表现在：1)检测范围广，能对方位在200米直径以内的停放车辆进行检测；2)安装维护无干扰，由于视频检测器往往是安装在停车场中部的顶上，因此安装及维护不会影响停车场的营业，也不需要开挖、破坏路面；3)维护方便低耗，传统的感应线圈检测器在损坏时，需要开挖路面进行维护，而视频检测设备发生问题时，可直接摘除或修理设备，减少了维护费用；4)检测参数丰富，不但能检测停车场内的车位占用情况，通过加入一些新的算法后还能够检测到停车场内的各种安全隐患，比如车辆的盗难、火灾等事故，这是一般的感应线圈检测器无法比拟的；5)可视性，能够将全方位的实时图像传给停车场的管理者，实现监视的职能；6)检测可靠性、准确度高，不会与传统的感应线圈检测器一样会有误动作或者误检测；7)统计计算方便，算法实现简单，特别适用于在大型停车场的管理；8)具有良好的先进性、可扩展性、可持续发展性，视频车位检测技术是智能交通系统的关键技术之一，其本身就能单独成为一个系统，通过网络能够与先进的车辆信息系统、等动态智能交通模块衔接，实现更多的功能。The beneficial effects of the present invention are mainly manifested in: 1) the detection range is wide, and the parked vehicles within a diameter of 200 meters can be detected; 2) the installation and maintenance do not interfere, because the video detector is often installed on the top of the middle part of the parking lot , so the installation and maintenance will not affect the business of the parking lot, and there is no need to excavate or damage the road surface; 3) The maintenance is convenient and low in consumption. When the traditional induction coil detector is damaged, the road surface needs to be excavated for maintenance, while the video detection equipment When a problem occurs, the equipment can be directly removed or repaired, reducing maintenance costs; 4) The detection parameters are rich, not only can detect the occupancy of parking spaces in the parking lot, but also can detect various parking spaces in the parking lot by adding some new algorithms. Potential safety hazards, such as vehicle theft, fire and other accidents, are unmatched by general induction coil detectors; 5) Visibility, which can transmit all-round real-time images to the manager of the parking lot to realize the monitoring function ;6) The detection reliability and accuracy are high, and there will be no misoperation or false detection like the traditional induction coil detector; 7) The statistical calculation is convenient, and the algorithm is simple to implement, especially suitable for the management of large parking lots; 8 ) has good advancement, scalability, and sustainable development. Video parking space detection technology is one of the key technologies of intelligent transportation systems. It can become a system by itself, and can communicate with advanced vehicle information systems, etc. The dynamic intelligent traffic module is connected to realize more functions.

(四)附图说明(4) Description of drawings

图1为三维立体空间反射到全方位视觉平面成像示意图；Fig. 1 is a schematic diagram of three-dimensional space reflection to omni-directional visual plane imaging;

图2为全方位视觉传感器的硬件组成示意图；Figure 2 is a schematic diagram of the hardware composition of the omnidirectional vision sensor;

图3为全方位视觉装置与一般的透视成像模型等价的透视投影成像模型示意图；3 is a schematic diagram of a perspective projection imaging model equivalent to an omnidirectional vision device and a general perspective imaging model;

图4为全方位视觉装置在水平方向上图像无形变模拟示意图；Fig. 4 is a schematic diagram of simulation without image deformation in the horizontal direction of the omnidirectional vision device;

图5为电子泊车诱导系统的结构功能框图；5 is a structural and functional block diagram of the electronic parking guidance system;

图6为基于全方位计算机视觉的电子泊车诱导系统的功能框图；Fig. 6 is the functional block diagram of the electronic parking guidance system based on omnidirectional computer vision;

图7为全方位视觉传感器的电子泊车诱导系统的处理流程图。Fig. 7 is a processing flowchart of the electronic parking guidance system of the omnidirectional vision sensor.

图8为基于全方位计算机视觉的电子泊车诱导系统实际应用例图。Fig. 8 is an example diagram of the practical application of the electronic parking guidance system based on omnidirectional computer vision.

(五)具体实施方式(5) Specific implementation methods

下面结合附图对本发明作进一步描述。The present invention will be further described below in conjunction with the accompanying drawings.

实施例1Example 1

参照图1～图8，一种基于全方位计算机视觉的电子泊车诱导系统，该电子泊车诱导系统包括微处理器6、用于监测停车场内停放车辆情况的全方位视觉传感器13、用于与外界通信的通信模块27，所述的全方位视觉传感器13安装在待监测的停车场的上部，全方位视觉传感器13通过USB接口连接微处理器6；Referring to Figures 1 to 8, an electronic parking guidance system based on omnidirectional computer vision, the electronic parking guidance system includes amicroprocessor 6, an omnidirectional visual sensor 13 for monitoring the situation of vehicles parked in a parking lot, and a In thecommunication module 27 communicating with the outside world, the omnidirectional vision sensor 13 is installed on the top of the parking lot to be monitored, and the omnidirectional vision sensor 13 is connected to themicroprocessor 6 through the USB interface;

所述的全方位视觉传感器13包括用于反射监控领域中物体的外凸反射镜面1、透明圆柱体3、摄像头5，所述的外凸反射镜面1朝下，所述的透明圆柱体3支撑外凸反射镜面1，黑色圆锥体2固定在折反射镜面1外凸部的中心，用于拍摄外凸反射镜面上成像体的摄像头5位于透明圆柱体的内部，摄像头5位于外凸反射镜面1的虚焦点上；The omni-directional visual sensor 13 includes aconvex mirror surface 1, atransparent cylinder 3, and acamera 5 for reflecting objects in the monitoring field, theconvex mirror surface 1 faces downward, and thetransparent cylinder 3 supportsConvex mirror surface 1,black cone 2 is fixed on the center of the convex portion ofcatadioptric mirror surface 1, thecamera 5 for photographing the imaging body on the convex reflective mirror surface is located inside the transparent cylinder, and thecamera 5 is located on the convexreflective mirror surface 1 on the virtual focus of

所述的微处理器包括：Described microprocessor comprises:

图像数据读取模块15，用于读取从视觉传感器传过来的停车场内车位的图像信息；Image data reading module 15, is used for reading the image information of the parking space in the parking lot that passes from vision sensor;

图像数据文件存储模块16，用于将读进来的图像信息通过文件方式保存在存储单元中；The image datafile storage module 16 is used to save the image information read in in the storage unit by means of a file;

虚拟车位框设置模块17，用于将读取的整个停车场的图像信息，按照车位分布情况设置与实际车位一一对应的虚拟车位框，并保存基准参考图像；The virtual parking spaceframe setting module 17 is used to set the virtual parking space frame corresponding to the actual parking space one to one according to the image information of the whole parking lot read according to the parking space distribution situation, and save the benchmark reference image;

传感器标定模块18，用于对全方位视觉传感器的参数进行标定，建立空间的实物图像与所获得的视频图像的线性对应关系；Thesensor calibration module 18 is used to calibrate the parameters of the omnidirectional visual sensor, and establish a linear correspondence between the physical image of the space and the obtained video image;

网络传输模块19，用于将图像信息通过网络向外界输出；Thenetwork transmission module 19 is used to output the image information to the outside world through the network;

颜色空间转化处理模块20，用于将图像数据读取模块采集的图像从RGB色彩空间转化到HSI色彩空间，转换的计算式为(18)：The color spaceconversion processing module 20 is used to convert the image collected by the image data reading module to the HSI color space from the RGB color space, and the conversion formula is (18):

$\mathrm{<span><span>I</span>I</span>}<span><span>=</span>=</span>\frac{\mathrm{<span><span>R</span>R</span>}<span><span>+</span>+</span>\mathrm{<span><span>G</span>G</span>}<span><span>+</span>+</span>\mathrm{<span><span>B</span><figure-callout\; label="B"\; filenames="A20061005047100271.png,A20061005047100301.png"\; state="\{\{state\}\}">B</figure-callout></span>}}{\mathrm{<span><span>3</span>3</span>}}$

$\mathrm{<span><span>H</span>h</span>}<span><span>=</span>=</span>\frac{\mathrm{<span><span>1</span>1</span>}}{\mathrm{<span><span>360</span>360</span>}}<span><span>[</span>[</span>\mathrm{<span><span>90</span>90</span>}<span><span>-</span>-</span>\mathrm{<span><span>Arc</span>Arc</span>}\mathrm{<span><span>tan</span>the\; tan</span>}<span><span>(</span>(</span>\frac{\mathrm{<span><span>F</span>f</span>}}{\sqrt{\mathrm{<span><span>3</span>3</span>}}}<span><span>)</span>)</span><span><span>+</span>+</span><span><span>\{</span>\{</span>\mathrm{<span><span>0</span>0</span>}<span><span>,</span>,</span>\mathrm{<span><span>G</span>G</span>}<span><span>></span>></span>\mathrm{<span><span>B</span>B</span>}<span><span>;</span>;</span>\mathrm{<span><span>180</span>180</span>}<span><span>,</span>,</span>\mathrm{<span><span>G</span>G</span>}<span><span><</span><</span>\mathrm{<span><span>B</span>B</span>}<span><span>\}</span>\}</span><span><span>]</span>]</span>$

$\mathrm{<span><span>S</span>S</span>}<span><span>=</span>=</span>\mathrm{<span><span>1</span>1</span>}<span><span>-</span>-</span><span><span>\{</span>\{</span>\frac{\mathrm{<span><span>min</span>min</span>}<span><span>(</span>(</span>\mathrm{<span><span>R</span>R</span>}<span><span>,</span>,</span>\mathrm{<span><span>G</span>G</span>}<span><span>,</span>,</span>\mathrm{<span><span>B</span>B</span>}<span><span>)</span>)</span>}{\mathrm{<span><span>I</span>I</span>}}<span><span>\}</span>\}</span><span><span>-</span>-</span><span><span>-</span>-</span><span><span>-</span>-</span><span><span>(</span>(</span>\mathrm{<span><span>18</span>18</span>}<span><span>)</span>)</span>$

其中，$F=\frac{2R-G-B}{G-B}$in,$f=\frac{2R-G-B}{G-B}$

上式中，H为HSI色彩空间的色调，S为HSI色彩空间的饱和度，I为HSI色彩空间的亮度，R为RGB色彩空间的红色；G为RGB色彩空间的绿色；B为RGB色彩空间的蓝色；In the above formula, H is the hue of the HSI color space, S is the saturation of the HSI color space, I is the brightness of the HSI color space, R is the red of the RGB color space; G is the green of the RGB color space; B is the RGB color space the blue;

颜色空间转化处理模块的输入端连接所述的虚拟车框检测模块；The input end of the color space conversion processing module is connected to the virtual car frame detection module;

虚拟车框检测模块21，用于对各个虚拟车位框，将所获得的当前帧现场视频图像与基准参考图像进行差值运算，图像相减的计算公式如式(17)表示：The virtual car frame detection module 21 is used for each virtual parking space frame, and the obtained current frame live video image and the benchmark reference image are carried out difference calculation, and the calculation formula of image subtraction is expressed as formula (17):

f_d(X，t₀，t_i)＝f(X，t_i)-f(X，t₀)                (17)f_d (X, t₀ , t_i )=f(X, t_i )-f(X, t₀ ) (17)

上式中，f_d(X，t₀，t_i)是实时拍摄到图像与基准参考图像间进行图像相减的结果；f(X，t_i)是实时拍摄到图像；f(X，t₀)是基准参考图像；In the above formula, f_d (X, t₀ , t_i ) is the result of image subtraction between the image captured in real time and the benchmark reference image; f(X, t_i ) is the image captured in real time; f(X, t₀ ) is the benchmark reference image;

如f_d(X，t₀，t_i)≥阈值时，判定为可疑有车事件；If f_d (X, t₀ , t_i )≥threshold, it is judged as a suspicious car-occupancy event;

如f_d(X，t₀，t_i)＜阈值时，判定为无可疑有车事件；If f_d (X, t₀ , t_i )<threshold value, it is determined that there is no suspicious vehicle occurrence;

连通区域计算模块22，用于在判定有可疑有车事件后，对当前图像进行标记，像素灰度为0的小区表示此小区无可疑有车，像素灰度为1则表示此小区有可疑有车，计算当前图像中的像素是否与当前像素周围相邻的某一个点的像素相等，如灰度相等判断为具有连通性，将所有具有连通性的像素作为一个连通区域；The connected area calculation module 22 is used to mark the current image after it is determined that there is a suspicious car event. A cell with a pixel grayscale of 0 means that there is no suspicious car in this cell, and a pixel gray level of 1 means that there is a suspicious car in this cell. Car, calculate whether the pixel in the current image is equal to the pixel of a certain point adjacent to the current pixel, if the gray level is equal, it is judged to have connectivity, and all the pixels with connectivity are regarded as a connected region;

车辆判断模块23，用于根据得到的连通区域，统计求出连通区域面积Si，并将连通区域面积与预设的阈值做比较：Thevehicle judging module 23 is used to statistically calculate the area Si of the connected area according to the obtained connected area, and compare the area of the connected area with a preset threshold:

如连通区域面积Si大于阈值S_阈，判定为在该车位上有车；If the area Si of the connected area is greater than the threshold S_threshold , it is determined that there is a car in the parking space;

如连通区域面积Si小于阈值S_阈，判定为在该车位上没有车；If the area Si of the connected area is smaller than the threshold S_threshold , it is determined that there is no car in the parking space;

车位信息发布模块25，用于根据各个虚拟车位框中的车辆判断，得到停车场内的车位占用信息，通过通信模块发布车位占用信息；Parking spaceinformation release module 25, for judging according to the vehicle in each virtual parking space frame, obtain the parking space occupancy information in the parking lot, release the parking space occupancy information by communication module;

车位信息更新模块24，用于根据当前监测的车位信息，跟上一次统计的车位信息做比较，如果车位占用信息有变化，更新发布车位占用信息；The parking spaceinformation updating module 24 is used for comparing the parking space information with the last statistical parking space information according to the currently monitored parking space information. If the parking space occupancy information changes, update and issue the parking space occupancy information;

车位预约处理模块26，用于管理人员根据预约车位情况预先设置车位预约情况，并将预约信息输入到车位信息发布模块，综合判断车位占用信息。The parking spacereservation processing module 26 is used for management personnel to pre-set the parking space reservation situation according to the reserved parking space situation, and input the reservation information into the parking space information release module to comprehensively judge the parking space occupancy information.

结合图1并参照图2，本发明的全方位视觉功能的配件的结构为：折反射面镜1、黑色圆锥体2、透明外罩圆柱体3、底座9所组成，所述的折反射面镜1位于圆柱体3的上端，且反射镜面的凸面伸入圆柱体内向下；所述的黑色圆锥体2固定在折反射面镜1的凸面的中心部；所述的折反射面镜1、黑色圆锥体2、圆柱体3、底座9的旋转轴在同一中心轴线上；所述的数码CCD摄像头5位于圆柱体2内的下方；所述的底座9上开有与所述的圆柱体2的壁厚相同的圆槽；所述的底座9上设有一个与数码摄像装置5的镜头4一样大小的孔，所述的底座9的下部配置微处理器6、存储器8以及显示器7。In conjunction with Fig. 1 and with reference to Fig. 2, the structure of the accessory of omnidirectional visual function of the present invention is: thecatadioptric mirror 1,black cone 2, transparentouter cover cylinder 3,base 9 are made up of, and describedcatadioptric mirror 1 is located at the upper end of thecylinder 3, and the convex surface of the mirror surface extends downward in the cylinder; theblack cone 2 is fixed at the center of the convex surface of thecatadioptric mirror 1; thecatadioptric mirror 1, black The rotating shaft ofcone 2,cylinder 3,base 9 is on the same central axis; Described digitalCCD camera head 5 is positioned at the below incylinder 2; A circular groove with the same wall thickness; thebase 9 is provided with a hole with the same size as thelens 4 of thedigital camera 5 , and the bottom of thebase 9 is equipped with amicroprocessor 6 , amemory 8 and adisplay 7 .

停车场中停放中的车辆处在相对静止状态属于静止对象，而出入停车场的车辆以及在停车场内走动中的人都属于运动对象，在获取这两种前景对象时都可以采用背景减算法图像处理方法，但是两者在背景模型的建立与更新策略是有所不同的，前者基本上要求图像背景不更新，以免将停放在停车场内的车辆作为背景；而后者则要求不断地更新图像背景，以便通过背景消减得到前景点集。所述的背景减算法在图7的求差影图得到虚拟车框检测模块21中进行。Vehicles parked in the parking lot in a relatively static state are static objects, while vehicles entering and leaving the parking lot and people walking in the parking lot are moving objects. The background subtraction algorithm can be used to obtain these two foreground objects Image processing methods, but the establishment and update strategies of the background model are different between the two. The former basically requires that the image background not be updated, so as not to use the vehicles parked in the parking lot as the background; while the latter requires continuous image updating. background in order to get the foreground point set by background subtraction. The background subtraction algorithm is performed in the virtual vehicle frame detection module 21 obtained from the difference shadow map in FIG. 7 .

所述的背景减算法也称为差分方法，是一种常用于检测图像变化和运动物体的图像处理方法。根据三维空间与图像像素的对应性关系把有光源点存在的那些像素部分检测出来，首先要有一个比较稳定的基准参考图像，并将该基准参考图像存储在计算机的存储器里，并通过上述的背景自适应法对基准参考图像进行动态更新，通过实时拍摄到图像与该基准参考图像间进行图像相减，相减的结果发生变化的区域亮度增强，图像相减的计算公式如式(17)表示，The background subtraction algorithm is also called the difference method, which is an image processing method commonly used to detect image changes and moving objects. According to the corresponding relationship between the three-dimensional space and the image pixels, to detect those pixels with light source points, first of all, there must be a relatively stable reference reference image, and the reference reference image is stored in the memory of the computer, and through the above-mentioned The background self-adaptive method dynamically updates the benchmark reference image, and performs image subtraction between the image captured in real time and the benchmark reference image, and the brightness of the region where the result of the subtraction changes is enhanced. The calculation formula of the image subtraction is as shown in formula (17) express,

f_d(X，t₀，t_i)＝f(X，t_i)-f(X，t₀)                (17)f_d (X, t₀ , t_i )=f(X, t_i )-f(X, t₀ ) (17)

式中f_d(X，t₀，t_i)是实时拍摄到图像与基准参考图像间进行图像相减的结果；f(X，t_i)是实时拍摄到图像，f(X，t₀)是基准参考图像。其中基准参考图像存放在图7的图像数据文件、虚拟车位存储模块17中，所述的基准参考图像是在停车场没有车辆停放情况下采集的图像。In the formula, f_d (X, t₀ , t_i ) is the result of image subtraction between the image captured in real time and the benchmark reference image; f(X, t_i ) is the image captured in real time, and f(X, t₀ ) is the benchmark reference image. Wherein the reference reference image is stored in the image data file of FIG. 7 and the virtual parkingspace storage module 17, and the reference reference image is an image collected when there is no vehicle parked in the parking lot.

考虑到停车场内的地面的颜色都接近于灰白色，与停放的车辆的颜色是有明显区别的，因此可以利用颜色模型来进行图像相减计算。在本发明中采用了HSI色彩空间模型，HSI色彩空间与红绿蓝RGB色彩空间的转化是在图7的颜色空间转化处理模块20中进行的，这是因为考虑到停车场中的光线会有强弱变化，而停车场的地面颜色与停放车辆的颜色的H与S分量是独立的，不会由于光线的强弱变化而变化。所以利用HSI色彩空间中的颜色的H与S分量进行实时拍摄到图像与基准参考图像的图像相减计算就能非常容易地得到停车场内的每个车位是否被占用的信息。Considering that the color of the ground in the parking lot is close to off-white, which is obviously different from the color of the parked vehicles, the color model can be used for image subtraction calculation. Adopted HSI color space model in the present invention, the conversion of HSI color space and red green blue RGB color space is carried out in the color spaceconversion processing module 20 of Fig. 7, and this is because the light in the parking lot will have The intensity changes, while the H and S components of the ground color of the parking lot and the color of the parked vehicle are independent, and will not change due to changes in the intensity of light. Therefore, by using the H and S components of the color in the HSI color space to subtract the images captured in real time and the benchmark reference image, it is very easy to obtain information about whether each parking space in the parking lot is occupied.

在求差影图后为了要得到每个车位上是否停有车辆的信息，还需要进行图像的预处理，所述的图像预处理主要是在图7的连通区域计算模块22中进行的，主要利用检查虚拟车位框区域中是否存在有连通区域方法。具体的做法是：在二维图像中，假设目标像素周围有m(m＜＝8)个相邻的像素，如果该像素灰度与这m个像素中某一个点A的灰度相等，那么称该像素与点A具有连通性。常用的连通性有4连通和8连通。4连通一般选取目标像素的上、下、左、右四个点。8连通则选取目标像素在二维空间中所有的相邻像素。将所有具有连通性的像素作为一个区域则构成了一个连通区域。In order to obtain the information of whether there is a vehicle parked on each parking space after calculating the difference shadow map, image preprocessing is also required. The image preprocessing is mainly carried out in the connected area calculation module 22 of Fig. 7, mainly Use the method of checking whether there is a connected area in the virtual parking space frame area. The specific method is: in a two-dimensional image, assuming that there are m (m<=8) adjacent pixels around the target pixel, if the gray level of this pixel is equal to the gray level of a certain point A in these m pixels, then The pixel is said to be connected to point A. The commonly used connectivity is 4-connectivity and 8-connectivity. 4 connectivity generally selects the upper, lower, left and right four points of the target pixel. 8-connectivity selects all adjacent pixels of the target pixel in two-dimensional space. Taking all connected pixels as a region constitutes a connected region.

所述的连通区域计算主要解决在图像处理过程中，一幅二值图像，其背景和目标分别具有灰度值0和1。我们将像素为0的小区表示此小区无物体存在，若为1则表示此小区有物体存在。由于有些车辆的玻璃与地面的颜色比较接近，所以可以采用连通成分标记法进行缺陷区域的合并。然后根据每个虚拟车位框中的连通区域统计求出其面积Si，当统计求出其面积Si超过阈值1时就认为是在该车位上有车，小于阈值1时就认为是在停车场内的视频噪声或者是丢弃物(纸张、塑料薄膜等)。The calculation of the connected regions mainly solves the problem that in a binary image, the background and the target have gray values of 0 and 1, respectively, during image processing. We use a cell with a pixel of 0 to indicate that there is no object in this cell, and a cell with a pixel of 1 to indicate that there is an object in this cell. Since the glass of some vehicles is close to the color of the ground, the connected component labeling method can be used to merge defect areas. Then calculate its area Si according to the statistics of the connected regions in each virtual parking space frame. When the statistically calculated area Si exceeds thethreshold value 1, it is considered that there is a car in the parking space, and when it is less than thethreshold value 1, it is considered to be in the parking lot. video noise or discarded objects (paper, plastic film, etc.).

所述的虚拟车位框是在系统投入运行前对全方位视觉传感器进行标定时进行制作的，具体做法是通过全方位计算机视觉传感器获得停车场中场内车位的图像信息，并将该所获得的车位图像信息显示在显示器上，然后根据所显示的停车场中各车位情况在计算机中设置虚拟车位框，并将虚拟车位框保存在一个文件中(图7中的图像数据文件、虚拟车位存储模块17)，要求所设置的虚拟车位框与实际停车场中各车位一一对应，并给以编号；在图8所示的停车场场景中共有50个实际车位，图7中的图像数据文件、虚拟车位存储模块18中也相应的保存着50个虚拟车位框的大小以及每个车位的编号等信息。The virtual parking space frame is produced when the omnidirectional vision sensor is calibrated before the system is put into operation. The specific method is to obtain the image information of the parking space in the parking lot through the omnidirectional computer vision sensor, and use the obtained The parking space image information is displayed on the display, and then a virtual parking space frame is set in the computer according to each parking space situation in the displayed parking lot, and the virtual parking space frame is saved in a file (image data file among Fig. 7, virtual parking space storage module 17), the virtual parking space frame that is required to be set corresponds to each parking space in the actual parking lot one by one, and is numbered; There are 50 actual parking spaces in the parking lot scene shown in Fig. 8, the image data file among Fig. 7, Information such as the size of 50 virtual parking space frames and the numbering of each parking space are also correspondingly stored in the virtual parkingspace storage module 18 .

停车场的使用者可以通过各种通信手段预约停车场中的车位，如通过互联网在表示着停车场内场景图上选定所想要停放的位置以及停放的时间段。预约停车场中的车位的处理是在图7中的车位预约处理模块26中进行的。The user of the parking lot can reserve the parking space in the parking lot by various means of communication, such as selecting the desired parking position and the time period for parking on the scene map in the parking lot through the Internet. The processing of reserving a parking space in the parking lot is carried out in the parking spacereservation processing module 26 in FIG. 7 .

一旦检测出停车场内的车位停放情况有变化或者车位已经被预约，在图7中的停车场车位信息更新模块24中要重新计算停车场内的车位占用情况，以便为车位信息发布模块25作好数据准备，在车位信息发布模块25中主要要完成两个任务，一个任务是为场内车辆诱导发布可停放的车位，这些信息用安放在停车场的入口处的大型显示屏幕显示可停放的车位，以便停车场的使用者能迅速找到要停放的车位；另一个任务是为场外车辆诱导发布可停放的车位数，并结合存放在停车场数据存储信息28中的有关停车场的地理位置等信息，随时将停车场的使用状况在可变信息显示板上以视觉的方式或通过广播以听觉的方式向驾驶员提供，也可以利用互联网、移动电话以及车载导航装置等方式发布，及时将停车场的使用情况更新设置于路侧的诱导信息板，如图6所示，诱导信息板显示的信息要明快易懂，比如“**停车场空位率为**％”，这些处理在图7中的网络通信模块27中完成。Once it is detected that the parking space parking situation in the parking lot has changed or the parking space has been reserved, the parking space occupancy situation in the parking lot will be recalculated in the parking spaceinformation update module 24 in FIG. Good data preparation. In the parking spaceinformation publishing module 25, there are mainly two tasks to be completed. One task is to guide and issue parking spaces for vehicles in the field. The information is displayed on the large display screen placed at the entrance of the parking lot Parking space, so that the user of the parking lot can quickly find the parking space to be parked; Another task is to guide and publish the number of parking spaces that can be parked for off-site vehicles, combined with the geographic location of the relevant parking lot stored in the parking lotdata storage information 28 and other information, the use status of the parking lot can be provided to the driver visually on the variable information display board or audibly through broadcasting at any time. The usage status of the parking lot is updated on the guidance information board set on the roadside, as shown in Figure 6. The information displayed on the guidance information board should be clear and easy to understand, such as "** parking lot vacancy rate **%", these processes are shown in Figure 6 Completed in thenetwork communication module 27 in 7.

实施例2Example 2

参照图1～图8，本实施例的颜色空间转化处理模块，用于将图像数据读取模块采集的图像从RGB色彩空间转化到(Cr，Cb)空间颜色模型，转换的计算式为(19)：Referring to Fig. 1～Fig. 8, the color space conversion processing module of the present embodiment is used for converting the image collected by the image data reading module from the RGB color space to the (Cr, Cb) space color model, and the conversion formula is (19 ):

Y＝0.29990*R+0.5870*G+0.1140*BY＝0.29990*R+0.5870*G+0.1140*B

Cr＝0.5000*R-0.4187*G-0.0813*B+128Cr＝0.5000*R-0.4187*G-0.0813*B+128

Cb＝-0.1787*R-0.3313*G+0.5000*B+128       (19)Cb＝-0.1787*R-0.3313*G+0.5000*B+128 (19)

上式中，Y代表(Cr，Cb)空间颜色模型的亮度，Cr、Cb是(Cr，Cb)空间颜色模型的两个彩色分量，表示色差；R表示RGB色彩空间的红色；G表示RGB色彩空间的绿色；B表示RGB色彩空间的蓝色。In the above formula, Y represents the brightness of the (Cr, Cb) space color model, Cr and Cb are the two color components of the (Cr, Cb) space color model, representing the color difference; R represents the red color of the RGB color space; G represents the RGB color The green of the space; B represents the blue of the RGB color space.

其余结构和工作过程与实施例1相同。All the other structures and working process are identical withembodiment 1.

Claims (5)

Translated fromChinese

1、一种基于全方位计算机视觉的电子泊车诱导系统，其特征在于：该电子泊车诱导系统包括微处理器、用于监测停车场内停放车辆情况的全方位视觉传感器、用于与外界通信的通信模块，所述的全方位视觉传感器安装在待监测的停车场的上部；1. An electronic parking guidance system based on omnidirectional computer vision, characterized in that: the electronic parking guidance system includes a microprocessor, an omnidirectional visual sensor for monitoring the situation of vehicles parked in the parking lot, and a communication with the outside world A communication module for communication, the omnidirectional vision sensor is installed on the top of the parking lot to be monitored;所述的全方位视觉传感器包括用于反射监控领域中物体的外凸反射镜面、透明圆柱体、摄像头，所述的外凸反射镜面朝下，所述的透明圆柱体支撑外凸反射镜面，黑色圆锥体固定在折反射镜面外凸部的中心，用于拍摄外凸反射镜面上成像体的摄像头位于透明圆柱体的内部，摄像头位于外凸反射镜面的虚焦点上；The omni-directional vision sensor includes a convex reflective mirror used to reflect objects in the monitoring field, a transparent cylinder, and a camera. The convex reflective mirror faces downward, and the transparent cylinder supports the convex reflective mirror, black The cone is fixed at the center of the convex portion of the catadioptric mirror, the camera for shooting the imaging body on the convex mirror is located inside the transparent cylinder, and the camera is located on the virtual focal point of the convex mirror;所述的微处理器包括：Described microprocessor comprises:图像数据读取模块，用于读取从视觉传感器传过来的停车场内车位的图像信息；The image data reading module is used to read the image information of the parking spaces in the parking lot transmitted from the visual sensor;图像数据文件存储模块，用于将读进来的图像信息通过文件方式保存在存储单元中；The image data file storage module is used to save the read image information in the storage unit by means of a file;虚拟车位框设置模块，用于将读取的整个停车场的图像信息，按照车位分布情况设置与实际车位一一对应的虚拟车位框，并保存基准参考图像；The virtual parking space frame setting module is used to set the virtual parking space frame corresponding to the actual parking space one-to-one according to the parking space distribution of the image information of the entire parking lot read, and save the benchmark reference image;传感器标定模块，用于对全方位视觉传感器的参数进行标定，建立空间的实物图像与所获得的视频图像的线性对应关系；The sensor calibration module is used to calibrate the parameters of the omnidirectional visual sensor, and establish the linear correspondence between the physical image of the space and the obtained video image;虚拟车框检测模块，用于对各个虚拟车位框，将所获得的当前帧现场视频图像与基准参考图像进行差值运算，图像相减的计算公式如式(17)表示：The virtual car frame detection module is used for each virtual parking space frame, and the obtained current frame live video image and the benchmark reference image are carried out difference calculation, and the calculation formula of image subtraction is expressed as formula (17):f_d(X，t₀，t_i)＝f(X，t_i)-f(X，t₀)        (17)f_d (X, t₀ , t_i )=f(X, t_i )-f(X, t₀ ) (17)上式中，f_d(X，t₀，t_i)是实时拍摄到图像与基准参考图像间进行图像相减的结果；f(X，t_i)是实时拍摄到图像；f(X，t₀)是基准参考图像；In the above formula, f_d (X, t₀ , t_i ) is the result of image subtraction between the image captured in real time and the benchmark reference image; f(X, t_i ) is the image captured in real time; f(X, t₀ ) is the benchmark reference image;如f_d(X，t₀，t_i)≥阈值时，判定为可疑有车事件；If f_d (X, t₀ , t_i )≥threshold, it is judged as a suspicious car-occupancy event;如f_d(X，t₀，t_i)＜阈值时，判定为无可疑有车事件；If f_d (X, t₀ , t_i )<threshold value, it is determined that there is no suspicious vehicle occurrence;连通区域计算模块，用于在判定有可疑有车事件后，对当前图像进行标记，像素灰度为0的小区表示此小区无可疑有车，像素灰度为1则表示此小区有可疑有车，计算当前图像中的像素是否与当前像素周围相邻的某一个点的像素相等，如灰度相等判断为具有连通性，将所有具有连通性的像素作为一个连通区域；The connected area calculation module is used to mark the current image after determining that there is a suspicious car event. A cell with a pixel grayscale of 0 means that there is no suspicious car in this cell, and a pixel grayscale of 1 means that there is a suspicious car in this cell , to calculate whether the pixel in the current image is equal to the pixel of a certain point adjacent to the current pixel, if the gray level is equal, it is judged to have connectivity, and all the connected pixels are regarded as a connected region;车辆判断模块，用于根据得到的连通区域，统计求出连通区域面积Si，并将连通区域面积与预设的阈值做比较：The vehicle judging module is used to statistically calculate the area Si of the connected area according to the obtained connected area, and compare the area of the connected area with a preset threshold:如连通区域面积Si大于阈值S阈，判定为在该车位上有车；If the area Si of the connected area is greater than the threshold S threshold, it is determined that there is a car in the parking space;如连通区域面积Si小于阈值S阈，判定为在该车位上没有车；If the area Si of the connected area is smaller than the threshold S threshold, it is determined that there is no car in the parking space;车位信息发布模块，用于根据各个虚拟车位框中的车辆判断，得到停车场内的车位占用信息，通过通信模块发布车位占用信息。The parking space information publishing module is used to obtain the parking space occupancy information in the parking lot according to the judgment of the vehicles in each virtual parking space frame, and release the parking space occupancy information through the communication module.2、如权利要求1所述的基于全方位计算机视觉的电子泊车诱导系统，其特征在于：所述的微处理器还包括：2. The electronic parking guidance system based on all-round computer vision according to claim 1, characterized in that: said microprocessor also includes:车位信息更新模块，用于根据当前监测的车位信息，跟上一次统计的车位信息做比较，如果车位占用信息有变化，更新发布车位占用信息。The parking space information update module is used to compare the current monitored parking space information with the last statistical parking space information, and update and release the parking space occupancy information if there is a change in the parking space occupancy information.3、如权利要求2所述的基于全方位计算机视觉的电子泊车诱导系统，其特征在于：所述的微处理器还包括：3. The electronic parking guidance system based on omnidirectional computer vision according to claim 2, characterized in that: said microprocessor also includes:车位预约处理模块，用于管理人员根据预约车位情况预先设置车位预约情况，并将预约信息输入到车位信息发布模块，综合判断车位占用信息。The parking space reservation processing module is used for management personnel to pre-set the parking space reservation situation according to the reserved parking space situation, and input the reservation information into the parking space information release module to comprehensively judge the parking space occupancy information.4、如权利要求1-3之一所述的基于全方位计算机视觉的电子泊车诱导系统，其特征在于：所述的微处理器还包括：4. The electronic parking guidance system based on omnidirectional computer vision according to any one of claims 1-3, characterized in that: the microprocessor further includes:颜色空间转化处理模块，用于将图像数据读取模块采集的图像从RGB色彩空间转化到HSI色彩空间，转换的计算式为(18)：The color space conversion processing module is used to convert the image collected by the image data reading module from the RGB color space to the HSI color space, and the conversion calculation formula is (18):$\mathrm{<span><span>I</span>I</span>}<span><span>=</span>=</span>\frac{\mathrm{<span><span>R</span>R</span>}<span><span>+</span>+</span>\mathrm{<span><span>G</span>G</span>}<span><span>+</span>+</span>\mathrm{<span><span>B</span>B</span>}}{\mathrm{<span><span>3</span>3</span>}}<span><span>-</span>-</span><span><span>-</span>-</span><span><span>-</span>-</span><span><span>(</span>(</span>\mathrm{<span><span>18</span>18</span>}<span><span>)</span>)</span>$$\mathrm{<span><span>H</span>h</span>}<span><span>=</span>=</span>\frac{\mathrm{<span><span>1</span>1</span>}}{\mathrm{<span><span>360</span>360</span>}}<span><span>[</span>[</span>\mathrm{<span><span>90</span>90</span>}<span><span>-</span>-</span>\mathrm{<span><span>Arc</span>Arc</span>}\mathrm{<span><span>tan</span>the\; tan</span>}<span><span>(</span>(</span>\frac{\mathrm{<span><span>F</span>f</span>}}{\sqrt{\mathrm{<span><span>3</span>3</span>}}}<span><span>)</span>)</span><span><span>+</span>+</span><span><span>\{</span>\{</span>\mathrm{<span><span>0</span>0</span>}<span><span>,</span>,</span>\mathrm{<span><span>G</span>G</span>}<span><span>></span>></span>\mathrm{<span><span>B</span>B</span>}<span><span>;</span>;</span>\mathrm{<span><span>180</span>180</span>}<span><span>,</span>,</span>\mathrm{<span><span>G</span>G</span>}<span><span><</span><</span>\mathrm{<span><span>B</span>B</span>}<span><span>\}</span>\}</span><span><span>]</span>]</span>$$\mathrm{<span><span>S</span>S</span>}<span><span>=</span>=</span>\mathrm{<span><span>1</span>1</span>}<span><span>-</span>-</span><span><span>\{</span>\{</span>\frac{\mathrm{<span><span>min</span>min</span>}<span><span>(</span>(</span>\mathrm{<span><span>R</span>R</span>}<span><span>,</span>,</span>\mathrm{<span><span>G</span>G</span>}<span><span>,</span>,</span>\mathrm{<span><span>B</span>B</span>}<span><span>)</span>)</span>}{\mathrm{<span><span>I</span>I</span>}}<span><span>\}</span>\}</span>$其中，$F=\frac{2R-G-B}{G-B}$in,$f=\frac{2R-G-B}{G-B}$上式中，H为HSI色彩空间的色调，S为HSI色彩空间的饱和度，I为HSI色彩空间的亮度，R为RGB色彩空间的红色；G为RGB色彩空间的绿色；B为RGB色彩空间的蓝色；In the above formula, H is the hue of the HSI color space, S is the saturation of the HSI color space, I is the brightness of the HSI color space, R is the red of the RGB color space; G is the green of the RGB color space; B is the RGB color space the blue;颜色空间转化处理模块的输入端连接所述的虚拟车框检测模块。The input end of the color space conversion processing module is connected to the virtual vehicle frame detection module.5、如权利要求1-3之一所述的基于全方位计算机视觉的电子泊车诱导系统，其特征在于：所述的微处理器还包括：5. The electronic parking guidance system based on omnidirectional computer vision according to any one of claims 1-3, characterized in that: the microprocessor further includes:颜色空间转化处理模块，用于将图像数据读取模块采集的图像从RGB色彩空间转化到(Cr，Cb)空间颜色模型，转换的计算式为(19)：The color space conversion processing module is used to convert the image collected by the image data reading module from the RGB color space to the (Cr, Cb) space color model, and the calculation formula for conversion is (19):Y＝0.29990*R+0.5870*G+0.1140*BY＝0.29990*R+0.5870*G+0.1140*BCr＝0.5000*R-0.4187*G-0.0813*B+128Cr＝0.5000*R-0.4187*G-0.0813*B+128Cb＝-0.1787*R-0.3313*G+0.5000*B+128    (19)Cb＝-0.1787*R-0.3313*G+0.5000*B+128 (19)上式中，Y代表(Cr，Cb)空间颜色模型的亮度，Cr、Cb是(Cr，Cb)空间颜色模型的两个彩色分量，表示色差；R表示RGB色彩空间的红色；G表示RGB色彩空间的绿色；B表示RGB色彩空间的蓝色。In the above formula, Y represents the brightness of the (Cr, Cb) space color model, Cr and Cb are the two color components of the (Cr, Cb) space color model, representing the color difference; R represents the red color of the RGB color space; G represents the RGB color The green of the space; B represents the blue of the RGB color space.

Images