技术领域Technical Field
本发明涉及图像处理技术领域，具体涉及基于图像处理的建筑工程质量检测方法。The present invention relates to the technical field of image processing, and in particular to a construction engineering quality detection method based on image processing.
背景技术Background technique
为了保证建筑工程质量，常利用计算机视觉技术对多种类型的建筑工程质量缺陷进行检测，建筑工程质量缺陷的类型包括检测结构缺陷、表面裂缝、施工误差等；该方法可以通过自动化和数字化的手段快速、精确地识别潜在的质量问题，提高建筑工程的质量管理效率。但是对施工场地建筑质量进行检测时，采集环境中的太阳光线等会对检测设备造成影响，检测设备对施工现场进行图像数据的采集时，采集的图像可能会存在部分区域过曝或者过暗的问题，导致采集的图像显示质量较差，检测系统对较差显示质量的图像进行分析时结果可能存在偏差或者出现严重的分析错误，从而影响建筑工程质量的最终检测结果。In order to ensure the quality of construction projects, computer vision technology is often used to detect various types of construction project quality defects, including structural defects, surface cracks, construction errors, etc. This method can quickly and accurately identify potential quality problems through automation and digital means, and improve the quality management efficiency of construction projects. However, when inspecting the quality of construction sites, the sunlight in the collection environment will affect the detection equipment. When the detection equipment collects image data from the construction site, some areas of the collected images may be overexposed or too dark, resulting in poor display quality of the collected images. When the detection system analyzes images with poor display quality, the results may be biased or serious analysis errors may occur, thereby affecting the final detection results of the construction project quality.
在建筑工程质量检测过程中，对于采集图像部分区域存在过曝或过暗影响图像显示质量的问题，现有的局部对比度增强算法可以很好地改善图像的视觉效果，该算法能够增加图像中像素的对比度，使得过暗或过曝的区域更加清晰可见。但是在该算法对采集的图像进行增强时，增强的增益参数则对图像的增强效果有着重要影响，较大的增益参数会产生更强的增益效果，增强对比度和细节，但可能导致过度增强损失细节，较小的增益参数则会减小增益效果，使得图像变得柔和，增强效果不够明显，导致建筑工程质量的检测结果不准确。In the process of building quality inspection, some areas of the captured image are overexposed or too dark, which affects the image display quality. The existing local contrast enhancement algorithm can improve the visual effect of the image. The algorithm can increase the contrast of the pixels in the image, making the dark or overexposed areas more clearly visible. However, when the algorithm enhances the captured image, the enhanced gain parameter has an important influence on the image enhancement effect. A larger gain parameter will produce a stronger gain effect, enhance the contrast and details, but may cause excessive enhancement and loss of details. A smaller gain parameter will reduce the gain effect, making the image soft and the enhancement effect not obvious enough, resulting in inaccurate inspection results of the building quality.
发明内容Summary of the invention
本发明提供基于图像处理的建筑工程质量检测方法，以解决现有的问题。The present invention provides a construction engineering quality detection method based on image processing to solve the existing problems.
本发明的基于图像处理的建筑工程质量检测方法采用如下技术方案：The construction engineering quality detection method based on image processing of the present invention adopts the following technical solutions:
本发明一个实施例提供了基于图像处理的建筑工程质量检测方法，该方法包括以下步骤：An embodiment of the present invention provides a construction project quality detection method based on image processing, the method comprising the following steps:
获取建筑工程图像；Acquire construction engineering images;
根据建筑工程图像得到灰度筛选序列；根据灰度筛选序列得到平均峰值距离；根据灰度筛选序列和平均峰值距离得到分块参数；根据分块参数将建筑工程图像分为若干个初始分块；A grayscale screening sequence is obtained according to the construction engineering image; an average peak distance is obtained according to the grayscale screening sequence; a block parameter is obtained according to the grayscale screening sequence and the average peak distance; and the construction engineering image is divided into a number of initial blocks according to the block parameter;
在建筑工程图像中，根据每个初始分块内像素点的梯度幅值，获得每个初始分块的背景类簇和内容类簇；根据每个初始分块的背景类簇和内容类簇，得到每个初始分块的初始增益参数；根据每个初始分块的初始增益参数对建筑工程图像进行增强，得到初始增强图像；In the construction engineering image, the background cluster and the content cluster of each initial block are obtained according to the gradient amplitude of the pixel points in each initial block; the initial gain parameter of each initial block is obtained according to the background cluster and the content cluster of each initial block; the construction engineering image is enhanced according to the initial gain parameter of each initial block to obtain an initial enhanced image;
在初始增强图像中，根据分块参数将初始增强图像分为若干个第一分块；根据每个第一分块内像素点的梯度幅值，得到每个第一分块的背景特征类簇和内容特征类簇；根据每个第一分块的背景特征类簇和内容特征类簇以及建筑工程图像中每个初始分块的内容类簇，得到初始增益参数的修正系数；In the initial enhanced image, the initial enhanced image is divided into a plurality of first blocks according to the block parameters; the background feature cluster and the content feature cluster of each first block are obtained according to the gradient amplitude of the pixel points in each first block; the correction coefficient of the initial gain parameter is obtained according to the background feature cluster and the content feature cluster of each first block and the content cluster of each initial block in the architectural engineering image;
根据初始增益参数的修正系数对建筑工程图像进行增强，得到最终增强图像；The architectural engineering image is enhanced according to the correction coefficient of the initial gain parameter to obtain a final enhanced image;
根据最终增强图像，得到建筑工程质量检测的结果。According to the final enhanced image, the result of construction project quality inspection is obtained.
进一步地，所述根据建筑工程图像得到灰度筛选序列，包括的具体步骤如下：Furthermore, the grayscale screening sequence is obtained according to the construction engineering image, and the specific steps include the following:
获取建筑工程图像的灰度直方图，将灰度直方图中每个灰度级的像素点数量按照灰度级依次排列，得到建筑工程图像的灰度序列，将灰度序列中第一个元素值到第一个非零元素值的区间，以及最后一个非零元素值到最后一个元素值的区间内所有元素值从灰度序列中删除，得到灰度筛选序列。A grayscale histogram of a construction engineering image is obtained, and the number of pixels of each grayscale level in the grayscale histogram is arranged in sequence according to the grayscale level to obtain a grayscale sequence of the construction engineering image. All element values in the interval from the first element value to the first non-zero element value in the grayscale sequence, and in the interval from the last non-zero element value to the last element value are deleted from the grayscale sequence to obtain a grayscale screening sequence.
进一步地，所述根据灰度筛选序列得到平均峰值距离，包括的具体步骤如下：Furthermore, the step of obtaining the average peak distance according to the grayscale screening sequence includes the following specific steps:
对于建筑工程图像的灰度筛选序列中的任意一个元素值，若所述元素值大于与其相邻的两个元素值，将所述元素值记为局部峰值；将每相邻两个局部峰值之间的距离分别记为一个峰值距离，计算所有峰值距离的平均值，记作平均峰值距离。For any element value in the grayscale screening sequence of the construction engineering image, if the element value is greater than the two adjacent element values, the element value is recorded as a local peak; the distance between each two adjacent local peaks is recorded as a peak distance, and the average value of all peak distances is calculated and recorded as the average peak distance.
进一步地，所述根据灰度筛选序列和平均峰值距离得到分块参数，包括的具体计算方式如下：Furthermore, the block parameters are obtained according to the grayscale screening sequence and the average peak distance, including the specific calculation method as follows:
式中，为分块参数；表示平均峰值距离；表示灰度筛选序列的第i个元素值；表示灰度筛选序列中所有元素值的均值；表示灰度筛选序列中元素值的数量；表示以自然常数为底的指数函数；表示预设的分块调节参数；表示向上取整符号；表示绝对值函数。In the formula, is the block parameter; represents the average peak distance; Represents the i-th element value of the grayscale screening sequence; Represents the mean value of all element values in the grayscale screening sequence; Represents the number of element values in the grayscale screening sequence; represents an exponential function with a natural constant as base; Indicates the preset block adjustment parameters; Indicates the rounding up symbol; represents the absolute value function.
进一步地，所述根据每个初始分块内像素点的梯度幅值，获得每个初始分块的背景类簇和内容类簇，包括的具体步骤如下：Furthermore, the step of obtaining the background cluster and the content cluster of each initial block according to the gradient amplitude of the pixel points in each initial block includes the following specific steps:
对任意一个初始分块内的像素点进行聚类，距离度量采用像素点的梯度幅值之间的差值绝对值，得到该初始分块的两个类簇；计算每个类簇的像素点的梯度幅值的平均值，将梯度幅值的平均值最小的类簇记为该初始分块的背景类簇，梯度幅值的平均值最大的类簇记为该初始分块的内容类簇。The pixels in any initial block are clustered, and the distance measurement uses the absolute value of the difference between the gradient amplitudes of the pixels to obtain two clusters of the initial block; the average gradient amplitude of the pixels in each cluster is calculated, and the cluster with the smallest average gradient amplitude is recorded as the background cluster of the initial block, and the cluster with the largest average gradient amplitude is recorded as the content cluster of the initial block.
进一步地，所述根据每个初始分块的背景类簇和内容类簇，得到每个初始分块的初始增益参数，包括的具体计算方式如下：Furthermore, the initial gain parameter of each initial block is obtained according to the background cluster and the content cluster of each initial block, and the specific calculation method is as follows:
式中，为建筑工程图像中第个初始分块的初始增益参数；表示建筑工程图像中第个初始分块的内容类簇内所有像素点的梯度幅值的平均值；、分别表示建筑工程图像中第个初始分块的背景类簇内所有像素点的灰度值的平均值和内容类簇内所有像素点的灰度值的平均值；表示建筑工程图像中第个初始分块的背景类簇内所有像素点的梯度幅值的平均值和内容类簇内所有像素点的梯度幅值的平均值的差值绝对值；表示归一化函数；为预设的增益系数。In the formula, For the construction engineering image The initial gain parameters of the initial blocks; Indicates the first The average value of the gradient amplitude of all pixels in the content cluster of the initial block; , They represent the first The average grayscale value of all pixels in the background cluster of the initial block and the average grayscale value of all pixels in the content cluster; Indicates the first The absolute value of the difference between the average value of the gradient amplitude of all pixels in the background cluster of the initial block and the average value of the gradient amplitude of all pixels in the content cluster; represents the normalization function; is the preset gain factor.
进一步地，所述根据每个初始分块的初始增益参数对建筑工程图像进行增强，得到初始增强图像，包括的具体步骤如下：Furthermore, the construction engineering image is enhanced according to the initial gain parameter of each initial block to obtain an initial enhanced image, and the specific steps include the following:
在建筑工程图像中，计算任意一个初始分块内所有像素点的灰度值的平均值，记为该初始分块的局部均值；将该初始分块内每个像素点的灰度值与所述局部均值的差值，记为每个像素点的局部对比度；计算任意一个像素点的局部对比度与该初始分块的初始增益参数的乘积，将所述乘积与所述局部均值的和值，记作该像素点的初始增强灰度值；In the architectural engineering image, the average value of the grayscale values of all pixels in any initial block is calculated and recorded as the local mean value of the initial block; the difference between the grayscale value of each pixel in the initial block and the local mean value is recorded as the local contrast of each pixel; the product of the local contrast of any pixel and the initial gain parameter of the initial block is calculated, and the sum of the product and the local mean value is recorded as the initial enhanced grayscale value of the pixel;
获取每个初始分块内每个像素点的初始增强灰度值；根据每个像素点的初始增强灰度值对建筑工程图像中每个像素点的灰度值进行替换，得到初始增强图像。The initial enhanced grayscale value of each pixel in each initial block is obtained; the grayscale value of each pixel in the building engineering image is replaced according to the initial enhanced grayscale value of each pixel to obtain an initial enhanced image.
进一步地，所述根据每个第一分块的背景特征类簇和内容特征类簇以及建筑工程图像中每个初始分块的内容类簇，得到初始增益参数的修正系数，包括的具体计算方式如下：Furthermore, the correction coefficient of the initial gain parameter is obtained according to the background feature cluster and the content feature cluster of each first block and the content cluster of each initial block in the construction engineering image, including the specific calculation method as follows:
式中，表示初始增强图像中第个第一分块的初始增益参数的修正系数；表示初始增强图像中第个第一分块内的内容特征类簇中所有像素点的梯度幅值的均值；表示建筑工程图像中第个初始分块的内容类簇中所有像素点的梯度幅值的均值；、分别表示初始增强图像中第个第一分块内的背景特征类簇内所有像素点的灰度值的平均值以及内容特征类簇内所有像素点的灰度值的平均值；为以自然常数为底的指数函数；为绝对值函数。In the formula, represents the first A correction coefficient of the initial gain parameter of the first block; represents the first The mean of the gradient amplitudes of all pixels in the content feature cluster in the first block; Indicates the first The mean of the gradient amplitudes of all pixels in the content cluster of the initial block; , They represent the first The average grayscale value of all pixels in the background feature cluster in the first block and the average grayscale value of all pixels in the content feature cluster; is an exponential function with a natural constant as base; is the absolute value function.
进一步地，所述根据初始增益参数的修正系数对建筑工程图像进行增强，得到最终增强图像，包括的具体步骤如下：Furthermore, the construction engineering image is enhanced according to the correction coefficient of the initial gain parameter to obtain a final enhanced image, which includes the following specific steps:
在建筑工程图像中，计算每个像素点的局部对比度与像素点所在初始分块的最终增益参数的乘积，将该乘积与像素点所在初始分块的局部均值的和值，记作每个像素点的最终增强灰度值；In the architectural engineering image, the product of the local contrast of each pixel and the final gain parameter of the initial block where the pixel is located is calculated, and the sum of the product and the local mean of the initial block where the pixel is located is recorded as the final enhanced grayscale value of each pixel;
获取每个初始分块内每个像素点的最终增强灰度值；根据每个像素点的最终增强灰度值对建筑工程图像中每个像素点的灰度值进行替换，得到最终增强图像。The final enhanced grayscale value of each pixel in each initial block is obtained; the grayscale value of each pixel in the building engineering image is replaced according to the final enhanced grayscale value of each pixel to obtain a final enhanced image.
进一步地，所述根据最终增强图像，得到建筑工程质量检测的结果，包括的具体步骤如下：Furthermore, the method of obtaining the result of the construction project quality inspection based on the final enhanced image includes the following specific steps:
将最终增强图像输入深度神经网络，得到最终增强图像中的隔离护栏区域；将当前天的最终增强图像中的隔离护栏区域像素点的个数与前一天的最终增强图像中的隔离护栏区域像素点的个数的比值，记作隔离护栏存在率；当隔离护栏存在率小于预设的存在阈值时，判断为隔离护栏损坏；当隔离护栏存在率大于或等于预设的存在阈值时，判断为隔离护栏正常。The final enhanced image is input into the deep neural network to obtain the isolation fence area in the final enhanced image; the ratio of the number of pixels in the isolation fence area in the final enhanced image of the current day to the number of pixels in the isolation fence area in the final enhanced image of the previous day is recorded as the isolation fence existence rate; when the isolation fence existence rate is less than the preset existence threshold When the isolation guardrail existence rate is greater than or equal to the preset existence threshold , it is judged that the isolation guardrail is normal.
本发明的技术方案的有益效果是：本发明根据灰度筛选序列和平均峰值距离得到分块参数，将建筑工程图像进行分块，对不同的分块进行自适应增强，提高了建筑工程图像的增强效果；根据每个初始分块的背景类簇和内容类簇，得到每个初始分块的初始增益参数，初步判断了每个分块的增强程度，为建筑工程图像的增强提高了数据依据；根据每个第一分块的背景特征类簇和内容特征类簇以及建筑工程图像中每个初始分块的内容类簇，得到初始增益参数的修正系数，提高了每个分块的增强程度的准确度，进一步提高了建筑工程图像的增强效果。至此本发明通过准确可信的初始增益参数的修正系数，对建筑工程图像进行增强，得到增强效果更好的最终增强图像，提高了建筑工程质量检测结果的准确性。The beneficial effects of the technical solution of the present invention are as follows: the present invention obtains the block parameters according to the grayscale screening sequence and the average peak distance, divides the construction engineering image into blocks, and adaptively enhances different blocks, thereby improving the enhancement effect of the construction engineering image; according to the background cluster and content cluster of each initial block, the initial gain parameter of each initial block is obtained, and the enhancement degree of each block is preliminarily judged, thereby improving the data basis for the enhancement of the construction engineering image; according to the background feature cluster and content feature cluster of each first block and the content cluster of each initial block in the construction engineering image, the correction coefficient of the initial gain parameter is obtained, thereby improving the accuracy of the enhancement degree of each block, and further improving the enhancement effect of the construction engineering image. So far, the present invention enhances the construction engineering image through the correction coefficient of the accurate and reliable initial gain parameter, obtains the final enhanced image with better enhancement effect, and improves the accuracy of the construction engineering quality detection result.
附图说明BRIEF DESCRIPTION OF THE DRAWINGS
为了更清楚地说明本发明实施例或现有技术中的技术方案，下面将对实施例或现有技术描述中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本发明的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其他的附图。In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
图1为本发明基于图像处理的建筑工程质量检测方法的步骤流程图。FIG1 is a flow chart showing the steps of a construction project quality detection method based on image processing according to the present invention.
具体实施方式Detailed ways
为了更进一步阐述本发明为达成预定发明目的所采取的技术手段及功效，以下结合附图及较佳实施例，对依据本发明提出的基于图像处理的建筑工程质量检测方法，其具体实施方式、结构、特征及其功效，详细说明如下。在下述说明中，不同的“一个实施例”或“另一个实施例”指的不一定是同一实施例。此外，一或多个实施例中的特定特征、结构或特点可由任何合适形式组合。In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the construction engineering quality detection method based on image processing proposed by the present invention, its specific implementation method, structure, characteristics and effects, in conjunction with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.
除非另有定义，本文所使用的所有的技术和科学术语与属于本发明的技术领域的技术人员通常理解的含义相同。Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
下面结合附图具体的说明本发明所提供的基于图像处理的建筑工程质量检测方法的具体方案。The specific scheme of the construction project quality detection method based on image processing provided by the present invention is described in detail below with reference to the accompanying drawings.
请参阅图1，其示出了本发明一个实施例提供的基于图像处理的建筑工程质量检测方法的步骤流程图，该方法包括以下步骤：Please refer to FIG1 , which shows a flowchart of a construction engineering quality detection method based on image processing provided by an embodiment of the present invention. The method comprises the following steps:
步骤S001：获取建筑工程图像。Step S001: Acquire construction project images.
需要说明的是，本实施例的主要目的是对建筑工程质量检测系统所采集到显示质量较差时的图像进行增强处理，以提高检测系统的准确性。It should be noted that the main purpose of this embodiment is to enhance the images collected by the construction engineering quality detection system when the display quality is poor, so as to improve the accuracy of the detection system.
具体的，在建筑工程场地布置工业相机，将工业相机镜头对准需要监测的建筑工程区域，每天对需要监测的建筑工程区域进行图像采集，将采集得到的RGB图像进行灰度化，将灰度化后的图像记为建筑工程图像。Specifically, an industrial camera is arranged at a construction site, and the lens of the industrial camera is aimed at the construction area that needs to be monitored. Image collection is performed on the construction area that needs to be monitored every day, and the collected RGB image is grayscaled, and the grayscale image is recorded as a construction image.
步骤S002：根据建筑工程图像得到灰度筛选序列；根据灰度筛选序列得到平均峰值距离；根据灰度筛选序列和平均峰值距离得到分块参数；根据分块参数将建筑工程图像分为若干个初始分块。Step S002: obtaining a grayscale screening sequence according to the construction engineering image; obtaining an average peak distance according to the grayscale screening sequence; obtaining a block parameter according to the grayscale screening sequence and the average peak distance; and dividing the construction engineering image into a number of initial blocks according to the block parameter.
需要说明的是，在建筑工程图像中，不同区域的对比度是不同的，此现象主要与采集图像时光线的强度以及建筑物的遮挡有关。建筑工程图像中不同区域的对比度不同，在对建筑工程图像利用局部对比度增强算法进行增强时，不同区域的增益参数不同。为了获取不同区域更加准确的增益参数，则需要将建筑工程图像进行分块处理。通过对每一分块内的区域进行分析，从而得到不同区域的增强的增益参数。在对图像进行分块时，图像的各个内容区间整体的对比度越小，分块数量越多，使用更多的分块可以更精细地调整每个区域的增益参数，以达到更好的对比度增强效果。It should be noted that in architectural images, the contrast of different areas is different. This phenomenon is mainly related to the intensity of light and the occlusion of buildings when the image is collected. The contrast of different areas in architectural images is different. When the architectural images are enhanced using the local contrast enhancement algorithm, the gain parameters of different areas are different. In order to obtain more accurate gain parameters for different areas, the architectural images need to be processed in blocks. By analyzing the areas within each block, the enhanced gain parameters of different areas are obtained. When the image is divided into blocks, the smaller the overall contrast of each content interval of the image, the more blocks there are, and the more blocks are used, the gain parameters of each area can be adjusted more finely to achieve better contrast enhancement effects.
具体的，获取建筑工程图像的灰度直方图，将灰度直方图中每个灰度级的像素点数量按照灰度级依次排列，得到建筑工程图像的灰度序列，将灰度序列中第一个元素值到第一个非零元素值的区间，以及最后一个非零元素值到最后一个元素值的区间内所有元素值从灰度序列中删除，得到灰度筛选序列；对于建筑工程图像的灰度筛选序列中的任意一个元素值，若该元素值大于与其相邻的两个元素值，将该元素值记为局部峰值；将每相邻两个局部峰值之间的距离分别记为一个峰值距离，计算所有峰值距离的平均值，记作平均峰值距离，特别说明的是，灰度筛选序列中第一个元素值及最后一个元素值不进行局部峰值的判断；分块参数的计算方式为：Specifically, a grayscale histogram of a construction engineering image is obtained, and the number of pixels of each grayscale level in the grayscale histogram is arranged in sequence according to the grayscale level to obtain a grayscale sequence of the construction engineering image. All element values in the interval from the first element value to the first non-zero element value in the grayscale sequence, and in the interval from the last non-zero element value to the last element value are deleted from the grayscale sequence to obtain a grayscale screening sequence; for any element value in the grayscale screening sequence of the construction engineering image, if the element value is greater than the two adjacent element values, the element value is recorded as a local peak value; the distance between each two adjacent local peak values is recorded as a peak distance, and the average value of all peak distances is calculated and recorded as the average peak distance. It is particularly noted that the first element value and the last element value in the grayscale screening sequence are not judged as local peaks; the calculation method of the block parameter is:
式中，为分块参数；表示平均峰值距离；表示灰度筛选序列的第i个元素值；表示灰度筛选序列中所有元素值的均值；表示灰度筛选序列中元素值的数量；表示以自然常数为底的指数函数；表示预设的分块调节参数，=10，以此为例进行叙述，其它实施方式中可设置为其它值，本实施例不进行限定；表示向上取整符号；表示绝对值函数。In the formula, is the block parameter; represents the average peak distance; Represents the i-th element value of the grayscale screening sequence; Represents the mean value of all element values in the grayscale screening sequence; Represents the number of element values in the grayscale screening sequence; represents an exponential function with a natural constant as base; Indicates the preset block adjustment parameters, =10, which is described as an example, and other values may be set in other implementation modes, which are not limited in this embodiment; Indicates the rounding up symbol; represents the absolute value function.
所需说明的是，的值越大，说明灰度直方图的波峰和波谷间的距离越远，建筑工程图像的整体对比度越高，分块的数量则越少；表示不同灰度级对应像素点数量的分布变化程度，该值越大则说明灰度直方图中不同灰度级对应像素点数量的分布越不均匀，说明建筑工程图像的对比度越大，那么分块时数量越少；的值越大，建筑工程图像的中像素点的灰度值的分布范围越大，建筑工程图像的对比度可能越大，分块数量则越少。It should be noted that The larger the value is, the farther the distance between the peak and the trough of the grayscale histogram is, the higher the overall contrast of the building engineering image is, and the fewer the number of blocks is; Indicates the distribution change degree of the number of pixels corresponding to different gray levels. The larger the value, the more uneven the distribution of the number of pixels corresponding to different gray levels in the grayscale histogram, which means that the greater the contrast of the building engineering image, the smaller the number of blocks. The larger the value of , the larger the distribution range of the grayscale values of the pixels in the construction engineering image, the greater the contrast of the construction engineering image may be, and the smaller the number of blocks.
进一步的，根据分块参数，将建筑工程图像分为个大小均匀的初始分块。Furthermore, according to the block parameters , the construction engineering images are divided into Initial blocks of uniform size.
步骤S003：在建筑工程图像中，根据每个初始分块内像素点的梯度幅值，获得每个初始分块的背景类簇和内容类簇；根据每个初始分块的背景类簇和内容类簇，得到每个初始分块的初始增益参数；根据每个初始分块的初始增益参数对建筑工程图像进行增强，得到初始增强图像。Step S003: In the construction engineering image, according to the gradient amplitude of the pixel points in each initial block, the background cluster and the content cluster of each initial block are obtained; according to the background cluster and the content cluster of each initial block, the initial gain parameter of each initial block is obtained; according to the initial gain parameter of each initial block, the construction engineering image is enhanced to obtain an initial enhanced image.
需要说明的是，不同的分块由于位置不同，其所受光线的影响不同，导致分块内各内容区域之间的对比度不同，不同分块的增益参数也不同。因此，为了获得不同分块对应的增益参数，对于每一分块内的像素点的梯度进行分析；对于同一分块内，梯度较大的像素点则可能为物体结构的边缘像素点，而梯度较小的像素点可能为背景或者物体表面的像素点，而同一分块内，属于同一物体结构区域的像素点的灰度值则可能是比较接近的。It should be noted that different blocks are affected by different light due to their different positions, resulting in different contrasts between content areas within the blocks, and different gain parameters for different blocks. Therefore, in order to obtain the gain parameters corresponding to different blocks, the gradient of the pixels in each block is analyzed; for the same block, the pixel with a larger gradient may be the edge pixel of the object structure, while the pixel with a smaller gradient may be the background or the surface pixel of the object, and the grayscale values of the pixels belonging to the same object structure area in the same block may be relatively close.
具体的，使用Sobel算子计算建筑工程图像中每一个像素点的梯度幅值，Sobel算子计算梯度幅值的过程为公知技术，具体方法在此不做介绍。利用K-means算法对任意一个初始分块内的像素点进行聚类，距离度量采用像素点的梯度幅值之间的差值绝对值，本实施例采用K=2进行叙述，得到该初始分块的两个类簇；K-means算法为公知技术，具体方法在此不做介绍；K-means算法的中文全称为k均值聚类算法，英文全称为k-means clusteringalgorithm。计算每个类簇的像素点的梯度幅值的平均值，将梯度幅值的平均值最小的类簇记为该初始分块的背景类簇，梯度幅值的平均值最大的类簇记为该初始分块的内容类簇；建筑工程图像中第个初始分块的初始增益参数的计算方式如下：Specifically, the Sobel operator is used to calculate the gradient amplitude of each pixel in the construction engineering image. The process of calculating the gradient amplitude by the Sobel operator is a well-known technology, and the specific method is not introduced here. The K-means algorithm is used to cluster the pixels in any initial block. The distance metric uses the absolute value of the difference between the gradient amplitudes of the pixels. This embodiment uses K=2 for description to obtain two clusters of the initial block; the K-means algorithm is a well-known technology, and the specific method is not introduced here; the full name of the K-means algorithm in Chinese is k-means clustering algorithm, and the full name in English is k-means clustering algorithm. Calculate the average value of the gradient amplitude of the pixels in each cluster, and record the cluster with the smallest average value of the gradient amplitude as the background cluster of the initial block, and the cluster with the largest average value of the gradient amplitude as the content cluster of the initial block; in the construction engineering image, The initial gain parameters of the initial blocks are calculated as follows:
式中，为建筑工程图像中第个初始分块的初始增益参数；表示建筑工程图像中第个初始分块的内容类簇内所有像素点的梯度幅值的平均值；、分别表示建筑工程图像中第个初始分块的背景类簇内所有像素点的灰度值的平均值和内容类簇内所有像素点的灰度值的平均值；表示建筑工程图像中第个初始分块的背景类簇内所有像素点的梯度幅值的平均值和内容类簇内所有像素点的梯度幅值的平均值的差值绝对值；表示线性归一化函数，归一化对象为所有初始分块的；为预设的增益系数，=2，本实施例以此为例进行叙述，其它实施方式中可设置为其它值。In the formula, For the construction engineering image The initial gain parameters of the initial blocks; Indicates the first The average value of the gradient amplitude of all pixels in the content cluster of the initial block; , They represent the first The average grayscale value of all pixels in the background cluster of the initial block and the average grayscale value of all pixels in the content cluster; Indicates the first The absolute value of the difference between the average value of the gradient amplitude of all pixels in the background cluster of the initial block and the average value of the gradient amplitude of all pixels in the content cluster; Represents a linear normalization function, the normalized object is all initial blocks ; is the preset gain factor, =2. This embodiment is described by taking this as an example. Other implementation modes may set it to other values.
所需说明的是，越小，说明建筑工程图像中第个初始分块的内容区域像素点的梯度幅值越小，内容区域不够突出，该初始分块的初始增益参数越大；越小，说明建筑工程图像中第个初始分块的背景类簇和内容类簇内像素点的梯度幅值差异越小，内容区域和背景区域之间的对比度较差，初始增益参数则应该越大；表示第个初始分块的背景区域和内容区域之间的总体颜色差异，其值越小，说明该初始分块内背景区域和内容区域之间颜色差异越小，内容区域和背景区域之间对比度可能越差，该初始分块的初始增益参数则越大。It should be noted that The smaller the size, the more The smaller the gradient amplitude of the pixel points in the content area of an initial block, the less prominent the content area is, and the larger the initial gain parameter of the initial block is; The smaller the size, the more The smaller the difference in gradient amplitude between the background cluster and the content cluster of the initial block, the worse the contrast between the content area and the background area, and the larger the initial gain parameter should be. Indicates The smaller the value of the overall color difference between the background area and the content area of an initial block, the smaller the color difference between the background area and the content area in the initial block, the worse the contrast between the content area and the background area may be, and the larger the initial gain parameter of the initial block.
根据上述方法，得到每个初始分块的初始增益参数。According to the above method, the initial gain parameter of each initial block is obtained.
进一步地，基于每个初始分块的初始增益参数利用局部对比度增强算法对建筑工程图像进行增强，将增强后的图像记作初始增强图像；增强的具体过程为：在建筑工程图像中，计算任意一个初始分块内所有像素点的灰度值的平均值，记为该初始分块的局部均值；将该初始分块内每个像素点的灰度值与局部均值的差值，记为每个像素点的局部对比度；计算任意一个像素点的局部对比度与该初始分块的初始增益参数的乘积，将乘积与局部均值的和值，记作该像素点的初始增强灰度值；根据上述方法得到每个初始分块内每个像素点的初始增强灰度值；根据每个像素点的初始增强灰度值对建筑工程图像中每个像素点的灰度值进行替换，得到初始增强图像。Furthermore, the construction engineering image is enhanced by using a local contrast enhancement algorithm based on the initial gain parameters of each initial block, and the enhanced image is recorded as the initial enhanced image; the specific process of enhancement is: in the construction engineering image, the average value of the grayscale values of all pixels in any initial block is calculated, and recorded as the local mean of the initial block; the difference between the grayscale value of each pixel in the initial block and the local mean is recorded as the local contrast of each pixel; the product of the local contrast of any pixel and the initial gain parameter of the initial block is calculated, and the sum of the product and the local mean is recorded as the initial enhanced grayscale value of the pixel; the initial enhanced grayscale value of each pixel in each initial block is obtained according to the above method; the grayscale value of each pixel in the construction engineering image is replaced according to the initial enhanced grayscale value of each pixel to obtain the initial enhanced image.
步骤S004：在初始增强图像中，根据分块参数将初始增强图像分为若干个第一分块；根据每个第一分块内像素点的梯度幅值，得到每个第一分块的背景特征类簇和内容特征类簇；根据每个第一分块的背景特征类簇和内容特征类簇以及建筑工程图像中每个初始分块的内容类簇，得到初始增益参数的修正系数。Step S004: In the initial enhanced image, the initial enhanced image is divided into a number of first blocks according to the block parameters; the background feature clusters and content feature clusters of each first block are obtained according to the gradient amplitude of the pixel points in each first block; the correction coefficient of the initial gain parameter is obtained according to the background feature clusters and content feature clusters of each first block and the content clusters of each initial block in the construction engineering image.
需要说明的是，上述的初始增强增益参数仅根据分块内图像的对比度进行分析得到的，增强效果可能不理想，故需要对增强后分块图像进一步分析，通过对初始增益参数进行调整，从而得到对分块图像更加准确的增强效果。It should be noted that the above initial enhancement gain parameters are obtained only based on the contrast of the image in the block, and the enhancement effect may not be ideal. Therefore, it is necessary to further analyze the enhanced block image and adjust the initial gain parameters to obtain a more accurate enhancement effect on the block image.
具体的，根据分块参数，将初始增强图像分为个大小均匀的第一分块；利用K-means算法对任意一个第一分块内的像素点进行聚类，距离度量采用像素点的梯度幅值之间的差值绝对值，本实施例采用K=2进行叙述，得到该第一分块的两个类簇；计算每个类簇内所有像素点的梯度幅值的平均值，将该第一分块内梯度幅值的平均值最小的类簇记为该第一分块的背景特征类簇，该第一分块内梯度幅值的平均值最大的类簇记为该第一分块的内容特征类簇；初始增强图像中第个第一分块的初始增益参数的修正系数的计算方式为：Specifically, according to the block parameters , the initial enhanced image is divided into first blocks of uniform size; clustering the pixels in any first block using the K-means algorithm, using the absolute value of the difference between the gradient amplitudes of the pixels as the distance metric, and this embodiment uses K=2 for description, to obtain two clusters of the first block; calculating the average value of the gradient amplitudes of all pixels in each cluster, recording the cluster with the smallest average value of the gradient amplitude in the first block as the background feature cluster of the first block, and recording the cluster with the largest average value of the gradient amplitude in the first block as the content feature cluster of the first block; the first pixel in the initial enhanced image is clustered by the K-means algorithm .... The correction coefficient of the initial gain parameter of the first block is calculated as follows:
式中，表示初始增强图像中第个第一分块的初始增益参数的修正系数；表示初始增强图像中第个第一分块内的内容特征类簇中所有像素点的梯度幅值的均值；表示建筑工程图像中第个初始分块的内容类簇中所有像素点的梯度幅值的均值；、分别表示初始增强图像中第个第一分块内的背景特征类簇内所有像素点的灰度值的平均值以及内容特征类簇内所有像素点的灰度值的平均值；为以自然常数为底的指数函数；为绝对值函数。In the formula, represents the first A correction coefficient of the initial gain parameter of the first block; represents the first The mean of the gradient amplitudes of all pixels in the content feature cluster in the first block; Indicates the first The mean of the gradient amplitudes of all pixels in the content cluster of the initial block; , They represent the first The average grayscale value of all pixels in the background feature cluster in the first block and the average grayscale value of all pixels in the content feature cluster; is an exponential function with a natural constant as base; is the absolute value function.
所需说明的是，表示第个初始分块增强前后类簇内像素点梯度幅值的变化程度，其值越小，说明该初始分块内像素点增强后可能属于内容区域但细节特征仍不够明显，初始增益参数的修正系数应该越大；表示第个初始分块增强前后类簇内像素点灰度值的变化程度，其值越大，说明增强后的两个类簇之间的对比度越大，增强的效果越好，初始增益参数的修正系数则越小。It should be noted that Indicates The change degree of the gradient amplitude of the pixels in the cluster before and after the initial block enhancement. The smaller the value, the more likely the pixels in the initial block belong to the content area after enhancement, but the detail features are still not obvious enough, and the larger the correction coefficient of the initial gain parameter should be. Indicates The larger the value is, the greater the contrast between the two clusters after enhancement, the better the enhancement effect is, and the smaller the correction coefficient of the initial gain parameter is.
根据上述方法，得到每个第一分块的初始增益参数的修正系数。According to the above method, the correction coefficient of the initial gain parameter of each first block is obtained.
步骤S005：根据初始增益参数的修正系数对建筑工程图像进行增强，得到最终增强图像。Step S005: enhancing the architectural engineering image according to the correction coefficient of the initial gain parameter to obtain a final enhanced image.
具体的，计算1与初始增强图像中第个第一分块的初始增益参数的修正系数的和值，将和值与建筑工程图像中第个初始分块的初始增益参数的乘积，记作建筑工程图像中第个初始分块的最终增益参数。Specifically, calculate 1 and the initial enhanced image The sum of the correction coefficients of the initial gain parameters of the first blocks is added to the first The product of the initial gain parameters of the initial blocks is recorded as The final gain parameter of the initial blocks.
进一步的，基于每个初始分块的最终增益参数利用局部对比度增强算法对建筑工程图像进行增强，将增强后的图像记作最终增强图像；增强的具体过程为：在建筑工程图像中，计算每个像素点的局部对比度与像素点所在初始分块的最终增益参数的乘积，将乘积与像素点所在初始分块的局部均值的和值，记作每个像素点的最终增强灰度值；根据上述方法得到每个初始分块内每个像素点的最终增强灰度值；根据每个像素点的最终增强灰度值对建筑工程图像中每个像素点的灰度值进行替换，得到最终增强图像。Furthermore, the construction engineering image is enhanced by using a local contrast enhancement algorithm based on the final gain parameter of each initial block, and the enhanced image is recorded as a final enhanced image; the specific process of enhancement is: in the construction engineering image, the product of the local contrast of each pixel and the final gain parameter of the initial block where the pixel is located is calculated, and the sum of the product and the local mean of the initial block where the pixel is located is recorded as the final enhanced grayscale value of each pixel; the final enhanced grayscale value of each pixel in each initial block is obtained according to the above method; the grayscale value of each pixel in the construction engineering image is replaced according to the final enhanced grayscale value of each pixel to obtain the final enhanced image.
步骤S006：根据最终增强图像，得到建筑工程质量检测的结果。Step S006: Obtain the result of the construction project quality inspection based on the final enhanced image.
本实施例使用的深度神经网络为DeepLabV3神经网络；使用的数据集为最终增强图像。The deep neural network used in this embodiment is the DeepLabV3 neural network; the data set used is the final enhanced image.
需要分割的像素点，共分为2类，即训练集对应标签标注过程为：单通道的语义标签，对应位置像素点属于背景类的标注为0，属于隔离护栏区域的标注为1。The pixels that need to be segmented are divided into two categories, that is, the labeling process of the training set is: single-channel semantic label, the corresponding position pixel belongs to the background class is labeled as 0, and the corresponding position pixel belongs to the isolation guardrail area is labeled as 1.
网络的任务是分类，所以使用的loss函数为交叉熵损失函数。The task of the network is classification, so the loss function used is the cross entropy loss function.
通过深度神经网络得到最终增强图像中的隔离护栏区域。The isolation fence area in the final enhanced image is obtained through a deep neural network.
将当前天的最终增强图像中的隔离护栏区域像素点的个数与前一天的最终增强图像中的隔离护栏区域像素点的个数的比值，记作隔离护栏存在率；当隔离护栏存在率小于预设的存在阈值时，判断为隔离护栏损坏；当隔离护栏存在率大于或等于预设的存在阈值时，判断为隔离护栏正常；预设的存在阈值=0.99，本实施例以此为例进行叙述，其它实施方式中可设置为其它值。The ratio of the number of pixels in the isolation fence area in the final enhanced image of the current day to the number of pixels in the isolation fence area in the final enhanced image of the previous day is recorded as the isolation fence existence rate; when the isolation fence existence rate is less than the preset existence threshold When the isolation guardrail existence rate is greater than or equal to the preset existence threshold When the isolation fence is judged to be normal, the preset existence threshold =0.99, this embodiment is described by taking this as an example, and other implementation modes may be set to other values.
需要说明的是，本实施例采用模型来呈现反比例关系及归一化处理，为模型的输入，实施者可根据实际情况设置反比例函数及归一化函数。It should be noted that this embodiment adopts Model to present inverse proportional relationship and normalization, As the input of the model, the implementer can set the inverse proportional function and normalization function according to the actual situation.
至此，本实施例完成。At this point, this embodiment is completed.
以上所述仅为本发明的较佳实施例而已，并不用以限制本发明，凡在本发明的原则之内，所作的任何修改、等同替换、改进等，均应包含在本发明的保护范围之内。The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410635000.2ACN118212478B (en) | 2024-05-22 | 2024-05-22 | Construction engineering quality inspection method based on image processing |
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410635000.2ACN118212478B (en) | 2024-05-22 | 2024-05-22 | Construction engineering quality inspection method based on image processing |
| Publication Number | Publication Date |
|---|---|
| CN118212478A CN118212478A (en) | 2024-06-18 |
| CN118212478Btrue CN118212478B (en) | 2024-07-30 |
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410635000.2AActiveCN118212478B (en) | 2024-05-22 | 2024-05-22 | Construction engineering quality inspection method based on image processing |
| Country | Link |
|---|---|
| CN (1) | CN118212478B (en) |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN119784758A (en)* | 2025-03-11 | 2025-04-08 | 深圳市索威尔科技开发有限公司 | A dynamic defect monitoring and early warning method and system based on machine vision rapid image recognition |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114298944A (en)* | 2021-12-30 | 2022-04-08 | 上海闻泰信息技术有限公司 | Image enhancement method, apparatus, device and storage medium |
| CN115457035A (en)* | 2022-11-10 | 2022-12-09 | 山东鲁旺机械设备有限公司 | Machine vision-based construction hanging basket welding quality detection method |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115601367B (en)* | 2022-12-15 | 2023-04-07 | 苏州迈创信息技术有限公司 | LED lamp wick defect detection method |
| CN116229870B (en)* | 2023-05-10 | 2023-08-15 | 苏州华兴源创科技股份有限公司 | Compensation data compression and decompression method and display panel compensation method |
| CN117593295B (en)* | 2024-01-18 | 2024-05-28 | 东莞市立时电子有限公司 | Nondestructive testing method for production defects of mobile phone data line |
| CN117876971B (en)* | 2024-03-12 | 2024-05-28 | 武汉同创万智数字科技有限公司 | Building construction safety monitoring and early warning method based on machine vision |
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114298944A (en)* | 2021-12-30 | 2022-04-08 | 上海闻泰信息技术有限公司 | Image enhancement method, apparatus, device and storage medium |
| CN115457035A (en)* | 2022-11-10 | 2022-12-09 | 山东鲁旺机械设备有限公司 | Machine vision-based construction hanging basket welding quality detection method |
| Publication number | Publication date |
|---|---|
| CN118212478A (en) | 2024-06-18 |
| Publication | Publication Date | Title |
|---|---|---|
| CN113689428B (en) | Mechanical part stress corrosion detection method and system based on image processing | |
| CN106875381B (en) | Mobile phone shell defect detection method based on deep learning | |
| CN116630813B (en) | Highway road surface construction quality intelligent detection system | |
| CN114694144B (en) | Intelligent identification and rating method for non-metallic inclusions in steel based on deep learning | |
| CN110726725A (en) | Transmission line hardware corrosion detection method and device | |
| CN115249246A (en) | A kind of optical glass surface defect detection method | |
| CN116030060B (en) | Plastic particle quality detection method | |
| CN115290053B (en) | Method for rapidly detecting construction abnormity of high-rise building | |
| CN118212478B (en) | Construction engineering quality inspection method based on image processing | |
| CN113313107A (en) | Intelligent detection and identification method for multiple types of diseases on cable surface of cable-stayed bridge | |
| CN116188468B (en) | HDMI cable transmission letter sorting intelligent control system | |
| CN118241529A (en) | A method for testing the smoothness of an intelligent road | |
| CN119006445B (en) | An automatic detection method for cable surface defects based on image processing | |
| CN119338823B (en) | Face image quality assessment method, system and computer-readable storage medium | |
| CN117036346B (en) | Silica gel sewage treatment intelligent monitoring method based on computer vision | |
| CN116402822B (en) | Concrete structure image detection method, device, electronic equipment and storage medium | |
| CN118799314A (en) | A detection information processing method for wind power and photovoltaic power generation equipment | |
| CN118229629A (en) | Edge detection method, system, equipment and medium for large-size liquid crystal display | |
| CN117974657A (en) | Cable surface defect detection method based on computer vision | |
| CN119274001B (en) | Quantitative recognition method of test paper color change based on improved MSRCR and random forest | |
| CN112991425B (en) | Water level extraction method, system and storage medium | |
| CN114120061A (en) | A small target defect detection method and system for power inspection scenarios | |
| CN110956640B (en) | A Method of Edge Point Detection and Registration in Heterogeneous Images | |
| CN118587496A (en) | Automatic identification system and method of parts processing accuracy based on computer vision | |
| CN116452526B (en) | A rice seed recognition and counting method based on image detection |
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right | Effective date of registration:20250422 Address after:450000, 1st Floor, Building 8, Jiayang Science and Technology Plaza, No. 151 Yangjin Road, Jinshui District, Zhengzhou City, Henan Province Patentee after:Xiangsheng (Henan) Testing Technology Co.,Ltd. Country or region after:China Address before:116000 room 816, free trade building, Dalian Free Trade Zone, Liaoning Province Patentee before:Dalian Boxun Technology Co.,Ltd. Country or region before:China | |
| TR01 | Transfer of patent right |