CN201256433Y - Tridimensional mechanical vision device based on bionic design - Google Patents

Abstract

The utility model relates to a visible sensation apparatus of stereo machine which is designed based on bionics, and is characterized in that a video frequency pick-up head array comprises a first pick-up head, a second pick-up head, a third pick-up head and a fourth pick-up head, which form the video frequency pick-up head array according to square symmetrical array, the four pick-up heads are respectively installed on four corners of the A surface o f the pick-up head array, four connectors of a pick-up head circuit are respectively corresponding to the four video frequency pick-up heads on the B surface of the pick-up head array, and the four pick-up heads are respectively connected with a four-way video frequency encoder through video frequency connecting heads. The visible sensation apparatus has the advantages that a group of the four pick-up heads form a bionic compound eye imaging sensor system which can produce two groups of matching images, namely direct coupling group matching images and cross-coupling group matching images, aggregating six pairs of matching images, thereby effectively obtaining object three-dimensional structures in images, and realizing dynamic automatic correcting to obtain the function of three-dimensional stereo information by using matching image algorithm.

Description

Three-dimensional machine vision device based on the bionics design

Technical field

The utility model is a kind of three-dimensional machine vision device that designs based on insect compound eye mechanism in the biology, belongs to computer vision measurement and technical field of video monitoring.

Background technology

At present in industrial video surveillance and monitoring and computer vision measurement application, high-definition picture often can provide needed material particular information, in order to obtain the three-dimensional information of image, the Vision Builder for Automated Inspection of forming by dual camera (stereo vision), this method adopted 2 independently and through the camera of overcorrect to form, and is to carry out three-dimensional structure in the three-dimensional Vision Builder for Automated Inspection to rebuild generally employing of institute; In addition, the Vision Builder for Automated Inspection that multi-cam is formed, usually adopt the camera more than 3, it is right that but each camera and another one camera can not be formed the image of optical axis symmetry one to one, therefore, common multi-cam method is then reused the combination of single camera often simply, does not have to set up the quadrate array arrange type insect compound eye type structure based on bionical mechanism, shortage system symmetry does not possess the bionic compound eyes function.

Summary of the invention

The utility model lacks system's symmetry at present stereo visual system, a kind of three-dimensional machine vision device based on the bionics design of arranging that designs based on the insect biological compound eye mechanism in the biology by square symmetry, be intended to overcome the deficiency that Vision Builder for Automated Inspection that existing dual camera or the camera more than 3 form exists, extract the 3-D solid structure information of object in the image of optimizing.

Technical solution of the present utility model: it is characterized in that the video frequency pick-up head battle array is by first camera, second camera, four video frequency pick-up heads of the 3rd camera and the 4th camera are formed, and according to square symmetry formula arrangement formation video camera head array, four cameras are installed in the position at four angles of camera array A face respectively, at camera array B face, first connector of camera circuit, second connector, the 3rd connector and the 4th connector, respectively with first camera, second camera, four video frequency pick-up heads of the 3rd camera and the 4th camera are corresponding; Four cameras connect head through video respectively and connect 4 road video encoders.

Described 4 road video encoders, its each road video encoder all adopts the SAA7114 chip of Phlips company, they are installed in the embedded video server of rear end, mainly finish the analog video signal of camera collection is changed to digitlization and video formats, its outputting data signals directly inserts the video frame memory in the embedded video server.

Advantage of the present utility model: design according to bionics principle, form three-dimensional machine vision device with one group of 4 camera, it can produce two assembly to image: promptly, directly the coupling assembly to image and mutual coupling assembly to image, amount to six pairs of pairing images, thereby obtain the object dimensional structure in the image effectively, thereby utilization pairing image algorithm is realized the dynamic auto function of obtaining the 3 D stereo information of object in the image with proofreading and correct.

Description of drawings

Accompanyingdrawing 1 is a system construction schematic diagram of the present utility model;

Accompanying drawing 2 is the positive schematic diagrames of settling of camera array of the present utility model;

Accompanying drawing 3 is that circuit connector of the present utility model is arranged schematic diagram;

Accompanying drawing 4 is camera of the present utility model and video-frequency connector annexation schematic diagram;

Accompanying drawing 5 is that direct coupling of the present utility model and mutual coupling assembly are to image; Wherein accompanying drawing 5-1 directly is coupled assembly to image graph, and accompanying drawing 5-2 is coupled assembly alternately to image graph;

Accompanying drawing 6 is functional-block diagrams of embedded video server D of the present utility model;

Accompanying drawing 7 is workflow diagrams of the present utility model.

Embodiment

Contrast accompanying drawing 1, A looks just (preceding) face that face is an imageing sensor, is the A face, and B looks the back that face is an imageing sensor, is the B face, and the housing of imageing sensor is C, and D is an embedded video server.The first camera I1, the second camera I2, the 3rd camera I3 and four cameras of the 4th camera I4 are installed in the position at four angles of camera array A face respectively, at the B face, the circuit connector B1 of camera, B2, B3 and B4, correspond respectively to the first camera I1, the second camera I2, the 3rd camera I3 and the 4th camera I4.

Contrast Fig. 2, its structure is that camera adopts CCD (Charge Coupled Device, charge coupled device) camera, by facing face (A face) as seen, the first camera I1 to the, four camera I4 arrange by quadrate array, four cameras are formed the high-resolution map image-position sensor, the center line of each camera is adjustable to the center in the front of imageing sensor, the parameter designing that is provided with commonly used is 25mm, the position each other, center that also is each camera is adjustable, is designed and sized to 2 * 25=50mm between the place-centric of camera apart.

Contrast Fig. 3, the circuit connector that is based on the three-dimensional machine vision device of bionics design is arranged schematic diagram, i.e. backsight (B looks face) schematic diagram.The circuit connector of the first camera I1 is B1, and the circuit connector of the second camera I2 is B2, and the circuit connector of the 3rd camera I3 is B3, and the circuit connector of the 4th camera I4 is B4.

Contrast Fig. 4, its structure is that the arrangement of the first camera I1 to the, four camera I4 is provided by Fig. 2, and each camera is one to one with being connected of video-frequency connector, and AA is the video connecting line among the figure, and BB is an audio connecting cord among the figure.

Contrast Fig. 5, what Fig. 5 represented is that the image that camera obtains is formed pairing image schematic diagram.Three-dimensional machine vision device is made up of four cameras of the first camera I1 to the, four camera I4, and each camera can be independently as a HD image working sensor.The image that per two parallel or vertical cameras obtain is formed the pairing image.These pairing images are referred to as directly to be coupled assembly to image, perhaps are referred to as to be coupled assembly alternately to image.Directly coupling group image is the first direct coupling group image camera (I1, I2), the second direct coupling group image camera (13, I4), the 3rd direct coupling group image camera (I1, I3), the 4th direct coupling group image camera (I2, I4) image is right, and mutual coupling group image is the first mutual coupling group image camera (I1, I4) and the second mutual coupling group image camera (12, I3) image is right.

Contrast Fig. 6, its structure mainly comprises formations such as jtag test protocol interface, direct access storage, Flash memory, synchronous DRAM SDRAM, ARM embedded system, RS485 communication interface, RS232 communication interface, 802.11 protocol processors and communication interface, general-purpose serial bus USB interface, the power management module of video encoder SAA7114, frame memory, video digital signal processor DSP, programmable logic device FPGA, standard.Video server D based on embedded system is designed to the composite construction that four camera signals of integrated collection are formed, any two camera independent assortments, and self-adapting calibration is hinted obliquely at world coordinate system, finishes the real-time computing function of quick stereo vision.Professional image/video digital signal processor (DSP) DM643 of TI company production, the high-end ARM embedded system PXA27X of Intel Company's production and the design platform that programmable logic device FPGA (field programmable gate array) combines are adopted in embedded video server D design.Communication interface adopts IEEE802.p/n standard design, the adaptive broadband wireless communications interface of compatible 5.9GHz and 2.4G.

The first camera I1, the second camera I2, four cameras of the 3rd camera I3 and the 4th camera I4 connect VIDE01～VIDE04 port that head is connected 4 road video encoders (adopting the SAA7114 of Phlips company) through video respectively.FPGA is intended to set up the express passway of high-speed image sampling unit and central processing unit, the main initialization to video encoder, the required sequential of dsp system and ARM system and logical relation, basic I/O signal, the control R/W signal etc. of producing able to programme realized.Dsp system finishes mainly that complicated high-performance stereoscopic vision is calculated in real time, the tracking of target, recognizer realization etc.High-end ARM embedded system is the coordinated management person of whole video server D, mainly is responsible for the exchange of communication, protocol analysis and data etc.

Contrast Fig. 7, the camera by operating personnel open the three-dimensional machine vision device that designs based on bionics obtains 4 road real time video images; If there is video image output, open system again then.If vision signal output is arranged, promptly obtain the real-time photography head array first camera I1, the second camera I2, the 3rd camera I3 and the 4th camera I4 export successfully, then enter next step; On the basis that obtains the output of 4 road parallel video images, obtain the image composition pairing image that per two parallel or vertical high-resolution map image-position sensors obtain under the software assistance at the embedded video server that is equipped with.

These pairing images be referred to as respectively directly to be coupled assembly to image and mutual coupling assembly to image.Directly the coupling assembly provides the image source of optimizing to image for extracting the image object 3-D solid structure.The mutual coupling assembly of obtaining simultaneously provides independently proof diagram image source of another one to image for extracting the image object 3-D solid structure.

Each video frequency pick-up head can be independently as a high definition pick-up lens job, per two parallel or vertical video frequency pick-up heads are formed the pairing imageing sensor, the image that these pairing imageing sensors obtain is referred to as directly to be coupled assembly to image, these direct coupling group images are the first direct coupling group image (I1, I2), the second direct coupling group image (13, I4), the 3rd direct coupling group image (I1, I3), the 4th direct coupling group image (I2, I4) image is right, and these direct coupling group images provide the image source of optimizing for extracting the image object three-dimensional structure.The high-resolution map image-position sensor that parallel with vertical video frequency pick-up head is formed forms combined crosswise pairing image, the image that the imageing sensor of these combined crosswise pairings obtains is referred to as to be coupled assembly alternately to image, mutual coupling group image is the first mutual coupling group image (I1, I4), the second mutual coupling group image (12, I3) image is right, these mutual coupling group images provide another independently proof diagram image source for extracting the image object three-dimensional structure, thereby have improved the accuracy of the object dimensional structure of obtaining in the image.

4 road video output signals of the video camera head array of arranging according to square symmetry are with parallel mode output, for the realtime graphic process software in the embedded video server of rear end provides data source.On the basis that obtains the output of 4 road parallel video images, the embedded video server that is equipped with in the rear end obtains per two images parallel or that per two vertical high-resolution map image-position sensors obtain and is referred to as to match image under the software assistance, these pairing images are referred to as directly to be coupled assembly to image, these direct coupling group images are to providing the image source of optimizing for extracting the image object three-dimensional structure, imageing sensor obtains mutual coupling assembly to image simultaneously, and these mutual coupling group images provide another independently proof diagram image source for extracting the image object three-dimensional structure.

The described first camera I1, the second camera I2, the 3rd camera I3 is connected with the embedded video server of four ports with the 4th camera I4, and each camera can be independently as a high definition pick-up lens job.The image that per two parallel or vertical high-resolution map image-position sensors obtain is formed the pairing image, these pairing images are referred to as directly to be coupled assembly to image, these direct coupling group images are the first direct coupling group image (I1, I2), the second direct coupling group image (13, I4), the 3rd direct coupling group image (I1, I3), and the 4th direct coupling group image (I2, I4) image is right, mutual coupling group image is the first mutual coupling group image (I1, I4), (12, I3) image is right for the second mutual coupling group image.

The video source hardware platform of direct coupling and mutual coupling is provided, can extracts formation stereo visual system data by different level, progressively and adaptively, obtain 3D (three-dimensional) Video stream information more accurately by embedded software.

Claims (4)

1, three-dimensional machine vision device based on the bionics design, it is characterized in that the video frequency pick-up head battle array is by first camera (I1), second camera (I2), four video frequency pick-up heads of the 3rd camera (I3) and the 4th camera (I4) are formed, and according to square symmetry formula arrangement formation video camera head array, four cameras are installed in the position at four angles of camera array A face respectively, at camera array B face, first connector (B1) of camera circuit, second connector (B2), the 3rd connector (B3) and the 4th connector (B4) are respectively with first camera (I1), second camera (I2), the 3rd camera (I3) is corresponding with the 4th camera (I4); First camera (I1), second camera (I2), four cameras of the 3rd camera (I3) and the 4th camera (I4) connect head through video respectively and are connected 4 road video encoders.

2, the three-dimensional machine vision device based on the bionics design according to claim 1, the center line that it is characterized in that described each camera is adjustable to the center in the front of imageing sensor, the parameter designing that is provided with commonly used is 25mm, is designed and sized to 2 * 25=50mm between the place-centric of camera apart.

3, the three-dimensional machine vision device based on bionics design according to claim 1 is characterized in that being connected by video connecting line AA, audio connecting cord BB between described each camera and video-frequency connector.

4, three-dimensional machine vision device based on the bionics design according to claim 1, it is characterized in that four the cameras compositions of described first camera (I1) to the 4th camera (I4), the image that per two parallel or vertical cameras obtain is formed the pairing image, directly coupling group image is the first direct coupling group image camera (I1, I2), the second direct coupling group image camera (13, I4), the 3rd direct coupling group image camera (I1, I3), the 4th direct coupling group image camera (I2, I4) image is right, mutual coupling group image is that (I1, I4) (12, I3) image is right with the second mutual coupling group image camera for the first mutual coupling group image camera.

Images