Research And Implementation Of Stereo Matching Algorithm Based On Binocular Vision

Computer vision is a science that uses computer to simulate human eyes,and binocular stereo vision has always been a hot spot of in computer vision research.Binocular stereo vision captures the same object from different angles through a binocular camera,and uses two-dimensional image and the principle of triangulation to restore the three-dimensional scene information of the object.Binocular stereo vision has been widely used in the fields of unmanned driving,aerospace monitoring,intelligent transportation,target recognition,etc.Stereo matching is the most critical step in the binocular stereo vision system.After extensive researched on various stereo matching algorithms at home and abroad,an improved gradient stereo matching algorithm is proposed,and three-dimensional reconstruction is completed by using the obtained disparity map.At the same time,the hardware structure of the improved gradient algorithm is studied,and the function modules of the algorithm are designed on FPGA.The main research content is:1.The basic theory of binocular stereo matching algorithm is researched,and common stereo matching algorithms are implemented,and then the advantages and disadvantages of various stereo matching algorithms are analyzed.2.The traditional gradient algorithm only includes the feature of the edge region of the object,so the false matching rate is high.In response to this problem,an improved gradient algorithm is proposed.Combining the intra-color correlation and adding the color change information of the pixel point on the basis of the traditional gradient measure,a more stable gradient cost is obtained for the image edge and the weak texture area.In the stage of parallax refinement,the mis-matching points are classified and treated,and the parallax correction based on the cross-support window method is used to further reduce the mis-match rate of the algorithm.3.Using the principle of triangulation,the three-dimensional reconstruction iscompleted used the disparity map obtained by stereo matching and is used to verify the practicability of the algorithm.4.Aiming at the characteristics of large amount of computation and high complexity in stereo matching algorithm,the hardware structure of improved gradient algorithm is studied.The hardware structure of stereo matching algorithm is based on field programmable gate array(FPGA),and image cache module,matching cost calculation,left and right parallax calculation,left and right consistency test,follow-up processing and other functional modules designed by Verilog HDL language.In this paper,Quartus II 11.0 is used as a development tool for stereo matching algorithm,and Modelsim 10.0c is used for functional simulation of each module of stereo matching algorithm.Simulation results are analyzed finally.