Research On Distance Measurement Based On Binocular Stereo Vision

One of the important ways for people to observe and understand the world is through vision.Most of the information is obtained through people's own vision system.Through simulating human vision,it has important theoretical significance and practical value in many fields,such as life,military and aerospace fields.In fact,computer vision technology realizes human visual function through computer,that is to realize the perception,recognition and positioning of the three-dimensional scene of the objective world.Computer and industrial camera are used to recognize,track and measure the target,and the image processing is further done.Computer software is used to process the image which is more suitable for human observation or transmission to various instruments for further detection.Binocular stereo vision is an important branch of computer vision.It collects images of different positions from the same scene by two identical cameras,and then makes stereo matching of the collected images to get the disparity map.Finally,it uses the disparity map to calculate the three-dimensional depth of the actual scene according to the ranging principle.The main research contents of this paper are divided into three aspects: firstly,analyze the basic principle of binocular stereo vision technology,deeply study the binocular stereo vision ranging system,build a software and hardware platform that can realize the ranging of close range target,and realize the ranging in the real time on the ranging interface;secondly,improve the corner detection in camera calibration algorithm by using gradient and The least square method can get more accurate sub-pixel corner position,and complete the sub-pixel corner precision processing;thirdly,in the stereo matching process,when calculating the matching cost,use 8 neighborhood to remove the maximum and minimum pixel values,the rest use the mean method to get the pixel values of the center point,and realize the cost through the dynamic planning of multiple paths in the semi global stereo matching Polymerization.In the parallax matching stage,the quality of the final parallax image is improved through the uniqueness detection,appropriate interpolation and left-right consistency detection of the rough parallax image.The hardware part of the platform mainly includes binocular camera and calibration board of high-precision float glass.The software part uses Visual Studio2015 and Matlab.Through the use of C + + and a large number of visual processingalgorithms in OpenCV library,the design of binocular stereo vision ranging is completed,and the ranging experiment is completed.Finally,the experimental results show that the binocular stereo vision realized in this paper can be used for accurate ranging in a certain range.