Research And Realization Of Double Camera Ranging System Based On Machine Vision

Machine vision,as a branch of the rapid development of artificial intelligence,mainly uses machines replace human eyes to measure and judge.Visual ranging is an important form of machine vision,which can effectively improve ranging speed and reduce labor costs.It is widely used in machine obstacle avoidance,industrial detection,robot positioning and navigation and other fields.This paper studies the dual-camera ranging system based on machine vision and performs real-time non-contact ranging on objects.A complete visual ranging system mainly includes: camera calibration,image processing,stereo matching and distance measurement.According to the principle of the ranging system,this paper uses two cameras to build an external image acquisition environment.Zhang Zhengyou calibration method is used to solve the internal and external parameters of two cameras.The distortion parameters and external parameters are combined with Bouguet distortion correction algorithm to correct the image.The image used for the actual measurement needs to be processed.Firstly,the image is grayed to get the gray image.Secondly,the image is smoothed and denoised to reduce the influence of image noise.Then,the edge segmentation method based on Canny operator is selected to segment the image,which can better distinguish the background from the image.Finally,using the Harris corner detection method with threshold and contrast enhancement,combined with the gradient operator for corner detection,which are ready for stereo matching.Stereo matching applies an improved Census transformation algorithm.The pixels gray value in the domain window are used to replace the median pixel gray value of the traditional algorithm,and the matching cost is calculated by the adaptive weight method.The resulting of disparity map is obtained by the left-right consistency check,the subpixel refinement method of quadratic polynomial difference and the median filtering.Finally,the distance measurement is realized by triangulation principle.According to the above work,a visual ranging system is built in the development environment of Microsoft Visual Studio and combined with computer vision library and Matlab calibration tools,and verified the system.Results show that the system has good accuracy and real-time performance,and has good application prospects.