Research And Application Of Binocular Visual Target Ranging In Cattle Manure Cleaning Robot System

This paper is the distance measuring system of binocular industrial camera,which is used in the project of "research and development of standardized cattle manure cleaning device".MV-EM120M industrial camera and AFT-1614MP industrial lens were mainly used to shoot the object,identify the object and measure the distance,and the robot arm equipped with the cow dung cleaning robot can accurately realize the snatch function.In order to meet the requirements of this design,this paper mainly establishes a complete binocular ranging system from three aspects:camera's two-target location,image stereo matching and binocular ranging.In order to better meet the needs of the project,this paper has made corresponding improvements to the dual-target and stereo matching parts of the ranging system.The paper mainly studies and experiments on the following tasks in the VS2010+OpenCV2.4.9 environment:(1)Camera Calibration:Due to the requirements of the experimental precision,the traditional checkerboard calibration plate was abandoned,and a higher precision AFT-CT430 calibration plate was selected.According to the calibration plate,an algorithm suitable for it was written to make the calibration result more accurate.(2)Stereo Matching:According to the actual application requirements,the detection feature points and construction descriptors studied are selected by SURF algorithm,and the violent matching method is used to perform preliminary matching of feature points.Because the number of mismatched points is too high,the next step is not efficient.Therefore,the limit constraint method is used to select the wrong matching points.Finally,the number of filter error point and the average time of filter error points are greatly improved.(3)Binocular ranging:According to the proposed matching method and the homography matrix model,the target object is determined,and the approximate center point of the target object is determined.Finally,the convergence binocular ranging method is used to obtain the information of the target object depth.Since the calibration and feature matching establish the foundation for speed and high precision,the final target of the project was determined to be 1.1 times higher than the overall processing speed of the ranging system,and the error rate was up to 1.25%.It can be confirmed from the experimental results that the proposed algorithm can be applied to the cattle manure cleaning device of the cattle farm.