Research On Work Piece Identification And Positioning Method Based On Machine Vision

With the proposal of "made in China 2025",the research and application of industrial robot and computer vision have developed rapidly.The traditional industrial robot system can’t deal with the problem about the change of target type and position due to the lack of perception and judgment ability.The industrial robot equipped with visual sensor can improve the perception ability of target recognition and position judgment,realize the recognition and positioning of target,and better meet the industrial needs.For the sake of solving the problem that the traditional robot is not able to recognize the type and position of the target casting,a casting recognition and positioning system on account of machine vision is advanced in this thesis.Taking the industrial robot as the carrier,this thesis studies and discusses the casting type recognition and location technology based on computer vision,so that the industrial robot can grasp and polish the target casting.In this thesis,the calibration methods of monocular camera and hand eye calibration are studied.Secondly,the image preprocessing method is optimized and improved,and then the traditional edge extraction algorithm is improved.Finally,the template matching technique based on casting shape is used to recognize and locate the casting.The main contents of this thesis are as follows:(1)In terms of camera calibration,this thesis studies the transformation relationship between the coordinate systems involved in camera imaging,analyzes the principles of Zhang’s calibration method and Nine-point hand-eye calibration,and calculates the internal and external parameter distortion coefficient of industrial camera and the transformation relationship matrix about the coordinate system of camera and robot through experiments.(2)In order to reduce the influence of external factors such as illumination and noise,this thesis proposes an improved image preprocessing method,optimizes and improves the Gaussian filtering algorithm,and designs a new hybrid filtering algorithm to remove the influence of illumination and noise on the image.Secondly,aiming at the shortcomings of traditional edge feature extraction methods,the Canny edge operator is optimized and improved.Combined with mathematical morphology and wavelet fusion algorithm,an improved edge extraction method is proposed to extract the edge contour of the target casting.(3)According to the characteristics of the target casting,this thesis designs a target contour template matching method.Firstly,collect the image containing the target casting,process it with the improved image preprocessing method to remove the influence of interference factors such as illumination and noise,then segment the region of interest containing the casting,extract the casting contour information in the region with the improved edge extraction algorithm,and make it as a template,The coordinates of the whole casting and the coordinates of the camera are determined by traversing the coordinates of the casting,and the coordinates of the whole object are identified by calculating the coordinates of the camera.In addition,the trajectory planning of the industrial robot is realized by off-line programming software to guide the robot to complete the grasping and grinding of the casting.Finally,the feasibility of the vision system designed in this thesis is verified and analyzed through simulation and physical experiments.The experimental results show that the vision system designed in this thesis can effectively identify and locate the types of target metal castings.