Wood Sorting And Defect Detection System Based On Visual Robot

Automated equipment can significantly improve the efficiency of woodworking,machine vision is the key to achieving automation technology.Nowadays machine vision localization algorithm is rich in function,however,for benefit,visual positioning method need to be more stable and reliable,efficient and fast.For the demand,we have developed a multi-target positioning algorithm.In order to realize the fully automated system,this paper also studies the hand-eye calibration of visual robot and the detection of wood surface defects.The system first calibrates the associated vision system and robot by hand-eye algorithm,then performs multi-target positioning and wood surface defect detection.Finally,the robot is sorted according to the target's positioning,rotation angle and defect characteristics.The main work of this paper is as follows:(1)Analysis the basic theory of camera imaging principle,including: with the distortion of the camera imaging model,the visual coordinate system between the conversion relationship and the camera calibration theory.Using these theories to des ign the software and build the experimental platform,completed the hand to eye calibration experiment.(2)A multi-target positioning and rotation angle recognition algorithm based on contour feature is proposed.The basic idea of this algorithm is derived from the matching of the smallest squares,which improves the matching rate of the algorithm and improves the matching rate by Fourier transform,Image pyramid and other strategies make the algorithm matching rate,accuracy,speed further improved.The algorithm has a good matching rate,and the pursuit of a certain degree of robustness,accuracy and speed,can efficiently identify wood parts,and other targets have good versatility.(3)The development of wood defect detection algorithm is studied.The surface defects of wood are extracted by Ostu threshold segmentation method,diffuse water method and morphological treatment,and the defects are positioned and polygonal approximation.The self-designed defect detection algorithm is compared with the related functions in the visual software Halcon.(4)Using the visual software developed by our research group,the experiment of sorting and sorting of wood based on visual robot was carried out,the success rate of capture was 88%.After conversthe visual coordinate to the robot coordinate,the system can automatically identifies the position and rotation angle,and guides the robot to classify the target.The experimental results have been applied in the cooperative enterprise.