Resign Of Objective Recognition Technology For JJR-1 Teach Robot

Intelligent robot is the main direction of technology development,it can obtain information from the outside,to respond the environment changings and complete the task which pro-program robot can not to be.Robot vision as the main means to get information is considered the most important sense of the robot ability like human.So the purpose of the subject is to make the robot in the 1980s with vision, which used for the identification of mechanical parts.Based on laboratory JJR-1-type teach robots,The main contents are as follows:1.The machine vision system was analyzed,the appropriate hardware was selected.Machine Vision System software architecture and workflow were given.2.The camera should be standardized with Zhengyou Zhang standard Method, because the obtained images from CCD have some distortion due to imaging systems limitations.JJR-1 robot hand-eye was standardized with vec and Tensor product (direct product of matrix) linearization solving non-linear equation CX=XD.The simulation results obtained high precision.3.Dicussed different denosising methods occording to the images with environment noise.The gray balance,median filter,Gaussian filter were used.4.A local automatic threshold value method is proposed to extract edge in this paper for the pretreated image.The method improves the precision and speed than the traditional operator.Hough transform was discussed and improved in this paper for the detection of linear and round mechanical parts.This method is faster than traditional Hough transform.Mechanical parts for the nearly straight-line contours of non-linear detection were picked up.New method was used to 20 the direction of 15x15 template operator radius of the circle 7 and find pixels in the scope of±2. Adopt two template matching methods to match images,which based on rough matching proceeds local matching precision.Experiments show that the speed significantly faster than the traditional template matching.