Research On Moving Workpieces Matching And Position Prediction Based On Images

With the continuous innovation of science and technology,the machine vision technology has been developed rapidly,and moving target predictive tracking technology has become an advanced subject.At present,the domestic machine vision technology has been widely used in food packaging,non-destructively testing and some high-risk production lines where are not suit for manual operation.But the moving target predictive tracking technology combined with machine vision is still in the research stage in some famous universities,and it is not well applied in practical engineering.What's more,the predictive tracking technology is the basis of video sequence processing and recognition,and it's also the guarantee of more advanced machine vision technology.Therefore,it is very important to research the moving target predictive tracking technology based on image sequence.In order to predict and track moving workpieces on the conveyor belt,the predictive tracking method combined with machine vision is used to deal with the problem.At first,according to the principle of camera imaging,the inside and outside parameters are calibrated by Zhang ZY checkerboard calibration method;Then,some processing are performed to ensure the matching and prediction proceeded smoothly,such as image preprocessing,edge detection,edge enhancement etc;According to the feature of workpieces,the Harris corner matching and Hough transform circle detection method are chose to complete all the matching work of workpieces,and add rectangular ROI to calculate their center moment so that the workpieces center could be obtained;For the predictive tracking problem of moving workpieces,Cam-shift algorithm is used to complete tracking task,which is based on the principle of Mean-shift algorithm.In order to ensure the requirements of accuracy and real-time,Kalman filter combined with Cam-shift algorithm is studied.Using Kalman filter could make the prediction of moving target more accurate and speed,when combined with the Cam-shift algorithm could complete the tracking successfully.On this basis,then proposed a new tracking algorithm based on linear prediction and Cam-shift algorithm,the linear prediction method is used as filtering estimation to instead of Kalman filter,and the prediction results would be substituted into Cam-shift estimation algorithm for tracking.The simulation results show that the matching method could match all the workpieces successfully and obtain the accurate center position,and the linear prediction combined with Cam-shift algorithm could not only ensure the workpieces predictive tracking accuratly,but also could deal with the predictive tracking problem which is in the complex background,and it could satisfy the real-time requirements better.