Research And Realization Of Target Tracking System Based On Active Vision

Target detection and tracking is the basement of visual analysis on moving target,and will become the focus of research with extensive application.After the study of object detection and tracking algorithm in common use,the paper put forward and design a kind of moving target detection and tracking algorithm based on the active vision,and complete the main function of the algorithm on embedded platform DM6437.On target detection,the paper introduces several detection algorithms.With the defect of incomplete detect the moving target using the frame difference method and the problem for the traditional static background difference method cannot updating background in real time,the paper puts forward to choose Adaboost algorithm or gaussian mixture background to detect target.Finaly,the effectiveness of the algorithm is tested by experiment.On target tracking,firstly,with the study of Kalman filter algorithm and Camshift algorithm,the fusion tracking algorithm on the Kalman filter and Camshift algorithm is designed.Secondly,according to the position and the area calculated by the target tracking algorithm,the paper propose and design the PTZ control algorithm,including the contral rules on pan,tilt and zoom to camera.Finally,combining with the moving target detection algorithm,it completes the automatic detection and tracking to moving targets,and the target is always in the center position of camera view at the right size.On the transplantation in embedded system,a moving target tracking system was designed and implemented on DM6437,including the construction of hardware platform,the acquisition and display for video image and the transplantation to the target tracking algorithm on active vision.Firstly,the embedded system realizes the video image acquisition and display,and stores the digital video images collected in SDRAM;The transplantation mainly completes the gaussian mixture background and the main function of fusion algorithm on Kalman filter and Camshift algorithm,and can calculate the position information of the target from the video images;Finaly,PTZ control algorithm was transplanted in the DM6437,and the control commands were sent through a serial port communication module.On the basis of the study on the active vision system and the test result for algorithm transplated on the DM6437,the follow conclusion can be attained: The tracking systemin based on active vision researched can detect and track target accurate and real time,and the moving target is always located in the center of the field,achieving the desired effect.