Motion Image Capture Identification And Tracking Of Mobile Robot

Robot positioning and monitoring systems play an important role in the development of intelligence. It has a very wide range of applications whether in military or civilian aspects. This paper repositions the robot and tracks it through the visual monitoring system under the circumanstance when Zig Bee is failed to position the mobile robot. Multi object recognition and tracking for mobile robots in the static background as well as single target tracking of mobile robots in the dynamic background are focused.First, this paper researched a method of the mobile robot target detection in the static background. After respective advantages and disadvantages of inter-frame difference and background subtraction were compared, the background subtraction based on the Median modeling was selected. The way to build value directly for RGB of the color image in the background modeling was optimized and the accuracy of the background modeling was improved.Secondly, the dynamic background was researched on the basis of detection. In this dynamic background, the method of mobile robot target detection mainly uses global motion compensation. It can solve the problem that the background changes greatly due to camera motion. Through the global motion compensation algorithm, the current frame of background can be the same as the previous one. Therefore, the moving targets can be detected through inter-frame difference algorithm.Thirdly, the target robots can be identified in a plurality of moving objects in the scene. SIFT feature point matching is the main method to complete the target identification. SIFT has the scale invariance and rotation invariance, the detected moving object was carried on SIFT extremum points matching with target template respectively. The mass center of the target object is the one which has the best matching degree. After identification, track moving targets. This paper used the pre-estimation of Kalman filter to track the target and improved the tracking stability and accuracy.Finally, a PC video surveillance platform was built. MFC interface development tools were used to develope a monitoring interface and procedures which were simple and highly maneuverable. And the real-time performance and robustness of the system were verified through the experiment of the mobile robot of image capture, identification and tracking.