| In this paper, a method of recognition and tracking of objects via the features of their contours and inner color regions is presented. A stimulant platform of random movement of multiple objects is created so as to implement the simulation of recognition and tracking of moving objects.Firstly, a platform of image procession is created, in which some basic functions are provided, such as reading and writing image file, image format conversion, image data separation and combination. On the basis, some functions of digit image procession are implemented, such as two-valued procession, histogram statistics and median filter about gray image, color transform about color image, image geometric transform. In addition, some functions of image region partition and feature extraction are implemented.Feature information of each object, including its contour data and features, inner color region features, is extracted and integrated in the form of multiple level chains and then saved in a stock file. When an object needs to be recognized, its features are extracted and matched with features of all stock objects, then the object can be made certain what it is. The relative rotation angle between the current object and the stock one is obtained by rotary match between their contours. Thus the recognition and stance positioning of object is implemented.A simulant platform of random movement of multiple objects is constructed with multiple threads method by the system. Multiple motive objects are displayed harmoniously on the screen. They can randomly change their motive direction, velocity and attitude. On the basis, the experiment of dynamic identification and tracking of moving objects is done, and the system can recognize the moving objects which appear randomly and record its movement path.The programming methods on digit camera image capture, image transfer through internet, serial port communication, etc. are researched correspondingly so as to realize a system with more perfect functions. |
PDF Full Text Request