Template-based Moving Object Tracking

In the contemporary era of information explosion, man receives all kinds of information which includes 70% vision information. So, digital image processing becomes more and more significant as the human being is cognizing and constructing the world. Tracking the moving object is an indispensable part of digital image processing, and it integrates digital image processing, automatic control and information science. In fact, it forms an technology which can recognize the object automatically, distill the object position information and track the moving object automatically from the digital image signal. In our daily life, tracking the moving object applies in the extensive field including military affairs, traffic and monitoring-controlling security and so on. We can forecast that the technologies of tacking become investigated deeply and apply more and more extensively while military affairs becomes prosperous, traffic becomes busy and security becomes important.On the all kinds of algorithms of tracking the moving object, the moving object tracking based on template applies extensively due to its simple algorithm and short processing time. However, the existing algorithms of the moving object tracking based on template are implemented in the image space, and that means using sub-image template matches the whole image. So, it leads to the blindness of searching scope and to the increment of the processing time.In this paper, firstly, we introduce the developmental status of tracking algorithms, the basic thought of tracking algorithms based on template and the existing tracking algorithms. After that, we propose the thought of how to translate the image to vector, and.in order to form the vector, we should distill the significant characteristics. So the matching will be implemented between the template vector and the image vector. Because of the changeability of the moving object, if the template vector is invariable, the departure of tracking course will appear as the moving object is changing and it probably leads to the failure of tracking finally. In order to increase the tracking veracity, we rectify the template to prevent the departure of tracking by implementing the algorithm of distilling the moving object's edge characteristics. At last, for the sake of inspecting the proposed algorithm in this paper, we take one video record as an experimental object. The result shows that we can track the moving object effectively and correctly through the algorithm and this algorithm increases the veracity of tracking and decreases the processing time. Of course, every algorithm has its advantages and disadvantages including the algorithm proposed in this paper, so we summarize the algorithm and propose what will be done in the future at the end of this paper.