The Research For Feature-based Algorithm Of Video Tracking

Object tracking is an important branch of computer vision, which combines advanced technologies and research achievements in image processing, pattern recognition, artificial intelligence, automatic control and other relative fields. The essence of object tracking is automatic recognizing moving target, deciding the state and the position of the target, and automatic tracking the target. Feature point extraction and tracking method is an important contents to study imaging-based target tracking,and avails to the research on target identification, detection and so on.In this paper, we introduce the match-based tracking methods, and discuss the correlation-based tracking methods and feature-based tracking methods. Then we discuss the theory of correlation-based tracking methods, matching criteria, the correlation-based tracking error analysis, searching region selection, as well as tracking methods exist. Next we discuss four feature-based tracking methods.The paper presents a tracking method based on Gabor wavelet feature and classification of target. This paper combines information of background and foreground (object), and this method extracts Gabor feature from background and foreground. Then we establish discriminant function, and use the function to classify background and foreground. The tracking becomes the foreground-background classification problem. The experimental results show that this approach is robust to the target with the zooming and illumination change.