Research Of Image Tracking Algorithms Based On Omni-direction Vision

With the development of computer technology, the ability of computing has been improved greatly, and the realization of human visual function by using computer has become one of the most popular topics in the computer field. However, the traditional vision tracking technology is restricted within the lens view, and it has many defects, such as bad real-time and target losing and so on. Therefore, the observation system has the feature of large blind spot, and it is difficult to be used in the occasion required large-scale observation. So, the research of omnidirectional vision (Omni-vision) tracking technology is very necessary.The paper analyzes the Omni-vision system particularly from practical applications and theoretical study. In the aspect of practical applications, the paper puts forward a new type of Omni-vision system with fisheye lens combines the merit of fisheye lens and the shortage of the traditional Omni-vision system, and designs a new novel Beacon Model. The fisheye lens has the feature of compact structure, small and difficult to damage. The traditional system is characteristic of image distortion, image detail losing and uneven density. The term of fish eye lens was first introduced by Maxwell. The view recorded of the fisheye lens which produces an omnidirectional image (omni-image) is similar to that seen by a fish under water. The omni-image brings a highly unavoidable inherent distortion while it provides hemisphere. In the theoretical research level, the paper adopts a method called Spherical Perspective Projection which is used for correction of such distorted image. Also, the improved mean shift algorithm which is a powerful technique for tracking objects in image sequences with complex background has been proved to be successful for the fast computation and effective tracking problems. Therefore, the system not only has the large perspective, but can be solved effectively the problems, such as poor quality collected images, target veiling, target losing, very bad real-time.Finally, through the application in mobile robots, the Omni-vision system was perfectly proved in the recognition and tracking functions, and can meet the requirements.