Application Of Robust Online Appearance Model In Tracking

Image tracking is an important branch of computer vision, which is significant both in theoretic area in reality. Especially in military, it has become one of the core techniques of Accuracy arm. Image tracking is difficult in such case as image signal is net definite or stable ,which is always influence by environmental problems such as complex background, air disturbance, all kinds noise, occlusion, illumination and so on. How to achieve reliable tracking performance from unreliable image is the main purpose of the tracking systems.Lack of suitable model is an important reason to influence the performance of tracking algorithms. Thomas F.El-Maraghi advised a robust and adaptive appearance model for motion-based tracking algorithms. The paper will discusses a set of problems on the model. Fist combines the model with correlative modal, utilizes ability of auto-exploring and identifying the stable signal of object of it to build an adaptive template. Then introduces phase-based WSL model, improve the phase features. And detrude more robust multi-scale feature based on phase. Latter build up a new WSL-like model. Next combine it with motion-based tracking algorithm to construct a new robust algorithm. Finally several variations on the model are discussed and three different resample strategies are compared.