A Solution for Improving the Effect of Tracking Moving-Object

摘要

Tracking moving-object is one of the key techniques in the intelligent surveillance system, which involves many subjects such as Computer Vision, Digital Image Processing and so on. In this paper, we firstly apply SVR model to invent the function of background confidence, which is used to estimate the reliability of pixel attribute. And then the color histogram and the mean-shift algorithm are applied to position the moving-object in video image. Pointing to the problems existed in the traditional tracking algorithm, we presents a solution to improve the tracking effect. The test proves that our solution achieves good actual results.