LPC-based object tracking in image processing

The tracking method based on local position color information (LPC) is a subregion model matching approach. A new color representation method is proposed firstly to eliminate illumination efficiently. And then the target region is divided into similar color subregion, in each subregion the histogram is calculated in certain color range, furthermore, the target can be represented by the position-weighted histogram. At last, the minimum cross entropy method is used as the similarity function between the target model and the target candidates. The target tracking algorithm has better effect in the experimental result.