Abstract: In order to solve the problem that the cumulative error caused by model updating on mean shift object tracking will become larger in the following tracking, a new object tracking method based on four channel non-separable wavelets is presented. After extracting the accurate edges of the target using the non-separable wavelets decomposition, mixed features of high frequency and low frequency sub-images on the accurate target are used to track object. During tracking, the target model and scale are updated based on the extracting edges information. And the fusion coefficients are updated considering the relativity of the sub-features. The experimental result shows that the proposed method has validity and accuracy when the scene and the shape of the target are changed. Compared with tracking methods without image segmentation, the proposed tracking method is more precise. When comparing with the tracking methods using CRF, it handles more quickly and has less error.

Key words: image segmentation