One of the conventional methods in the field of image tracking of non-rigid targets is to use a repetitive procedure called average transfer in determining the central mode position of the target. The display of the target in the average transfer tracker is based on the More
One of the conventional methods in the field of image tracking of non-rigid targets is to use a repetitive procedure called average transfer in determining the central mode position of the target. The display of the target in the average transfer tracker is based on the histogram of spatial interpolation feature with a direction-independent kernel. The most critical challenge in the medium transfer detector is the kernel scaling. So far, no efficient and perfect method to adjust or adapt the kernel dimensions when the target dimensions change has been presented. Another problem of the average transmission detector occurs when facing a target with a variable radiation pattern. In this article, with the approach of solving these problems, the average transmission tracking algorithm with strong adaptive scaling is presented, while it solves the problem of the average transmission algorithm in the face of changes in the radiation pattern of the target by adapting the target model in each frame. In the proposed method, the dimensions of the window in the next frame are set first by using the power calculation method resulting from the time-space derivatives of the intensity of the image pixels. Then, the results of the window scaling in the next frame are applied to the average transfer detector. The results show that the use of the proposed algorithm, while reducing the target positioning error in comparison with the standard average transfer algorithm, also shows a significant efficiency against the changes of contrast 2 and target radiation pattern.
Manuscript profile
