Moving target detection based on integration of saliency in time and frequency domains

Moving target detection in time domain by the traditional Gaussian Mixture Model algorithm contains problems such as, existing shadows and large calculation. To refine the issues mentioned above, a novel method is presented combining Frequency-Tuned and Gaussian Mixture Model. In this paper, the images were pre-processed. In addition, the dynamic interesting saliency map was detected by Frequency-Tunes algorithm. Then, the detected saliency map could be further detected by Gaussian Mixture Model to get the final conclusions. The results prove that the proposed method can advance velocity and eliminate the shadow.