Optimized motion detection method based on modified three-frames and stationary wavelets

摘要

In visual surveillance, the object detection is a fundamental step for further processing such as segmentation, tracking, and extraction of a scene’s contextual information. A lot of approaches of motion detection have been proposed in the literature. In this paper, we have proposed a new method which is more efficient and computationally inexpensive compared to the others ones to detect moving objects in video sequence. The proposed algorithm begins with computation of difference temporal using difference between three frames successive. And then, we propose to apply to fixed number of alternate frames centralized around the actual frames instead of making difference images using the traditional approach. The new approach helps to reduce the computational complexity without reducing quality of the obtained images. After calculating the modified three-frame difference, the obtained results are decomposed using discrete stationary wavelet transform 2D and the coefficients are thresholded using Birge-Massart strategy in order to extract the foreground. The evaluation tests show that the proposed approach reaches the better performance of detection than the other approaches.