Improved moving object detection algorithm based on adaptive background subtraction

Moving object detection is the first step in video-surveillance that aims to detect the moving objects to be classified and tracked. There are many challenges in moving object detection such as lighting changes, dynamic backgrounds, occlusions, and shadows. Many complicated algorithms were proposed in the literature to face these challenges at the cost of increasing the processing time that may deteriorate the performance of the whole surveillance system. SOBS is an efficient algorithm using self-organization approach that was presented in [1] and improved in [2]. In this paper, we introduce a new algorithm based on adaptive background subtraction that has reduced computation complexity, hence, lower processing time while maintaining a competitive performance in terms of the recall and precision parameters. The proposed algorithm construct background model and compares its pixels with the current images to identify foreground/background pixels and minimizes the number of updated pixels in background model to reduce the processing time. The processing time is decreased by 74% compared to the SOBS algorithm. In addition, the proposed algorithm has competitive performance with respect to state-of-the-art algorithms in moving object detection.