EVBS-CAT: enhanced video background subtraction with a controlled adaptive threshold for constrained wireless video surveillance

Moving object detection (MOD) has gained significant attention for its application in advanced video surveillance tasks. Region-of-Interest (ROI) detection algorithms are essential prerequisites for various applications, ranging from video surveillance to adaptive video coding. The simplicity and efficiency of MOD methods are critical when targeting energy-constrained systems, such as Wireless Multimedia Sensor Networks (WMSN). The challenge is always to reduce computational costs while preserving high detection accuracy. In this article, we present EVBS-CAT, an Enhanced Video Background Subtraction with a Controlled Adaptive Threshold selection method for low-cost surveillance systems. The proposed moving object detection method utilizes background subtraction (BS) with morphological operations and adaptive thresholding. We evaluate the algorithm using the Change Detection 2012 dataset. Through a computational complexity analysis of each step, we demonstrate the efficiency of the proposed MOD technique for embedded WMSN. The algorithm yields promising results compared to state-of-the-art MOD techniques in the context of embedded wireless surveillance.