Low complexity object detection with background subtraction for intelligent remote monitoring

Advancements in digital technologies have enabled cost-effective deployments of visual sensor nodes that can detect a motion event in the coverage area. The real world field remote monitoring image capture conditions, for example, in security and ecological studies are affected by wind, rain, snow, sunlight etc. and are seldom ideal. Motion detection is a precursor to subsequent intelligent processing on the image to extract information. Less complex object detection techniques often rely on maintaining a background image and subtracting foreground image, purporting to have an object in it, to create a difference image to determine the presence of a moving object. Correct object detection is critical as otherwise resulting false positive (without a moving object) images needlessly invoke further processing, storage and analysis. In this paper, we review background subtraction techniques and propose an image differencing technique that can significantly reduce the algorithm complexity along with other associated advantages. The results of proposed reduced image subset are provided to highlight the benefits.