Facilitating visual surveillance with motion detections

Visual surveillance is playing an ever increasing role in criminal detection because of a rapid deployment of surveillance cameras. Motion detection, which refers to the process of detecting a change in the position of an object in relation to the background or the change in the background in relation to the object, has become one of the enabling techniques to facilitate visual surveillance.

This paper parallelizes a motion detection algorithm using a cluster of inexpensive computing devices. Custom region of interest is implemented to enhance the performance and accuracy of the motion detection algorithm. The performance of the parallelized algorithm is evaluated from both the scalability in computation and the accuracy in motion detection. Performance evaluation results show that the enhanced algorithm achieves higher accuracy in motion detection with reduced execution times in computation.