Efficient video analytics to overcome command latency for Mega-pixel video transmission over ethernet

The Mega-pixel digital cameras for the surveillance system has the advantage of high-resolution videos but at the same time has the disadvantage of large latency time to transmit over the network. Which eventually results in incapability of desired object tracking. This problem can be over come by intelligent Video Analytics at the video acquisition end. Real time efficient video analytics (VA) for surveillance system need to make decision with in camera processing unit or at camera edge processing unit. This paper presents a novel approach to generate a full view mosaic for a surveillance sequence to help detect moving objects and extract MPEG-7 motion descriptors. The algorithm has been evaluated with several test sequence simulating dynamic environment with camera and object motion.