Abstract :
The future architecture being proposed in this paper is shown. It is one in which digitization, processing, compression, and networking have been integrated into the video sensor. The video sensor itself becomes an intelligent node on the network and the need for an intermediate computer has been eliminated. As shown, remote computers are still involved in hosting application software, providing some processing, and for long-term storage of acquired video. In this architecture, the image processing has been partitioned such that the processor in the video camera performs part of the processing (feature extraction) and a remote computer performs the other part (feature matching). In addition to distributing the processing, more effective use is made of the network bandwidth in that only the feature definitions are sent across the network instead of the entire image
