An FPGA Based Efficient Surveillance System: A Split Processing Approach

Abstract

This research work proposes the design of a field programmable gate array (FPGA) based efficient surveillance robot. A bifurcated processing approach has been developed with an implementation of a part of video processing algorithms on the ZYNQ XC7Z010-1CLG400C FPGA (on the surveillance robot) and another part of the controller system. Video processing filters on the video signals to be transmitted are mapped onto the peripherals of the surveillance robot whereas detection and tracking algorithms have been applied to Control Unit system. Wi-Fi module on the surveillance robot transmits the Processed signals. The control unit system receives the transmitted video signals via IEEE 802.11 protocol. The control on robot has been achieved by the gaming console at control unit system with the manual feedback provided by a human. Wireless control with the portable controller has also been enabled with the help of an application developed on the Android device. Unlike conventional Surveillance systems where video processing algorithms are applied at a single stage, proposed research work implements video processing algorithms at two different stages one at the Programmable logic of node Machine and other at the control unit system. Experimental work has been carried out on MATLAB. It has been observed from the experiment that method shows optimum detection rate, accuracy, specificity and false positive rate to ensure accurate tracking for a single object and minimum latency due to the implementation of sophisticated algorithms on control unit doing the complete processing on video data. It can be used as an alternate to the fixed hardware surveillance systems in military applications with re-configurable hardware ensuring minimal hardware resources and more power efficiency.