FPGA Based High Performance Optical Flow Computation Using Parallel Architecture

The proposed work describes a highly parallel architecture for high performance optical flow computation. This system implements the efficient Lucas and Kanade algorithm with multi-scale extension for the computation of large motion estimations. This work deals with the architecture, evaluation of the accuracy and system performance. It also has extension to the original L&K algorithm. So the capable of working is larger than the standard mono scale approaches. In this proposed system, Matlab and Modelsim simulation are selected for local optical flow algorithms due to their potential for a high-performance massive parallelization. The results are obtained with a throughput of one pixel per clock cycle along the whole processing scheme by using the finepipeline based architecture.