Efficient hardware implementation of PSO-based object tracking system

This paper proposes hardware design, modelling and implementation for object tracking system targeting video sequence based on Particle Swarm Optimization (PSO). The system uses Structural SIMilarity index (SSIM) as PSO´s fitness function. In this paper, we describe Particle Swarm Optimisation algorithm and Structural SIMilarity index, and then we show how it has been simplified for hardware implementation and achieved better results than its software implementation counterpart. Finally, we present the proposed algorithm´s performance that has been tested and evaluated over different video sequences, showing its adaption for tracking in real-time.