Real-Time Target Tracking Based on SOPC

Abstract

The Mean Shift is an algorithm which is a non-parametric probability density gradient estimation. It can be widely applied to the moving target tracking. This paper presents a target tracking method of mean shift based on SOPC. Through designing the parallel algorithm of the system, to take into account the cost of FPGA hardware and complexity of the model, we improve the real-time of the target tracking. Meanwhile, through modeling based on DSP Builder, instead of preparing FPGA hardware description language code, we reduce system designing complexity and improve the designing efficiency of the electronic system . The results show that the real-time of the target tracking is proportional to the Parallelism of design, and the system can meet with the real-time of high-speed visual tracking.