PSO-based multiple people tracking

In tracking applications, the task is a dynamic optimization problem which may be influenced by the object state and the time. In this paper, we present a robust human tracking by the particle swarm optimization (PSO) algorithm as a search strategy. We separate our system into two parts: human detection and human tracking. For human detection, considering the active camera, we do temporal differencing to detect the regions of interest. For human tracking, avoid losing tracking from unobvious movement of moving people, we implement the PSO algorithm. The particles fly around the search region to get an optimal match of the target. The appearance of the targets is modeled by feature vector and histogram. Experiments show the effectiveness of the proposed method.