Face Tracking Based on Particle Filter with Differential Evolution Algorithm

Background: Face tracking technology has been paid more and more attention with the development of intelligent techniques. Many methods have been used for face tracking, including the particle filter algorithm and Particle Swarm Optimization (PSO). However, these methods generally ignore the motive and the optimal solution of the particle. Methods: A particle filter based on the improved Differential Evolution Algorithm (DEF) is proposed, preventing the particles from falling into the local optimum and considering the mobility of the face tracking problem. The performance is better than the particle swarm optimization algorithm by replacing resampling in particle filtering, which is caused by the use of the niche technology to solve the problem of moving peak optimization. Results: The results show that the proposed algorithm is better than the multi-agent co-evolution of particle filter, adaptive PSO particle filter and other current methods. Although the proposed algorithm increases the tracking time, it can achieve about 13 frames per second of the tracking speed. Conclusion: A particle filter algorithm has been combined with the improved DEF and niche technology to avoid the local optimum and to solve the moving peak optimization problem. The proposed algorithm provides the best tracking accuracy with an acceptable tracking speed.