Face Orientation Estimation by Particle Swarm Optimization

Traditional methods for 2D-3D feature based pose estimation problem require two inputs, and they can not work well due to lack of correspondences of input images. In order to solve the pose estimation problem more effectively and rapidly, this paper proposed a 6-point template which overcomes the shortcomings of traditional approaches, and use particle swarm optimization to find the optimal solutions. Experiments demonstrate that the proposed method is effective, anti-noising, and rapid.