Research And Implementation Of Head Pose Estimation Algorithm Based On Monocular Vision

Head Pose Estimation is to estimate the three orientation parameters of the head relative to the camera from the images or video information,that is,to estimate 3 rotation angle pitch,yaw and roll.This technology has been widely used in the fields of Monitoring System,Vehicle Safety Driving Assistant System,Game entertainment and so on.Based on this,this paper has carried on the thorough research to the head posture estimation problem under monocular vision.The main work of this paper is as follows:1.the key algorithms involved in the whole Head Pose Estimation are discussed in detail,including Face Detection,Facial Feature Point Model and Head Pose Parameter Estimation.2.the Original POSIT Algorithm is derived from the Geometric Angle,this paper,the POSIT Algorithm is derived from the Algebraic Angle.3.This paper emphasizes on two general Head Pose Estimation Algorithms-POSIT Algorithm and EPNP Algorithm to estimate Head Pose Estimation,we study the influence of the noise,number of points and Parameter changes under different angles,the results show that POSIT Algorithm in accuracy is worse than that of EPNP Algorithm,but the former has significant advantages in speed,so we need to select the appropriate algorithm considering speed and accuracy in the actual problem.In addition,this paper sets up a complete Head Pose Estimation online demonstration system.