Robust Head Pose Estimation Algorithm For Embedding Video System

recognition and computer vision research, has attracted more and more attention and achieved a rapid development in recent years. For its value in production and high accuracy, head pose estimation and concern behavior analysis has become the fastest growing research direction. Through the template matching, machine learning, morphological characteristics, track prediction methods, we can get the head pose information in 3D space. Combined with accurate face detection, it is widely used in security monitoring, national defense, information society, consumer electronics and other related areas of national economy.But in real applications, we often have only a normal camera for image and video capture, with the challenge like strong lighting environment change, the noise caused by camera equipment, complex and changing background environment and the shield of the object. The main purpose of this paper is to develop a robust head pose estimation algorithm.With the help of Histograms of oriented gradients(HOG) and piecewise processing, we propose an accurate piecewise HOG-SVR head pose estimation algorithm, which has an amazing estimation error, far better than the result of classical Adaboost and SVR method. Finally, we detect the occluded area of face according to the local response of the HOG block, and propose a blocking-Gaussian-HOG-SVR algorithm to handle occlusion issues.At last, we finish the simulation and implementation of the proposed algorithm in PC platform, and build an integrated verification system on TI OMAP3530 platform.