A SIFT-point distribution-based method for head pose estimation

Estimating the head pose of a person in a video or image sequence is a challenging problem in computer vision. In this paper, we present a new technique on how to estimate the human face pose from a video sequence, by creating a probabilistic model based on the scale invariant features of the face. This method consists of four major steps: (1) the face is detected using the basic CAMSHIFT algorithm, (2) a training dataset is created for each face pose, (3) the distinctive invariant features of the training and test face image sets are extracted using the scale-invariant feature transform (SIFT) algorithm, (4) a kernel density estimate (KDE) of SIFT points on each image is generated. Pose classification is achieved by nearest-neighbor search using a KDE overlap measure. Results indicate that the proposed method is robust, accurate, not computationally expensive, and can successfully be used for pose estimation.

[1]  Gary Bradski,et al.  Computer Vision Face Tracking For Use in a Perceptual User Interface , 1998 .

[2]  M. Trivedi,et al.  A Two-level Pose Estimation Framework Using Majority Voting of Gabor Wavelets and Bunch Graph Analysis , 2004 .

[3]  David Beymer,et al.  Face recognition under varying pose , 1994, 1994 Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.

[4]  William T. Freeman,et al.  Example-based head tracking , 1996, Proceedings of the Second International Conference on Automatic Face and Gesture Recognition.

[5]  Roberto Cipolla,et al.  Determining the gaze of faces in images , 1994, Image Vis. Comput..

[6]  Mohan M. Trivedi,et al.  Head Pose Estimation in Computer Vision: A Survey , 2009, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[7]  Larry S. Davis,et al.  An anthropometric shape model for estimating head orientation , 1997 .

[8]  G LoweDavid,et al.  Distinctive Image Features from Scale-Invariant Keypoints , 2004 .

[9]  Thomas Vetter,et al.  Face Recognition Based on Fitting a 3D Morphable Model , 2003, IEEE Trans. Pattern Anal. Mach. Intell..

[10]  J. Crowley,et al.  Estimating Face orientation from Robust Detection of Salient Facial Structures , 2004 .

[11]  Sridha Sridharan,et al.  Efficient constrained local model fitting for non-rigid face alignment , 2009, Image Vis. Comput..

[12]  Roberto Cipolla,et al.  Fast visual tracking by temporal consensus , 1996, Image Vis. Comput..

[13]  James L. Crowley,et al.  Head Pose Estimation on Low Resolution Images , 2006, CLEAR.

[14]  R. Stiefelhagen.  Estimating Head Pose with Neural Networks-Results on the Pointing 04 ICPR Workshop Evaluation Data , 2004 .

[15]  P. J. Green,et al.  Density Estimation for Statistics and Data Analysis , 1987 .

[16]  Harry Wechsler,et al.  Face pose discrimination using support vector machines (SVM) , 1998, Proceedings. Fourteenth International Conference on Pattern Recognition (Cat. No.98EX170).

[17]  Yuxiao Hu,et al.  Evaluation of Head Pose Estimation for Studio Data , 2006, CLEAR.