Modeling of Pose Effects in Oriented Filter Responses for Head Pose Estimation

We propose an approach for view angle invariant recognition of 3D objects, based on modeling the variations of local feature values as function of view angle. In recognition stage we can compute the probabilities for any pixel that there is certain feature in a given pose angle. Any maximum likelihood or posterior based estimation methods can then be applied to infer the objects and their view parameters. We demonstrate the method with piecewise linear model for the pose effects, to recognize the location and pose of a head from the two eyes.

[1]  William T. Freeman,et al.  Presented at: 2nd Annual IEEE International Conference on Image , 1995 .

[2]  Alex Pentland,et al.  View-based and modular eigenspaces for face recognition , 1994, 1994 Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.

[3]  U. Halici,et al.  Intelligent biometric techniques in fingerprint and face recognition , 2000 .

[4]  Norbert Krüger,et al.  Face recognition by elastic bunch graph matching , 1997, Proceedings of International Conference on Image Processing.

[5]  John G. Daugman,et al.  Complete discrete 2-D Gabor transforms by neural networks for image analysis and compression , 1988, IEEE Trans. Acoust. Speech Signal Process..

[6]  Christoph von der Malsburg,et al.  Analysis and synthesis of human faces with pose variations by a parametric piecewise linear subspace method , 2001, Proceedings of the 2001 IEEE Computer Society Conference on Computer Vision and Pattern Recognition. CVPR 2001.