Multi-pose Face Recognition Based on Block Adaptation

Considering that the SRC algorithm cannot solve the error offset problem between the testing and training samples, we propose one method as the affine transformation and partition by integrating into the linear reconstruction model, which were named as block adaptive multi-pose face recognition algorithm (BA-SRC). In this method, we first model the pose change using the affine transformation model for the face after the human face was blocked. Then we estimate the initial value of the affine transformation parameter by minimizing the image block reconstruction error, and then compensate the local area error caused by pose change, so as to improve the performance of face recognition. The experiments show that the algorithm proposed in this paper is very robust to pose change, and has a good recognition result.