Recognition of partially occluded face by error detection with logarithmic operator and KPCA
Xiaolin Chen, Shunfang Wang, Xiaoli Ruan · 2016
Occluded images often affected the recognition rates in face recognition, thus the occlusion should be checked out and given a little weighting coefficient so as to weaken its impact on the recognition rate as much as possible. The traditional algorithms often use the reconstruction error operator based on principal component analysis (PCA) to estimate the weight for occluded face, which often need iterative computation and consequently high time complexity. Moreover there are many difficulties for noise problems and threshold selection in iteration. In order to overcome the shortcoming, this paper suggested a new research method of face recognition on the basis of error detection and kernel principal component analysis (KPCA) algorithm under partial occlusion. First, face images were divided into several regions. Then, the weight of each area is calculated by a new logarithmic transformed Gaussian error operator instead of the traditional error operator. Finally, each region's features are extracted by KPCA, which is a nonlinear transformation and mapping method. All features are fused with their weights to realize the final classification and recognition. The experimental results with the AR facial database indicated that the new algorithm was of great robustness and validity.