Aiming at inherent problems of Illumination changes and occlusion in face recognition

a face recognition method combined with kernel function and blocking is proposed. Face images are divided into block images

each sub-block is mapped to a higher dimension space by Gaussian kernel. Each sub-block is combined with class specific dictionary learning to get the corresponding reconstruction error of each class based on local kernel collaborative representation. According to the reciprocal of the reconstruction error

the process from local discrimination to global recognition is completed by voting. Experimental results on Extend Yale B

AR and CMU PIE face database show that the proposed method not only has high recognition accuracy

but also has strong robustness to illumination changes and occluded face images.