Kernel Collaborative Representation Classification Based on Adaptive Dictionary Learning

Abstract

In recent years, with the progress of technology, face recognition is used more and more widely in various fields. The classification algorithm based on sparse representation has made a great breakthrough in face recognition. However, face images are often affected by different poses, lighting, and expression changes, so test samples are often difficult to represent with limited original training samples. Due to the conventional dictionary learning methods lacking adaptability, we propose a kernel collaborative representation classification based on adaptive dictionary learning. In this paper, the coarse to fine sparse representation is related to the adaptive dictionary learning problem. First, the labeled atom dictionary is learned from each kind of training samples by sparse approximation. Based on this assumption, we use an efficient algorithm to generate an adaptive dictionary that is related with the test sample. Then, based on the adaptive class dictionary, the kernel collaborative representation is used to realize the inter class competition classification. The kernel function is combined with the coarse to fine sparse representation to extract the non-linear factors such as facial expression change, posture, illumination, occlusion and so on. The kernel collaborative representation is used to realize the inter class competition classification. The main advantage of this approach is to combine coarse to fine kernel collaborative representation with dictionary learning to generate adaptive dictionaries that approximate to the test image. Experimental results demonstrate that the proposed appraoch outperforms some previous state-of-the-art dictionary learning methods and sparse coding methods in face recognition.