Discriminative Weighted Non-negative Sparse Low-rank Representation Classifier for Robust Face Recognition

Aiming at the traditional face recognition method, when the training samples and test samples are in complex application scenarios, the face recognition performance is degraded by the interference factors such as light changes, pollution and occlusion. This paper proposes a robust face recognition based on discriminative weighted non-negative sparse low-rank representation classification algorithm (WDNSLRRC). Based on the non-negative sparse low rank representation classification algorithm (NSLRRC), structural inconsistency constraints and singular values with different weights are assigned by adaptive weighted kernel norms. Different classes of samples may have the same features, and structural inconsistencies may inhibit these same features while retaining independent features. Weighted kernel norm (WNNM) is a low rank algorithm for constrained matrix singular value sparsity. The superiority of the algorithm in face recognition performance is proved in different face databases.