Discriminant Collaborative Representation based on Classification
Ting-ting Shan, Mingyan Jiang · 2018 IEEE International Conference of Safety Produce Informatization (IICSPI) · 2018
Collaborative Representation Classification (CRC) is a kind of reconstruction representation algorithm, it is extensively studied by many researchers. CRC is improved by sparse representation, but it is more efficient than sparse representation, especially in pattern recognition. In this paper, we explored CRC and improved the algorithm. We applied nearest neighbor criteria to CRC, and calculated a regularized Tikhonov matrix, which constrains the coefficient vector to obtain more local sample information. And we added a l2-norm regularized term into the objective function, it reduces the correlation between the different categories, and results in a more discriminant sparse coefficients. We also provided the most efficient solution to the novel algorithm. The method increases the discriminant of the coefficients while preserving the sparsity of the coefficients and improves the accuracy of image classification. At the end of this article, the effectiveness of the algorithm is proved by some experimental comparisons.