Palmprint Recognition Using SC-LNMF Model in Gabor Domain

A novel palmprint recognition method utilizing the local non-negative matrix factorization (LNMF) with sparse constraint (SC-LNMF) in the transform domain of 2D-Gabor wavelet is mainly discussed in this paper. And to extract more texture features of palmprint images, a modified 2D-Gabor kernel function is also used here. It is known that the common LNMF method can successfully extract an image’s local feature, but it does not consider the sparse control of feature vectors and feature coefficients. Therefore, to solve the problem and improve feature recognition accuracy, the SC-LNMF model is proposed and performed in the 2D-Gabor wavelet transform domain. Furthermore, for SC-LNMF features, the classification task can be implemented by selecting appropriate classifiers. Experimental results show that this palmprint recognition algorithm proposed here is efficiency in application.