Denoising of Facial Images Using Non-Negative Matrix Factorization with Sparseness Constraint

In this paper, we have studied denoising of facial images using non-negative matrix factorization with sparseness constraint. We have considered gaussian noise with zero mean and salt-pepper noise for our study. This type of noise partakes during image acquisition and are additive in nature. Here, we have also seen how this algorithm is able to compress the database and the possible areas to use this algorithm. We have validated this algorithm with standard error parameter of relative root mean square error. We also found out that the algorithm is not much affected with noise.