Nonnegative Matrix Factorization Approach for Image Reconstruction

Abstract

Nonnegative Matrix Factorization (NMF) has been proved to be a powerful method in data processing and has also shown success in applications such as feature extraction and image representation. In this paper, we propose two symmetric matrix-based methods, Symcom and Symize, to achieve square strategy (SQR) in SQR-NMF. This integration process allows the matrix to preserve symmetry property associated with images to enhance image reconstruction. Simulation results show that Symcom performs better on super wide or super long data matrices and Symize achieves better results on symmetrical data matrices.