Risk Minimization Approach for Image Restoration Using L2 Penalty in EM Framework
Ravi Pratap Singh, M. K. Singh · 2023
This paper presents a modified expectation maximization (EM) technique for the restoring the image from its degraded model. The proposed methodology comprises the concept of minimization of the risk associated with the L2penalty in the maximization step (M-step) of the algorithm. This method optimizes the worst case error, and at the same time also ensures the uniform shrinkage of the wavelet coefficients by using proper regularization parameter. The proposed methodology achieves 8.77 dB which is larger than the ISNR achieved by the state-of-the art methods.