Noise level estimation using gradients of image blocks
Shuzhen Wang, Hong Liu, Kun Xie, Zimin Chen, Jingang Zhang · 2016
High-quality image denoising is required for many image processing applications. Our work concentrates on portraying a fast noise process method using PCA referring to weaked textured image patches, with patches generated from the noisy image corrupted by zero-mean Gaussian noise. The main contribution lies in the selection of specified blocks. We adopt a new measurement according to the gradients of the input patches ignoring the complex calculations of eigenvalues in previous articles. The proposed approach has high stability, accuracy and computational efficiency and even works perfectly in all kinds of scenes.