Pixelwise Estimation of Signal-Dependent Image Noise Using Deep Residual Learning

Figure 4

(a) Whole network architecture. The input is a noisy RGB image with three channels, which goes through a stack of residual blocks followed by a convolution and Relu [32] activation. The output is a noise-level channel. (b) Structure of the residual block. Firstly, the c-channel input tensor is convolved to a -channel tensor. Then, the -channel tensor goes through a residual structure with times of convolution and Relu and produces a w-channel output tensor.

(a)

(b)