Research on Image Denoising Based on Gamma Transformation
Hanmin Ye, Wenjie Liu, Shiming Huang · 2019
In order to solve the problem that the neural network style transfer method produces a lot of noise in the process of image synthesis, a method of neural network image denoising based on gamma transform is presented in this paper. The VGG -19 network is used to obtain the composite images of each round, and the pixels of the images are modified by gamma transformation to achieve the purpose of image demolishing. Taking the gamma transformation as a part of the loss function of neural network, and combining the content loss function and style function, the weighted algebra of the three loss functions is taken as the total loss function of neural network. The experimental results show that the noise of the composite image is successfully suppressed and the number of interactions of image generation is reduced.