Image Deblurring Analysis Based on Deep Learning Algorithm
Xiaotian Liu · 2022 IEEE International Conference on Electrical Engineering, Big Data and Algorithms (EEBDA) · 2022
Nowadays, people take pictures more and more in daily life. However, since the objects they photograph are not always static, they will produce blur and impair image quality. In most cases, people are unlikely to get another opportunity to take pictures, so in order to solve the problem of deblurring blurred images, this article uses an algorithm based on Generative Adversarial Networks (GAN) to achieve this goal, mainly using DeblurGANv2 Code. By comparing the effects of using the Resnet, FPN-Mobile-net and FRN-Inception structures in the Generator, this article shows a relatively effective method of defuzzification, that is, using FPN-Mobile-net in the Generator, and in the generator Use Double-GAN discriminator.