Image defogging method and device based on perception discrimination enhanced generative adversarial network

Abstract

The invention discloses an image defogging method and device based on a perception discrimination enhanced generative adversarial network, and the method comprises the steps: collecting a sample: obtaining a clear and fogless image through a high-definition camera, synthesizing a foggy image through an optical model and depth prior information, and obtaining a foggy image pair and a fogless imagepair; constructing a generation network, dividing the collected data samples into a training set and a test set, and inputting the foggy images in the training set into the generation network to generate defogged images; constructing a perception discrimination enhanced adversarial network, sending the generated defogged image and the original fog-free image to the adversarial network, and discriminating whether the generated defogged image is true or false; after a given secondary round is iterated, obtaining an optimal model; and inputting the test image into the optimal model, and carryingout defogging processing. The method provided by the invention is superior to other latest image defogging algorithms, and a fog-free image with higher quality can be generated.

Description

technical field [0001] The invention belongs to the technical field of image defogging, and more specifically, relates to an image defogging method and device based on a perceptually enhanced generative adversarial network. Background technique [0002] Fog is an atmospheric phenomenon caused by very small particles in the air that obscure the transparency of the atmosphere. In computer vision, fog can lead to severe degradation of image quality, which in turn affects the performance of subsequent image analysis algorithms. [0003] Existing image defogging methods can be roughly divided into two categories, the first is based on prior defogging methods. He et al. assumed that in a clear natural image, at least one of the RGB channels is close to zero, so they proposed an image dehazing method based on the dark channel prior (Single Image Haze Removal Using Dark Channel Prior), but the processing is similar to Such methods may fail when using atmospherically lit scene obje...

AI Technical Summary

Problems solved by technology

However, most of the existing image defogging methods blindly increase the network depth in order to improve network performance, ignoring the fact that the model is difficult to converge during the training process, resulting in gradient disappearance.

Such an image defogging algorithm is difficult to achieve good visual effects on the reconstructed image

Images