Research Article
Research of Image Dehazing Based on Image Gradient Distortion Prior
@INPROCEEDINGS{10.1007/978-3-030-72795-6_28, author={Chuangeng Tian and Jiamin Cheng and Lei Chen}, title={Research of Image Dehazing Based on Image Gradient Distortion Prior}, proceedings={Simulation Tools and Techniques. 12th EAI International Conference, SIMUtools 2020, Guiyang, China, August 28-29, 2020, Proceedings, Part II}, proceedings_a={SIMUTOOLS PART 2}, year={2021}, month={4}, keywords={Dehazing image Prior model Gradient distortion model Laplace distortion model Poisson equation}, doi={10.1007/978-3-030-72795-6_28} }- Chuangeng Tian
Jiamin Cheng
Lei Chen
Year: 2021
Research of Image Dehazing Based on Image Gradient Distortion Prior
SIMUTOOLS PART 2
Springer
DOI: 10.1007/978-3-030-72795-6_28
Abstract
In view of the problem of image quality degradation caused by the scattering of outdoor atmospheric particles in dusty weather, this paper proposed a research of dehazing image based on gradient distribution prior. Firstly, the method is to obtain a prior model of the gradient distribution of images from a large number of high quality natural image data sets. Secondly, the gradient distribution of hazy image is changed to make the infinite approximation the learned prior model. Finally, the Poisson equation is used to solve the reconstructed image to obtain dehazing image. According to a large number of experimental data, it shows that this research can remove the haze of image effectively. Compared with the most advanced methods, the image processed by this algorithm not only retains more details of the original image, but also improves the image definition largely.
