Premium
14.3: A real‐time video defogging algorithm
Author(s) -
Di Lu,
Limin Yan
Publication year - 2021
Publication title -
sid symposium digest of technical papers
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.351
H-Index - 44
eISSN - 2168-0159
pISSN - 0097-966X
DOI - 10.1002/sdtp.14387
Subject(s) - computer science , artificial intelligence , computer vision , pyramid (geometry) , distortion (music) , frame (networking) , sky , image (mathematics) , algorithm , mathematics , geography , telecommunications , amplifier , geometry , bandwidth (computing) , meteorology
This paper proposes a fast and efficient real‐time video defogging algorithm based on pyramid model, which solves the problems existing in the defogging algorithms, such as over‐estimation of transmittance, color distortion of sky area, and poor real‐time performance. Pyramid subsampling is used to obtain reduced images. The coarse transmittance of the reduced image is obtained by introducing pseudo‐defogging image and the confidence of dark primary color as correction factors. The image was returned to the original size, and the texture was refined with the combined bilateral filtering. Finally, the atmospheric light scattering model and the inter‐frame video defogging theory were combined to restore the degraded video. The experimental results show that this method has a complete defogging effect and can efficiently defog a variety of scenes. The improvement of the algorithm in PSNR and MSSIM is 15.7% and 11.6%, which effectively solves the problems of over‐estimation of transmittance, color distortion in the sky area, etc. Besides, it is fast and stable in operation, and suitable for real‐time video defogging.