TY - GEN
T1 - MBA-RAINGAN
T2 - 2022 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2022
AU - Shen, Yiyang
AU - Feng, Yidan
AU - Wang, Weiming
AU - Liang, Dong
AU - Qin, Jing
AU - Xie, Haoran
AU - Wei, Mingqiang
N1 - Publisher Copyright:
© 2022 IEEE
PY - 2022
Y1 - 2022
N2 - Rain severely degrades the visibility of scene objects, especially when images are captured through the glass under rainy weather. We observe three intriguing phenomena: 1) rain is a mixture of raindrops, rain streaks and rainy haze; 2) the depth from the camera determines the degree of object visibility, where objects nearby and far away are visually blocked by rain streaks and rainy haze, respectively; and 3) raindrops on the glass randomly affect the object visibility of the whole image space. However, existing solutions and benchmark datasets lack full consideration of the mixture of rain (MOR). In this paper, we originally consider that the overall object visibility is determined by MOR, and enrich the RainCityscapes by considering real-world raindrops to construct the MOR dataset, named RainCityscapes++. To solve the practical rain removal problem arisen from MOR, we formulate a new rain imaging model and propose a multi-branch attention generative adversarial network (MBA-RainGAN). Extensive experiments show clear improvements of our approach over SOTAs on RainCityscapes++.
AB - Rain severely degrades the visibility of scene objects, especially when images are captured through the glass under rainy weather. We observe three intriguing phenomena: 1) rain is a mixture of raindrops, rain streaks and rainy haze; 2) the depth from the camera determines the degree of object visibility, where objects nearby and far away are visually blocked by rain streaks and rainy haze, respectively; and 3) raindrops on the glass randomly affect the object visibility of the whole image space. However, existing solutions and benchmark datasets lack full consideration of the mixture of rain (MOR). In this paper, we originally consider that the overall object visibility is determined by MOR, and enrich the RainCityscapes by considering real-world raindrops to construct the MOR dataset, named RainCityscapes++. To solve the practical rain removal problem arisen from MOR, we formulate a new rain imaging model and propose a multi-branch attention generative adversarial network (MBA-RainGAN). Extensive experiments show clear improvements of our approach over SOTAs on RainCityscapes++.
KW - Bidirectional coordinate attention
KW - GAN
KW - Image deraining
KW - MBA-RainGAN
KW - Mixture of rain
UR - https://www.scopus.com/pages/publications/85131257187
U2 - 10.1109/ICASSP43922.2022.9746588
DO - 10.1109/ICASSP43922.2022.9746588
M3 - Conference contribution
AN - SCOPUS:85131257187
T3 - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
SP - 3418
EP - 3422
BT - 2022 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2022 - Proceedings
Y2 - 22 May 2022 through 27 May 2022
ER -