Density-Aware Haze Image Synthesis by Self-Supervised Content-Style Disentanglement

doi:10.1109/TCSVT.2021.3130158

CASIA OpenIR > 多模态人工智能系统全国重点实验室 > 先进时空数据分析与学习

	Density-Aware Haze Image Synthesis by Self-Supervised Content-Style Disentanglement
	Zhang, Chi 1; Lin, Zihang 1; Xu, Liheng1 ; Li, Zongliang 1; Tang, Wei 2; Liu, Yuehu 1; Meng, Gaofeng3,4,5 ; Wang, Le1 ; Li, Li 6
发表期刊	IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY
ISSN	1051-8215
	2022-07-01
卷号	32 期号:7 页码:4552-4572
通讯作者	Liu, Yuehu(liuyh@xjtu.edu.cn)
摘要	The key procedure of haze image synthesis with adversarial training lies in the disentanglement of the feature involved only in haze synthesis, i.e., the style feature, from the feature representing the invariant semantic content, i.e., the content feature. Previous methods introduced a binary classifier to constrain the domain membership from being distinguished through the learned content feature during the training stage, thereby the style information is separated from the content feature. However, we find that these methods cannot achieve complete content-style disentanglement. The entanglement of the flawed style feature with content information inevitably leads to the inferior rendering of haze images. To address this issue, we propose a self-supervised style regression model with stochastic linear interpolation that can suppress the content information in the style feature. Ablative experiments demonstrate the disentangling completeness and its superiority in density-aware haze image synthesis. Moreover, the synthesized haze data are applied to test the generalization ability of vehicle detectors. Further study on the relation between haze density and detection performance shows that haze has an obvious impact on the generalization ability of vehicle detectors and that the degree of performance degradation is linearly correlated to the haze density, which in turn validates the effectiveness of the proposed method.
关键词	Feature extraction Image synthesis Scattering Generative adversarial networks Atmospheric modeling Training Testing Haze synthesis unsupervised image-to-image translation self-supervised disentanglement
DOI	10.1109/TCSVT.2021.3130158
收录类别	SCI
语种	英语
资助项目	National Key Research and Development Project of New Generation Artificial Intelligence of China[2018AAA0102504] ; National Natural Science Foundation of China[61973245]
项目资助者	National Key Research and Development Project of New Generation Artificial Intelligence of China ; National Natural Science Foundation of China
WOS研究方向	Engineering
WOS类目	Engineering, Electrical & Electronic
WOS记录号	WOS:000819817700037
出版者	IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
引用统计	被引频次：8[WOS] [WOS记录] [WOS相关记录]
文献类型	期刊论文
条目标识符	http://ir.ia.ac.cn/handle/173211/49195
专题	多模态人工智能系统全国重点实验室_先进时空数据分析与学习
通讯作者	Liu, Yuehu
作者单位	1.Xi An Jiao Tong Univ, Inst Artificial Intelligence & Robot, Xian 710049, Shaanxi, Peoples R China 2.Univ Illinois, Dept Comp Sci, Chicago, IL 60607 USA 3.Chinese Acad Sci, Natl Lab Pattern Recognit, Inst Automat, Beijing 100190, Peoples R China 4.Chinese Acad Sci, Ctr Artificial Intelligence & Robot, HK Inst Sci & Innovat, Beijing 100190, Peoples R China 5.Univ Chinese Acad Sci, Sch Artificial Intelligence, Beijing 100049, Peoples R China 6.Tsinghua Univ, Dept Automat, BNRist, Beijing 100084, Peoples R China
推荐引用方式 GB/T 7714	Zhang, Chi,Lin, Zihang,Xu, Liheng,et al. Density-Aware Haze Image Synthesis by Self-Supervised Content-Style Disentanglement[J]. IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY,2022,32(7):4552-4572.
APA	Zhang, Chi.,Lin, Zihang.,Xu, Liheng.,Li, Zongliang.,Tang, Wei.,...&Li, Li.(2022).Density-Aware Haze Image Synthesis by Self-Supervised Content-Style Disentanglement.IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY,32(7),4552-4572.
MLA	Zhang, Chi,et al."Density-Aware Haze Image Synthesis by Self-Supervised Content-Style Disentanglement".IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS FOR VIDEO TECHNOLOGY 32.7(2022):4552-4572.