PSD: Principled Synthetic-to-Real Dehazing Guided by Physical Priors

Chen, Zeyuan; Wang, Yang‐Chao; Yang, Yang; Liu, Dong

doi:10.1109/cvpr46437.2021.00710

Cited by 201 publications

(135 citation statements)

References 33 publications

Supporting

Mentioning

132

Contrasting

Unclassified

Order By: Relevance

“…In the section, we evaluate the proposed method by conducting experiments on both synthetic datasets and real world images. All the results are compared against nine state-of-the-art dehazing methods: DCP [9], AOD-NET [20], PSD [7], CAP [43], non-local [3], MGBL [41], DehazeNet [5], PFF [24], GCA [6]. In addition, we conduct ablation studies to demonstrate effectiveness of each module of our approach.…”

Section: Methodsmentioning

confidence: 99%

“…And they need lots of reasoning time. However, the recent deep learning-based dehazing model [7,20,41] have the problem that the dehazing images are fuzzy and texture details are not obvious. In Figure 7, the dehazing images generated by our algorithm not only have small color distortion, but also retain a large number of texture details.…”

Section: Visual Comparisonsmentioning

confidence: 99%

“…Results: Firstly, the network with Tucker reconstruction regularization term fits the mapping relationship better in the image dehazing (see Table 2). Additionally, after we re-Input CAP [43] non-local [3] DCP [9] PSD [7] DehazeNet [5] AOD [20] MGBL [41] PFF [24] GCA [6] 3, with the increase of Taylor term, the targeted value decreases because the resolution of the first term is low, and part of the effective information is lost, so the first term of Taylor's expansion cannot be fitted well.…”

Section: Ablation Studymentioning

confidence: 99%

“…3) The recent approaches [7,41] have made some effective attempts at the high-resolution image dehazing task, but they still lack sufficient interpretability and can only prove their validity by limited qualitative analysis of the modules.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Single UHD Image Dehazing via Interpretable Pyramid Network

Xiao¹,

Zheng²,

Chen³

et al. 2022

Preprint

View full text Add to dashboard Cite

Currently, most single image dehazing models cannot run an ultra-high-resolution (UHD) image with a single GPU shader in real-time. To address the problem, we introduce the principle of infinite approximation of Taylor's theorem with the Laplace pyramid pattern to build a model which is capable of handling 4K hazy images in real-time.The N branch networks of the pyramid network correspond to the N constraint terms in Taylor's theorem. Low-order polynomials reconstruct the low-frequency information of the image (e.g. color, illumination). High-order polynomials regress the high-frequency information of the image (e.g. texture). In addition, we propose a Tucker reconstructionbased regularization term that acts on each branch network of the pyramid model. It further constrains the generation of anomalous signals in the feature space. Extensive experimental results demonstrate that our approach can not only run 4K images with haze in real-time on a single GPU (80FPS) but also has unparalleled interpretability. The developed method achieves state-of-the-art (SOTA) performance on two benchmarks (O/I-HAZE) and our updated 4KID dataset while providing the reliable groundwork for subsequent optimization schemes.

show abstract

Section: Methodsmentioning

confidence: 99%

Section: Visual Comparisonsmentioning

confidence: 99%

Section: Ablation Studymentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Single UHD Image Dehazing via Interpretable Pyramid Network

Xiao¹,

Zheng²,

Chen³

et al. 2022

Preprint

View full text Add to dashboard Cite

show abstract

“…To our knowledge, the test time training (fine-tuning) can be dated back to (Y. Sun et al, 2020), and similar methodology has been used in image dehazing problem (Chen, Wang, Yang, & Liu, 2021). The core idea of our fine-tuning strategy is to first synthesize a pseudo normal-light image Îsyn from the low-light image itself, and then update the network by minimizing the following loss function:…”

Section: Self-supervised Fine-tuning At Test Timementioning

confidence: 99%

Low-light Image Enhancement by Retinex Based Algorithm Unrolling and Adjustment

Liu¹,

Xie²,

Zhao³

et al. 2022

Preprint

View full text Add to dashboard Cite

Motivated by their recent advances, deep learning techniques have been widely applied to low-light image enhancement (LIE) problem. Among which, Retinex theory based ones, mostly following a decomposition-adjustment pipeline, have taken an important place due to its physical interpretation and promising performance. However, current investigations on Retinex based deep learning are still not sufficient, ignoring many useful experiences from traditional methods. Besides, the adjustment step is either performed with simple image processing techniques, or by complicated networks, both of which are unsatisfactory in practice. To address these issues, we propose a new deep learning framework for the LIE problem. The proposed framework contains a decomposition network inspired by algorithm unrolling, and adjustment networks considering both global brightness and local brightness sensitivity. By virtue of algorithm unrolling, both implicit priors learned from data and explicit priors borrowed from traditional methods can be embedded in the network, facilitate to better decomposition. Meanwhile, the consideration of global and local brightness can guide designing simple yet effective network modules for adjustment. Besides, to avoid manually parameter tuning, we also propose a self-supervised fine-tuning strategy, which can always guarantee a promising performance. Experiments on a series of typical LIE datasets demonstrated the effectiveness of the proposed method, both quantitatively and visually, as compared with existing methods.

show abstract

SCPA‐Net: Self‐calibrated pyramid aggregation for image dehazing

Chen

et al. 2022

Computer Animation & Virtual

View full text Add to dashboard Cite

Dehazing as an important image processing field has developed for many years, there exist many excellent methods for exploring more complex networks to solve this problem. In this paper, instead of designing a complex network structure, we propose a novel dehazing network based on the consideration of enhancing feature aggregation and feature representation abilities of dehazing architecture. Specifically, we propose a self-calibrated pyramid aggregation network (SCPA-Net) for image dehazing, which is based on an encoder-decoder architecture. In the encoder, we build a self-attention block as a unit to aggregate information from a neighborhood to adapt to its content. In the decoder, we introduce a self-calibration block to capture long-range spatial and channel dependencies to produce more discriminative representations. Finally, to learn the scale information, the pyramid upsampling structure is applied to aggregate the multiscale self-calibrated attentive features. Experimental results show our SCPA-Net can achieve impressive dehazing performance.

show abstract

PSD: Principled Synthetic-to-Real Dehazing Guided by Physical Priors

Cited by 201 publications

References 33 publications

Single UHD Image Dehazing via Interpretable Pyramid Network

Single UHD Image Dehazing via Interpretable Pyramid Network

Low-light Image Enhancement by Retinex Based Algorithm Unrolling and Adjustment

SCPA‐Net: Self‐calibrated pyramid aggregation for image dehazing

Contact Info

Product

Resources

About