Gated Context Aggregation Network for Image Dehazing and Deraining

Chen, Dongdong; He, Mingming; Fan, Qingnan; Liao, Jing; Zhang, Liheng; Hou, Dongdong; Liu, Yuan; Hua, Gang

doi:10.1109/wacv.2019.00151

Cited by 494 publications

(267 citation statements)

References 49 publications

Supporting

Mentioning

265

Contrasting

Order By: Relevance

“…Qu et al [9] proposed an enhanced Pix2Pix dehazing network (EPDN) which used a phased dehazing module to enhance the dehazing effect. Chen et al [34]. proposed an end-to-end gated context aggregation network (GCANet), which used a smooth expansion technique to eliminate gridding artifacts caused by negligible parameters of the expanded convolution kernel, and leverage a gated sub-network to fuse the features from different levels.…”

Section: B Image Dehazing Methods Based On Learningmentioning

confidence: 99%

Iterative Residual Network for Image Dehazing

Hua

Fan

2020

IEEE Access

View full text Add to dashboard Cite

In this paper, we propose an Iterative Residual Network. By designing the calculation unit, we put the hazy image into the calculation unit to perform an iterative operation that can stitch the hazy image in stages with the unit output and substitute it into the calculation. After multiple iterations, a clean image can be generated. We introduce Long Short-Term Memory network and Residual ideas in the design process of the computing unit to further optimize the model. The Long Short-Term Memory network can be used to connect computing units at different stages. The use of residual block connection in the deep processing of the computing unit can preserve the original features of the image and prevent the model from overfitting. This model directly generates hazy-free images in an end-to-end manner and does not rely on the estimations of the transmission map and atmospheric light. Experiments show that Iterative Residual Network can effectively remove the haze in the image. In the test of the synthetic dataset and the real dataset, Iterative Residual Network is superior to the existing methods in terms of PSNR, SSIM, FADE and subjective visual effects.

show abstract

Section: B Image Dehazing Methods Based On Learningmentioning

confidence: 99%

Iterative Residual Network for Image Dehazing

Hua

Fan

2020

IEEE Access

View full text Add to dashboard Cite

show abstract

“…2. Furthermore, a state-of-the-art dehazing and deraining method [8] will be supplemented under our proposed method [63] for the advantage in the challenge of pristine scene restoration, especially for accumulated rain. Figure 2 shows an overview of the proposed algorithm.…”

Section: Proposed Methodsmentioning

confidence: 99%

Rain Streaks and Snowflakes Removal for Video Sequences via Motion Compensation and Matrix Completion

Zhou

Shimada

2020

SN COMPUT. SCI.

View full text Add to dashboard Cite

Image and video deraining tasks aim to reconstruct original scenes, from which human vision and computer vision systems can better identify objects and more details present in images and video sequences. This paper proposes a three-step method to detect and remove rain streaks, even snowflakes from great majority video sequences, using motion compensation and low-rank matrix completion method. Firstly, we adopt the optical flow estimation between consecutive frames to detect the motion of rain streaks. We then employ the online dictionary learning for sparse representation technique, and SVM classifier to eliminate parts that are not rain streaks. Finally, we reconstruct the video sequence by using low-rank matrix completion techniques. In particular, by introducing the image dehazing network(GCANet) to our proposed method, the heavy rain caused dense rain accumulation and blurry phenomenon can be worked out well. The experimental results demonstrate the proposed algorithm and perform qualitatively and quantitatively better in several image quality metrics, boosting the best published PSNR metric by 4.47%, 6.05% on two static video sequences and 12.13% on a more challenging dynamic video sequence. In addition, to demonstrate the generality of the proposed method, we further apply it to two challenge tasks, which also achieves state-of-the-art performance.

show abstract

“…[34] established a rain imaging model based on scene depth effect, and then constructed depth-guided attention rain removal network to guide the main network to output residual image, which is mainly suitable for RainCityscapes images with depth information. Chen et al [20] proposed the smoothed dilated convolution end-to-end gated network, which tackles the gridding artifacts by the smoothed dilated convolution and merges the features of different levels via the gated sub-network.…”

Section: Related Work a Rain Removal Researchmentioning

confidence: 99%

“…where  is trade-off parameter between ( ) and ( ). [18] and Gated Context Aggregation Network(GCANet) [20]. In the end, we make analysis in the ablation study.…”

Section: Loss Functionmentioning

confidence: 99%

“…Recently, the data-driven rain removal methods [5,[17][18][19][20] have attracted people's attention, which benefits from the powerful non-linear mapping function of deep CNN, without tedious handcrafted features or prior knowledge. Although those deep networks have achieved significant performance in single image de-raining, it is hard to strike the balance between low model parameters and advanced performance.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Wavelet Based Deep Recursive Pyramid Convolution Residual Network for Single Image Rain Removal

Gong

Wang

2020

IEEE Access

View full text Add to dashboard Cite

Image rain removal aims to separate the background image from the rainy image. During the past three years, the image rain removal with deep convolutional neural networks has achieved impressive performance. However, how to reach tradeoff between high de-raining performance and low model parameters is still a challenge. To address the issue, the paper is devoted to exploring a novel method based on wavelet deep recursive pyramid convolution residual network (WDRPRN), in which discrete wavelet transform is embedded to decompose the rainy image in different frequency domains, and the deep recursive pyramid convolution residual network (DRPRN) can well predict the residual coefficients between rainy image and clean image. In addition, compared with other neural networks, the DRPRN adopts recursive model that can cost fewer parameters. Plentiful of experiments on synthetic and real-world datasets show that the proposed method is significantly superior to the recent state-of-the-art algorithms. INDEX TERMS Image rain removal, wavelet transform, residual coefficients, low model parameters.

show abstract

Gated Context Aggregation Network for Image Dehazing and Deraining

Cited by 494 publications

References 49 publications

Iterative Residual Network for Image Dehazing

Iterative Residual Network for Image Dehazing

Rain Streaks and Snowflakes Removal for Video Sequences via Motion Compensation and Matrix Completion

Wavelet Based Deep Recursive Pyramid Convolution Residual Network for Single Image Rain Removal

Contact Info

Product

Resources

About