.png){kind=icon}
Return
Single image deraining using modified bilateral recurrent network (modified_BRN)
.png){kind=icon}
PRE
AI
DOI：10.1007/s11042-023-15276-2.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
The process of reinstating a clean background to an image that has been destroyed by multiple rain streaks and rain built up is called Image Deraining. We propose a single recurrent network first that begins by iteratively unfolding one shallow-residual network and then uses a recurrent layer to transfer the in-depth properties across stages. The traditional SRN (Single Recurrent Network) was used to learn both residual mapping and direct mapping for the removal of unwanted rain-streaks and anticipating a clean backdrop. With the combining of the SRNs into modified Bilateral Recurrent Network (BRN), the rain-streak layer and the backdrop can be exploited. Hence, we put forward a model using bilateral LSTMs (Long Short-Term Memory) that can transmit deep-features of rain-streak layer and backdrop layer between stages, as well as introduce the inter-play between SRNs, resulting in a BRN. The proposed modified_BRN performs better over the sophisticated methods on real-world and synthetic datasets, such as the popular datasets: Rain100H, Rain 100 L and Rain 12. The comparative analysis of the experimental results has been analysed on the two standard parameters: PSNR (Peak Signal-to-Noise Ratio) and SSIM (Structural Similarity Index Measure).
Keywords:
Image deraining
Recurrent network
Convolutional neural network
LSTM
Residual mapping
Direct mapping
Rain-streak layer
Backdrop
Journal
IF:
3
Papers:
2.0W
Citations:
3.2W
Organization
Cited Papers
ShareSaveElectronic structure and morphology of dark oxides on zinc generated by electrochemical treatment
OAAI
OAAIShareSave
ShareSaveShareSaveDiagnostic and prognostic significance of squamous cell carcinoma antigen in non-small cell lung cancer.png)
Lung Cancer
Lung CancerIF0
PREAI
ShareSave
View more