CRNet: A Detail-Preserving Network for Unified Image Restoration and Enhancement Task

Kangzhen Yang,Tao Hu,Kexin Dai,Genggeng Chen,Yu Cao,Wei Dong,Peng Wu,Yanning Zhang,Qingsen Yan

2024-04-22

Abstract:In real-world scenarios, images captured often suffer from blurring, noise, and other forms of image degradation, and due to sensor limitations, people usually can only obtain low dynamic range images. To achieve high-quality images, researchers have attempted various image restoration and enhancement operations on photographs, including denoising, deblurring, and high dynamic range imaging. However, merely performing a single type of image enhancement still cannot yield satisfactory images. In this paper, to deal with the challenge above, we propose the Composite Refinement Network (CRNet) to address this issue using multiple exposure images. By fully integrating information-rich multiple exposure inputs, CRNet can perform unified image restoration and enhancement. To improve the quality of image details, CRNet explicitly separates and strengthens high and low-frequency information through pooling layers, using specially designed Multi-Branch Blocks for effective fusion of these frequencies. To increase the receptive field and fully integrate input features, CRNet employs the High-Frequency Enhancement Module, which includes large kernel convolutions and an inverted bottleneck ConvFFN. Our model secured third place in the first track of the Bracketing Image Restoration and Enhancement Challenge, surpassing previous SOTA models in both testing metrics and visual quality.

Computer Vision and Pattern Recognition,Image and Video Processing

What problem does this paper attempt to address?

The problem that this paper attempts to solve is that in real - world scenarios, due to lighting conditions, exposure time limitations and sensor limitations, the captured images are often affected by various forms of image degradation such as blurring and noise, and usually only low - dynamic - range images can be obtained. In order to obtain high - quality images, researchers have attempted various image restoration and enhancement operations, such as denoising, deblurring and high - dynamic - range imaging. However, performing only a single type of image enhancement still cannot produce satisfactory image results.

In response to this challenge, the paper proposes a method named Composite Refinement Network (CRNet), which utilizes multi - exposure images to achieve unified image restoration and enhancement tasks. By fully integrating information - rich multi - exposure inputs, CRNet can perform unified image restoration and enhancement. To improve the quality of image details, CRNet explicitly separates and strengthens high - and low - frequency information through pooling layers and effectively fuses the information of these frequencies using specially designed Multi - Branch Blocks. In addition, CRNet also adopts a High - Frequency Enhancement Module, including large - kernel convolution and ConvFFN with an inverted bottleneck structure, to increase the receptive field and fully integrate input features.

Through these techniques, CRNet achieved third place in the first track of the Bracket Image Restoration and Enhancement Challenge, surpassing the performance of previous state - of - the - art models in test metrics and visual quality.