R2RNet: Low-light image enhancement via Real-low to Real-normal Network

Hai, Jiang; Zhu, Xuan; Yang, Ren; Hao, Yutong; Zou, Fengzhu; Lin, Fang Cheng; Han, Songchen

doi:10.1016/j.jvcir.2022.103712

Cited by 89 publications

(42 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In order to increase the enhancement performance of the network, we have also made some improvements to the network. Lv et al (2018) and Hai et al (2023) have proposed a combination of multiple loss functions as the final loss function, but they have not considered the fact that medical image lesions are diverse in shape, size, and structure. Thus, we proposed a new loss function to enhance lesions and smooth other regions.…”

Section: Discussionmentioning

confidence: 99%

LRENet: a location-related enhancement network for liver lesions in CT images

Guo,

Wang,

Agaian

et al. 2024

Phys. Med. Biol.

View full text Add to dashboard Cite

Objective.Liver cancer is a major global health problem expected to increase by more than 55% by 2040. Accurate segmentation of liver tumors from Computed Tomography (CT) images is essential for diagnosis and treatment planning. However, this task is challenging due to the variations in liver size, the low contrast between tumor and normal tissue, and the noise in the images. Approach. In this study, we propose a novel method called Location-Related Enhancement Network (LRENet) that can enhance the contrast of liver lesions in CT images and facilitate their segmentation. LRENet consists of two steps: (1) locating the lesions and the surrounding tissues using a morphological approach and (2) enhancing the lesions and smoothing the other regions using a new loss function.Main results.We evaluated LRENet on two public datasets (LiTS and 3Dircadb01) and the dataset collected from a collaborative hospital (Liver cancer dateset), and compared it with state-of-the-art methods regarding several metrics. The results of the experiments showed that our proposed method outperformed the compared methods on three datasets in several metrics. We also trained the Swin-Transformer network on the enhanced datasets and showed that our method can improve the segmentation performance of both liver and lesions. Significance. Our method has potential applications in clinical diagnosis and treatment planning, as it can provide more reliable and informative CT images of liver tumors.

show abstract

Section: Discussionmentioning

confidence: 99%

LRENet: a location-related enhancement network for liver lesions in CT images

Guo,

Wang,

Agaian

et al. 2024

Phys. Med. Biol.

View full text Add to dashboard Cite

show abstract

“…This approach achieves promising results, and a subsequent improved version, KinD++, was also proposed. Jiang et al proposed the R2R network [18], which adds a denoising network to the original decomposition and relighting networks. This network aims to enhance brightness and contrast in the spatial domain while preserving the detail information of the original image in the frequency domain.…”

Section: Deep Learning Methodsmentioning

confidence: 99%

Forest nighttime image enhancement based on improved SCI light estimation and reflection optimization

Zhang,

Li,

Zhang

2023

IET Image Processing

View full text Add to dashboard Cite

The images captured by forest nighttime imaging devices are prone to degradation and noise issues due to environmental constraints. This paper proposes a novel framework based on the self‐calibrating illumination framework to improve image quality. The framework consists of four specific steps: (1) addressing non‐uniform degradation by estimating illuminance components through non‐uniform MaxRGB, (2) extracting rich information using multi‐scale inflationary convolution, (3) analyzing the relationship between different input levels and the initial level and progressively adjusting the luminance components using a channel self‐calibration module to achieve continuous improvement in luminance, and (4) designing a reflectance optimization module that enhances illuminance while suppressing noise through a regularization model. To evaluate the proposed method, a dataset comprising 1000 forest night images is constructed using an image simulation strategy. The effectiveness of this method was validated on both synthetic and real datasets, demonstrating its ability to significantly enhance nighttime forest images. When compared to six other algorithms, this method outperformed them in eight evaluation metrics, including a peak signal‐to‐noise ratio metric of 22.839 and a structural similarity metric of 0.909. The experimental results clearly indicate that this method excels in enhancing image quality and fidelity.

show abstract

“…The second convolutional layer uses dilation kernels with dilation rates of 3 to extract global contextual information by increasing the receptive field of the network. Inspired by the authors' findings of [21], we adopted the same denoising network architecture. The network has a modified U-Net architecture.…”

Section: B Low-light Enhancementmentioning

confidence: 99%

“…The max-pooling layers were replaced with convolutional layers to conserve feature information. Additionally, the network utilizes "deepnarrow" architecture [50], by replacing the sub-modules of U-Net with Residual Modules (RM) [21].…”

Section: B Low-light Enhancementmentioning

confidence: 99%

“…Another Retinexbased framework is KinD++ [20], proposed by Zhang et al, who refined their previous model by adding a multi-scale illumination attention module for reflectance restoration. R2R-Net [21] denoises the reflectance and then uses it to adjust the illumination. Despite their effectiveness, the first two Retinex models cause chroma noise when lighting dark areas, and R2R-Net produces outputs with diminished color contrast.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Iterative Retinex-Based Decomposition Framework for Low Light Visibility Restoration

et al. 2023

View full text Add to dashboard Cite

Images captured in low-light conditions often suffer from low brightness, low signal-to-noise ratio, low contrast, a narrow gray range, and color distortion, which can significantly impact human perception and limit the performance of various computer vision applications. Most existing low-light image restoration methods require assistance with a color cast, local over-exposure, glow, and artificial light sources. This paper proposes a new framework called RSD-Net, incorporating several innovative blocks, including a novel iterative Retinex network decomposition and enhancement algorithms, to improve the visibility and quality of images captured in low-light or nighttime conditions. We have extensively evaluated our proposed method on various benchmarking datasets and under different real-world scenarios, including challenging conditions such as glow, artificial light sources, low illumination, and noise. Moreover, we have evaluated our method on a face detection algorithm using extremely dark images and compared its performance with other state-of-the-art methods. The simulation results show that our proposed framework achieves a noticeable improvement compared to other low-quality image restoration techniques and enhances face detection accuracy in low-quality environments. The proposed framework has the potential to substantially impact human perception and enhance the performance of numerous computer vision applications.

show abstract

R2RNet: Low-light image enhancement via Real-low to Real-normal Network

Cited by 89 publications

References 40 publications

LRENet: a location-related enhancement network for liver lesions in CT images

LRENet: a location-related enhancement network for liver lesions in CT images

Forest nighttime image enhancement based on improved SCI light estimation and reflection optimization

Iterative Retinex-Based Decomposition Framework for Low Light Visibility Restoration

Contact Info

Product

Resources

About