A Learning-to-Rank Approach for Image Color Enhancement

Yan, Jianzhou; Lin, Stephen; Kang, Sing Bing; Tang, Xiaoou

doi:10.1109/cvpr.2014.382

Cited by 85 publications

(93 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…An increasing amount of efforts focus on investigating learning-based enhancement methods since the pioneering work of Bychkovsky et al [15], which provides the first and largest MIT-Adobe FiveK dataset consisting of input/output image pairs for tone adjustment. Yan et al [17] achieved automatic color enhancement by tackling a learning-to-rank problem, while Yan et al [25] enabled semantic-aware image enhancement. Recently, Lore et al [26] presented a deep autoencoderbased approach for enhancing low-light images.…”

Section: Histogram-based Methodsmentioning

confidence: 99%

“…Early approaches work by performing histogram equalization [7], [8], [9], or by designing intensity mapping functions [10], [11], [12], [13], while many subsequent approaches [1], [2], [3], [4] rely on the Retinex model [14] to enhance photos. Others learn data-driven photo adjustment by utilizing either traditional machine arXiv:1907.10992v1 [cs.CV] 25 Jul 2019 (a) Input (b) NPE [1] (c) WVM [2] (d) JieP [3] (e) LIME [4] (f) HDRNet [5] (g) DPE [6] (h) Ours learning techniques [15], [16], [17], or the deep neural networks [5], [18], [6]. However, as demonstrated in Fig.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Enhancing Underexposed Photos Using Perceptually Bidirectional Similarity

Zhang

Nie

Zhu

et al. 2021

IEEE Trans. Multimedia

View full text Add to dashboard Cite

This paper addresses the problem of enhancing underexposed photos. Existing methods have tackled this problem from many different perspectives and achieved remarkable progress. However, they may fail to produce satisfactory results due to the presence of visual artifacts such as color distortion, loss of details and uneven exposure, etc. To obtain high-quality results free of these artifacts, we present a novel underexposed photo enhancement approach in this paper. Our main observation is that, the reason why existing methods induce the artifacts is because they break a perceptual consistency between the input and the enhanced output. Based on this observation, an effective criterion, called perceptually bidirectional similarity (PBS) is proposed for preserving the perceptual consistency during enhancement. Particularly, we cast the underexposed photo enhancement as PBS-constrained illumination estimation optimization, where the PBS is defined as three constraints for estimating the illumination that can recover the enhancement results with normal exposure, distinct contrast, clear details and vivid color. To make our method more efficient and scalable to high-resolution images, we introduce a sampling-based strategy for accelerating the illumination estimation. Moreover, we extend our method to handle underexposed videos. Qualitative and quantitative comparisons as well as the user study demonstrate the superiority of our method over the state-of-the-art methods.

show abstract

Section: Histogram-based Methodsmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Enhancing Underexposed Photos Using Perceptually Bidirectional Similarity

Zhang

Nie

Zhu

et al. 2021

IEEE Trans. Multimedia

View full text Add to dashboard Cite

show abstract

“…During the process, exemplar images are given or retrieved from a database and the target image's color distribution is translated into the exemplar images' color distribution. In the learning-based approaches [1,19,20], a mapping function from the source color distribution to the target color distribution is learned from training data. Recent work [20] handles the high non-linearity of this problem with a deep neural network.…”

Section: Introductionmentioning

confidence: 99%

Distort-and-Recover: Color Enhancement Using Deep Reinforcement Learning

Park

Lee

Yoo

et al. 2018

2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition

203

149

View full text Add to dashboard Cite

Learning-based color enhancement approaches typically learn to map from input images to retouched images. Most of existing methods require expensive pairs of input-retouched images or produce results in a noninterpretable way. In this paper, we present a deep reinforcement learning (DRL) based method for color enhancement to explicitly model the step-wise nature of human retouching process. We cast a color enhancement process as a Markov Decision Process where actions are defined as global color adjustment operations. Then we train our agent to learn the optimal global enhancement sequence of the actions. In addition, we present a 'distort-and-recover' training scheme which only requires high-quality reference images for training instead of input and retouched image pairs. Given high-quality reference images, we distort the images' color distribution and form distorted-reference image pairs for training. Through extensive experiments, we show that our method produces decent enhancement results and our DRL approach is more suitable for the 'distortand-recover' training scheme than previous supervised approaches. Supplementary material and code are available at https

show abstract

“…[1] proposed a local adjustment method that finding candidate images in dataset and searching for the best transformation of each pixel. [5] proposed a learning-to-rank method to enhance images step-by-step like human photographers.…”

Section: Introductionmentioning

confidence: 99%

A Coarse-to-Fine Framework for Learned Color Enhancement with Non-Local Attention

Shan

Zhang

Chen

2019

2019 IEEE International Conference on Image Processing (ICIP)

View full text Add to dashboard Cite

Automatic color enhancement is aimed to adaptively adjust photos to expected styles and tones. For current learned methods in this field, global harmonious perception and local details are hard to be well-considered in a single model simultaneously. To address this problem, we propose a coarse-tofine framework with non-local attention for color enhancement in this paper. Within our framework, we propose to divide enhancement process into channel-wise enhancement and pixel-wise refinement performed by two cascaded Convolutional Neural Networks (CNNs). In channel-wise enhancement, our model predicts a global linear mapping for RGB channels of input images to perform global style adjustment. In pixel-wise refinement, we learn a refining mapping using residual learning for local adjustment. Further, we adopt a non-local attention block to capture the long-range dependencies from global information for subsequent fine-grained local refinement. We evaluate our proposed framework on the commonly using benchmark and conduct sufficient experiments to demonstrate each technical component within it.

show abstract

A Learning-to-Rank Approach for Image Color Enhancement

Cited by 85 publications

References 18 publications

Enhancing Underexposed Photos Using Perceptually Bidirectional Similarity

Enhancing Underexposed Photos Using Perceptually Bidirectional Similarity

Distort-and-Recover: Color Enhancement Using Deep Reinforcement Learning

A Coarse-to-Fine Framework for Learned Color Enhancement with Non-Local Attention

Contact Info

Product

Resources

About