Investigating Haze-Relevant Features in a Learning Framework for Image Dehazing

By Ketan Tang, Jianchao Yang, Jue Wang. 2014 IEEE Conference on Computer Vision and Pattern Recognition, 2014

588 citations recorded by the source corpus.

Subfields: Computer Vision and Pattern Recognition, Media Technology

View Paper: https://www.semanticscholar.org/paper/c1a665a0fd386d5653c0d4f7c1fff686f8d87912

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Media Technology

Citations in the corpus

Gated Fusion Network for Single Image Dehazing (2018, background): “[34] combined four types of haze-relevant features with Random Forest to estimate the transmission.”
Model Adaptation with Synthetic and Real Data for Semantic Dense Foggy Scene Understanding (2018, background): “Recent approaches also rely on trainable architectures [53], which have evolved to end-to-end models [34,59].”
A Fast Single Image Haze Removal Algorithm Using Color Attenuation Prior (2015, methodology): “’s method for preparing the training data [51], we collect the haze-free images from Google Images and Flickr and use them to produce the synthetic depth maps and the corresponding hazy images for obtaining enough training samples.”
Single Image Dehazing via Conditional Generative Adversarial Network (2018, background): “estimate the the transmission map by learning multi-scale haze relevant features [35].”
Densely Connected Pyramid Dehazing Network (2018, methodology): “Inspired by the previous methods that use multi-level features for estimating the transmission map [33, 5, 42, 1, 24], we propose a densely connected encoderdecoder structure that makes use of the features from multiple layers of a CNN, where the dense block is used as the basic structure.”
DehazeNet: An End-to-End System for Single Image Haze Removal (2016, methodology): “In [17], Random Forests (RF) are used to investigate haze-relevant features for haze-free image restoration.”
GridDehazeNet: Attention-Based Multi-Scale Network for Image Dehazing (2019, background): “Representative works along this line of research include [36, 5, 9, 37, 42].”
I-HAZE: a dehazing benchmark with real hazy and haze-free indoor images (2018, background): “Fusion-based single image dehazing approaches [13], [23] have also achieved visually pleasant results without explicit transmission estimation and, more recently, several machine learning based methods have been introduced [17], [24], [25].”
Dense-Haze: A Benchmark for Image Dehazing with Dense-Haze and Haze-Free Images (2019, background): “More recently, several machine learning dehazing methods [19], [20], [21], [22], [23], [24], [25] have been introduced in the literature.”
O-HAZE: A Dehazing Benchmark with Real Hazy and Haze-Free Outdoor Images (2018, background): “Recently many strategies [15, 39, 20, 41, 25, 6, 5, 1, 16, 2, 14, 40] addressed this problem , by considering different kinds of priors to estimate the transmission.”
Single Image Dehazing via Multi-scale Convolutional Neural Networks with Holistic Edges (2019, methodology): “A variety of multi-scale haze-relevant features are analyzed by Tang et al. (2014) in a regression framework based on random forests (Breiman et al. 2001).”
Underwater Image Enhancement by Dehazing With Minimum Information Loss and Histogram Distribution Prior (2016, unknown): “become too dark or too bright after being restored by our underwater image dehazing algorithm, an adaptive exposure map [38] is employed to adjust our results for better visual quality.”
Semantic Foggy Scene Understanding with Synthetic Data (2017, unknown): “Step 2 is simple: we use the method proposed in [29] with the improvement of [63].”
D-HAZY: A dataset to evaluate quantitatively dehazing algorithms (2016, background): “While earlier dehazing approaches employ additional information such as multiple images [3] or a rough estimate of the depth [4], recent techniques have tackled this problem by using only the information of a single hazy input image [5], [6], [7], [8], [9], [10], [11], [12], [13], [14], [15], [16].”
Benchmarking Single-Image Dehazing and Beyond (2017, background): “[19], [20] discovered the dark channel prior (DCP) to more reliably calculate the transmission matrix, followed by many successors.”
Non-local Image Dehazing (2016, methodology): “Patch-based methods take great care to avoid artifacts by either using multiple patch sizes [19] or taking into consideration patch overlap and regularization using connections between distant pixels [3].”
Single Image Dehazing via Multi-scale Convolutional Neural Networks (2016, methodology): “[2] in a regression framework based on random forests.”
NH-HAZE: An Image Dehazing Benchmark with Non-Homogeneous Hazy and Haze-Free Images (2020, background): “Most of the early single image dehazing methods solve the optical model of Koschmieder [26] by searching for different priors that capture statistical properties of the hazy and non-hazy images [20, 38, 24, 40, 27, 10, 1, 21, 19, 39].”
Enhanced Pix2pix Dehazing Network (2019, background): “[18] systematically investigated a variety of haze-relevant priors in a regression framework to learn the best prior combination for image dehazing.”