O-HAZE: A Dehazing Benchmark with Real Hazy and Haze-Free Outdoor Images

By Codruta Orniana Ancuti, Cosmin Ancuți, Radu Timofte, C. De Vleeschouwer. 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), 2018

704 citations recorded by the source corpus.

Haze removal or dehazing is a challenging ill-posed problem that has drawn a significant attention in the last few years. Despite this growing interest, the scientific community is still lacking a reference dataset to evaluate objectively and quantitatively the performance of proposed dehazing methods. The few datasets that are currently considered, both for assessment and training of learning-based dehazing techniques, exclusively rely on synthetic hazy images. To address this limitation, we introduce the first outdoor scenes database (named O-HAZE) composed of pairs of real hazy and corresponding haze-free images. In practice, hazy images have been captured in presence of real haze, generated by professional haze machines, and O-HAZE contains 45 different outdoor scenes depicting the same visual content recorded in haze-free and hazy conditions, under the same illumination parameters. To illustrate its usefulness, O-HAZE is used to compare a representative set of state-of-the-art dehazing techniques, using traditional image quality metrics such as PSNR, SSIM and CIEDE2000. This reveals the limitations of current techniques, and questions some of their underlying assumptions.

Subfields: Computer Vision and Pattern Recognition

View Paper: https://www.semanticscholar.org/paper/fb07f990b8247b46fec6dcd2a065a5f7e553db39

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

Citations in the corpus

Cycle-Dehaze: Enhanced CycleGAN for Single Image Dehazing (2018, methodology): “Therefore, the scaling process on images has a larger effect on I-HAZE [6] and O-HAZE [7] datasets according to PSNR and SSIM metrics.”
Seeing Through Fog Without Seeing Fog: Deep Multimodal Sensor Fusion in Unseen Adverse Weather (2019, background): “The folding of entropy and application of the sigmoid generates a multiplication matrix in the interval [0,1].”
Dense-Haze: A Benchmark for Image Dehazing with Dense-Haze and Haze-Free Images (2019, background): “Basically, Dense-Haze extends the O-HAZE [31] dataset that has been used recently for the ﬁrst single image dehazing challenge ever organized [32].”
NH-HAZE: An Image Dehazing Benchmark with Non-Homogeneous Hazy and Haze-Free Images (2020, background): “The NTIRE challenges employed new realistic image dehazing datasets (with haze and haze-free images): I-HAZE [5], O-HAZE [8] and DENSE-HAZE [4].”
Multi-Scale Boosted Dehazing Network With Dense Feature Fusion (2020, background): “O-HAZE [3] PSNR↑ 16.”
Advancing Image Understanding in Poor Visibility Environments: A Collective Benchmark Study (2019, methodology): “video restoration, such as DIV2K [108] and MANGA109 [29] for image super-resolution, PolyU [123] and Darmstadt [89] for denoising, RawInDark [13] and LOL dataset [120] for low light enhancement, HazeRD [141], OHAZE [3] and IHAZE [2] for dehazing, Rain100L/H [127] and Rain800 [134] for rain streak re”
“Double-DIP”: Unsupervised Image Decomposition via Coupled Deep-Image-Priors (2018, methodology): “We evaluated our framework on the O-HAZE dataset [4] and compared it to unsupervised and self-supervised dehazing algorithms.”
NAS-DIP: Learning Deep Image Prior with Neural Architecture Search (2020, methodology): “We present the visual comparisons with DoubleDIP [22] on the O-HAZE dataset [4].”
Vision Transformers for Single Image Dehazing (2022, neutral): “DehazeFormer-L [16, 16, 16, 12, 12] [48, 96, 192, 96, 48] [2, 4, 4, 2, 2] [1/4, 1/2, 3/4, 0, 0] [2, 4, 6, 1, 1] ConvBlock”
You Only Look Yourself: Unsupervised and Untrained Single Image Dehazing Neural Network (2020, methodology): “Besides, a variety of priors are employed based on different observations/assumptions, such as color attenuation prior (CAP) (Zhu et al., 2015), non-local color prior (NCP) (Berman et al., 2016) and haze-line prior (HLP) (Berman et al., 2017).”