A new approach for training a physics-based dehazing network using synthetic images

被引:3
|
作者
Del Gallego, Neil Patrick [1 ,2 ]
Ilao, Joel [1 ,3 ]
Cordel, Macario, II [1 ,4 ]
Ruiz, Conrado, Jr. [1 ,5 ]
机构
[1] De La Salle Univ, 2401 Taft Ave, Manila 1004, Metro Manila, Philippines
[2] De La Salle Univ, Graph Animat Multimedia & Entertainment GAME Lab, Manila, Metro Manila, Philippines
[3] De La Salle Univ, Ctr Automat Res CAR, Manila, Metro Manila, Philippines
[4] De La Salle Univ, Dr Andrew L Tan Data Sci Inst, Manila, Metro Manila, Philippines
[5] Univ Ramon Llull, Grp Recerca Tecnol Media, La Salle, Barcelona, Catalonia, Spain
关键词
Image dehazing; Deep neural network; Physics-based dehazing; Unlit image priors;
D O I
10.1016/j.sigpro.2022.108631
中图分类号
TM [电工技术]; TN [电子技术、通信技术];
学科分类号
0808 ; 0809 ;
摘要
In this study, we propose a new approach for training a physics-based dehazing network, using RGB images and depth maps gathered from a 3D urban virtual environment, with simulated global illumination and physically-based shaded materials. Since 3D scenes are rendered with depth buffers, full image depth can be extracted based on this information, using a custom shader, unlike the extraction of real-world depth maps, which tend to be sparse. Our proposed physics-based dehazing network uses generated transmission and atmospheric maps from RGB images and depth maps from the virtual environment. To make our network compatible with real-world images, we incorporate a novel strategy of using unlit image priors during training, which can also be extracted from the virtual environment. We formulate the training as a supervised image-to-image translation task, using our own DLSU-SYNSIDE (SYNthetic Single Image Dehazing Dataset), which consists of clear images, unlit image priors, transmission, and atmospheric maps. Our approach makes training stable and easier as compared to unsupervised approaches. Experimental results demonstrate the competitiveness of our approach against state-of-theart dehazing works, using known benchmarking datasets such as I-Haze, O-Haze, and RESIDE, without our network seeing any real-world images during training. The DLSU-SYNSIDE dataset and source code can be accessed through this link: https://neildg.github.io/SynthDehazing/. (c) 2022 Elsevier B.V. All rights reserved.
引用
收藏
页数:16
相关论文
共 50 条
  • [21] Physics-based Preoperative approach planning using hybrid virtual bodies
    Nakao, M
    Kuroda, T
    Komori, M
    Oyama, H
    Komeda, M
    MEDICINE MEETS VIRTUAL REALITY 12: BUILDING A BETTER YOU: THE NEXT TOOLS FOR MEDICAL EDUCATION, DIAGNOSIS , AND CARE, 2004, 98 : 262 - 264
  • [22] A Ship Firefighting Training Simulator with Physics-Based Smoke
    Tao, Rui
    Ren, Hongxiang
    Zhou, Yi
    JOURNAL OF MARINE SCIENCE AND ENGINEERING, 2022, 10 (08)
  • [23] A Physics-based prognostics approach for Tidal Turbines
    Ewing, Fraser
    Thies, Philipp R.
    Shek, Jonathan K.
    Bittencourt, Claudio
    2019 IEEE INTERNATIONAL CONFERENCE ON PROGNOSTICS AND HEALTH MANAGEMENT (ICPHM), 2019,
  • [24] A physics-based approach to modelling grassland fires
    Mell, William
    Jenkins, Mary Ann
    Gould, Jim
    Cheney, Phil
    INTERNATIONAL JOURNAL OF WILDLAND FIRE, 2007, 16 (01) : 1 - 22
  • [25] PHYSICS-BASED APPROACH TO WAVE STATISTICS AND PROBABILITY
    Babanin, Alexander V.
    PROCEEDINGS OF THE ASME 32ND INTERNATIONAL CONFERENCE ON OCEAN, OFFSHORE AND ARCTIC ENGINEERING - 2013, VOL 2A, 2013,
  • [26] District Heating Network Demand Prediction Using a Physics-Based Energy Model with a Bayesian Approach for Parameter Calibration
    Chen, Si
    Friedrich, Daniel
    Yu, Zhibin
    Yu, James
    ENERGIES, 2019, 12 (18)
  • [27] Physics-Based Satellite-Derived Bathymetry (SDB) Using Landsat OLI Images
    Kim, Minsu
    Danielson, Jeff
    Storlazzi, Curt
    Park, Seonkyung
    REMOTE SENSING, 2024, 16 (05)
  • [28] Inversion using adaptive physics-based neural network: Application to magnetotelluric inversion
    Alyousuf, Taqi
    Li, Yaoguo
    GEOPHYSICAL PROSPECTING, 2022, 70 (07) : 1252 - 1272
  • [29] Simulation of via interconnects using physics-based models and microwave network parameters
    Rimolo-Donadio, Renato
    Stepan, Andrzej J.
    Bruens, Heinz-Dietrich
    Drewniak, James L.
    Schuster, Christian
    2008 IEEE WORKSHOP ON SIGNAL PROPAGATION ON INTERCONNECTS, 2008, : 68 - +
  • [30] PHYSICS-BASED SONIC IR CRACK LENGTH ESTIMATION USING THERMAL IMAGES ALONE
    Abu-Nabah, Bassam A.
    Al-Said, Samer M.
    PROCEEDINGS OF 2022 49TH ANNUAL REVIEW OF PROGRESS IN QUANTITATIVE NONDESTRUCTIVE EVALUATION, QNDE2022, 2022,