ODD-Net: a hybrid deep learning architecture for image dehazing

Abstract Haze can significantly reduce visibility and contrast of images captured outdoors, necessitating the enhancement of images. This degradation in image quality can adversely affect various applications, including autonomous driving, object detection, and surveillance, where poor visibility ma...

Full description

Saved in:
Bibliographic Details
Main Authors: C. S. Asha, Abu Bakr Siddiq, Razeem Akthar, M. Ragesh Rajan, Shilpa Suresh
Format: Article
Language:English
Published: Nature Portfolio 2024-12-01
Series:Scientific Reports
Subjects:
Online Access:https://doi.org/10.1038/s41598-024-82558-6
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850145536125960192
author C. S. Asha
Abu Bakr Siddiq
Razeem Akthar
M. Ragesh Rajan
Shilpa Suresh
author_facet C. S. Asha
Abu Bakr Siddiq
Razeem Akthar
M. Ragesh Rajan
Shilpa Suresh
author_sort C. S. Asha
collection DOAJ
description Abstract Haze can significantly reduce visibility and contrast of images captured outdoors, necessitating the enhancement of images. This degradation in image quality can adversely affect various applications, including autonomous driving, object detection, and surveillance, where poor visibility may lead to navigation errors and obscure crucial details. Existing dehazing techniques face several challenges: spatial methods tend to be computationally heavy, transform methods often fall short in quality, hybrid methods can be intricate and demanding, and deep learning methods require extensive datasets and computational power. To overcome these challenges, we present ODD-Net, a hybrid deep learning architecture. Our research introduces a comprehensive data set and an innovative architecture called Atmospheric Light Net (A-Net) to estimate atmospheric light, using dilated convolution, batch normalisation, and ReLU activation functions. Furthermore, we develop T-Net to measure information transmission from objects to the camera, using multiscale convolutions and nonlinear regression to create a transmission map. The integrated architecture combines pre-trained A-Net and T-Net models within the atmospheric scattering model. Comparative analysis shows that ODD-Net provides superior dehazing quality, especially in transmission map estimation and depth measurement, surpassing state-of-the-art methods such as DCP, GMAN, DehazeNet, and LCA. Our quantitative analysis reveals that ODD-Net achieves the highest performance in terms the quality metrics compared. The proposed method demonstrates notable quantitative and qualitative improvements over existing techniques, setting a new standard in image dehazing.
format Article
id doaj-art-f035fbac49d042e4a87f0e93990093ff
institution OA Journals
issn 2045-2322
language English
publishDate 2024-12-01
publisher Nature Portfolio
record_format Article
series Scientific Reports
spelling doaj-art-f035fbac49d042e4a87f0e93990093ff2025-08-20T02:28:04ZengNature PortfolioScientific Reports2045-23222024-12-0114111710.1038/s41598-024-82558-6ODD-Net: a hybrid deep learning architecture for image dehazingC. S. Asha0Abu Bakr Siddiq1Razeem Akthar2M. Ragesh Rajan3Shilpa Suresh4Department of Mechatronics, Manipal Institute of Technology, Manipal Academy of Higher EducationDepartment of Mechatronics, Manipal Institute of Technology, Manipal Academy of Higher EducationDepartment of Mechatronics, Manipal Institute of Technology, Manipal Academy of Higher EducationDepartment of Electronics and Communication Engineering, Amrita Vishwa VidyapeethamDepartment of Mechatronics, Manipal Institute of Technology, Manipal Academy of Higher EducationAbstract Haze can significantly reduce visibility and contrast of images captured outdoors, necessitating the enhancement of images. This degradation in image quality can adversely affect various applications, including autonomous driving, object detection, and surveillance, where poor visibility may lead to navigation errors and obscure crucial details. Existing dehazing techniques face several challenges: spatial methods tend to be computationally heavy, transform methods often fall short in quality, hybrid methods can be intricate and demanding, and deep learning methods require extensive datasets and computational power. To overcome these challenges, we present ODD-Net, a hybrid deep learning architecture. Our research introduces a comprehensive data set and an innovative architecture called Atmospheric Light Net (A-Net) to estimate atmospheric light, using dilated convolution, batch normalisation, and ReLU activation functions. Furthermore, we develop T-Net to measure information transmission from objects to the camera, using multiscale convolutions and nonlinear regression to create a transmission map. The integrated architecture combines pre-trained A-Net and T-Net models within the atmospheric scattering model. Comparative analysis shows that ODD-Net provides superior dehazing quality, especially in transmission map estimation and depth measurement, surpassing state-of-the-art methods such as DCP, GMAN, DehazeNet, and LCA. Our quantitative analysis reveals that ODD-Net achieves the highest performance in terms the quality metrics compared. The proposed method demonstrates notable quantitative and qualitative improvements over existing techniques, setting a new standard in image dehazing.https://doi.org/10.1038/s41598-024-82558-6Image dehazingDeep retinexLCADense depthA NetT Net
spellingShingle C. S. Asha
Abu Bakr Siddiq
Razeem Akthar
M. Ragesh Rajan
Shilpa Suresh
ODD-Net: a hybrid deep learning architecture for image dehazing
Scientific Reports
Image dehazing
Deep retinex
LCA
Dense depth
A Net
T Net
title ODD-Net: a hybrid deep learning architecture for image dehazing
title_full ODD-Net: a hybrid deep learning architecture for image dehazing
title_fullStr ODD-Net: a hybrid deep learning architecture for image dehazing
title_full_unstemmed ODD-Net: a hybrid deep learning architecture for image dehazing
title_short ODD-Net: a hybrid deep learning architecture for image dehazing
title_sort odd net a hybrid deep learning architecture for image dehazing
topic Image dehazing
Deep retinex
LCA
Dense depth
A Net
T Net
url https://doi.org/10.1038/s41598-024-82558-6
work_keys_str_mv AT csasha oddnetahybriddeeplearningarchitectureforimagedehazing
AT abubakrsiddiq oddnetahybriddeeplearningarchitectureforimagedehazing
AT razeemakthar oddnetahybriddeeplearningarchitectureforimagedehazing
AT mrageshrajan oddnetahybriddeeplearningarchitectureforimagedehazing
AT shilpasuresh oddnetahybriddeeplearningarchitectureforimagedehazing