Effective dual stage fusion model for single image dehazing

Haze is an atmospheric phenomenon that greatly decreases the visibility of images captured by digital camera. The aim of the image dehazing is to restore the scene radiance from the hazy image. This paper proposes dual stage fusion model for an effective dehazing. In the proposed method, hazy image...

Full description

Saved in:
Bibliographic Details
Published inPattern analysis and applications : PAA Vol. 28; no. 3
Main Authors Varshney, Vikas, Panda, Jeebananda, Gupta, Rashmi
Format Journal Article
LanguageEnglish
Published London Springer London 01.09.2025
Springer Nature B.V
Subjects
Computer Science
Digital cameras
Digital imaging
Effectiveness
Haze
Image filters
Image restoration
Original Article
Pattern Recognition
Qualitative analysis
Image dehazing
Laplacian filter
Transmittance
Dual stage fusion
Color correction
Dark channel prior (DCP)
Online AccessGet full text

Cover

More Information
Summary:Haze is an atmospheric phenomenon that greatly decreases the visibility of images captured by digital camera. The aim of the image dehazing is to restore the scene radiance from the hazy image. This paper proposes dual stage fusion model for an effective dehazing. In the proposed method, hazy image is first pre-processed using Laplacian filter and Dark Channel Prior (DCP). To capture the edge information and reduce the noise in the input image, the Laplacian filtered image is combined with the dark channel image. Then, the output image obtained from DCP algorithm (without soft matting) is enhanced through dual stage fusion technique. The first stage fusion is the linear fusion of the two images obtained by applying AGCWD and WAAHE, and the second stage fusion is the DWT based fusion of the first stage fusion image and ACMHE image. Proposed work is implemented on various datasets such as O-HAZE, BeDDE and NH-HAZE. To see the effectiveness of proposed work, different images of same size with varied haze dense are used. AGCWD and WAAHE in dual stage fusion preserves small details and colors. Moreover, ACMHE gives more enhanced image in terms of contrast and sharpness. Both qualitative and quantitative analysis have been done on several images and compared with some existing dehaze methods. Results show that the technique proposed in the paper provides satisfactory result for dense hazy image also.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1433-7541
1433-755X
DOI:10.1007/s10044-025-01510-0