A naive but effective post-processing approach for Dark Channel Prior (DCP)

Danny Ngo Lung Yao and Abdullah Bade and Iznora Aini Zolkifly and Paridah Daud (2023) A naive but effective post-processing approach for Dark Channel Prior (DCP). Springer, 169. pp. 67-76. ISSN 2367-4512

[img] Text
FULLTEXT.pdf
Restricted to Registered users only

Download (3MB) | Request a copy

Abstract

Dark Channel Prior (DCP) is originally introduced to remove the haze effects from a digital image. Though the effectiveness of the DCP approach on haze removal, the DCP approach often leads the recovered image to become darker even though the haze effects had been removed. The images with the same or similar levels of color mean are likely without the problem of color shifts. Therefore, we have created a color mean adjustment method to adjust for the color shift by balancing the means of the color channels. CLAHE is employed in this study to boost the image’s contrast, while the color mean adjustment method is utilized to smooth its final appearance. Throughout the experiments like visual comparison analysis, Peak Signal-to-Noise Ratio (PSNR), and Universal Quality Index (UQI), our proposed method proved to be a highly effective post-processing approach for the DCP approach as it suppresses more image noises, enhance image quality, and, more importantly, allows the DCP approach to be used on underwater images. Besides, our proposed method also resolves the dark look issue of the DCP approach.

Item Type: Article
Keyword: Color mean adjustment, CLAHE, DCP
Subjects: N Fine Arts > NC Drawing. Design. Illustration > NC1-1940 Drawing. Design. Illustration > NC730-758 Technique
N Fine Arts > NC Drawing. Design. Illustration > NC1-1940 Drawing. Design. Illustration > NC950-(996) Illustration
Department: FACULTY > Faculty of Social Sciences and Humanities
Depositing User: JUNAINE JASNI -
Date Deposited: 05 Aug 2025 15:21
Last Modified: 05 Aug 2025 15:21
URI: https://eprints.ums.edu.my/id/eprint/44759

Actions (login required)

View Item View Item