Dark Channel Prior Dehazing Combined with Image Quality Assessment Applied to Haze Detection.

Saved in:

Bibliographic Details
Title:	Dark Channel Prior Dehazing Combined with Image Quality Assessment Applied to Haze Detection.
Authors:	Chih-Ping Yen¹ peter@mail.cpu.edu.tw, Cheng-Tan Tung² tung@mail.cpu.edu.tw, Hsin-Hsiung Kao² kao@mail.cpu.edu.tw, Chen-Yu Li² cyli@mail.cpu.edu.tw
Source:	IAENG International Journal of Computer Science. May2025, Vol. 52 Issue 5, p1463-1471. 9p.
Subjects:	Intelligent transportation systems, Image databases, Pearson correlation (Statistics), Traffic safety, Structural models
Abstract:	Driving in hazy conditions with low visibility can easily lead to accidents due to delayed reactions. Consequently, integrating image processing techniques into existing traffic CCTV systems offers a cost-effective solution for real-time haze detection and early warning systems. This study develops a novel haze detection model by combining the dark channel prior dehazing algorithm with advanced image quality assessment metrics, ensuring accurate haze degree classification. Experimental results demonstrate that this model combined with the Structural SIMilarity (SSIM) image quality assessment method can effectively capture the changes in luminance, contrast, and structure of haze images to distinguish different haze degrees. Therefore, the performance Accuracy @1 reaches 96.55%. In addition, the correlation coefficients Pearson, Spearman, and Kendall also proved once again that the correlation between the ranking results calculated by SSIM and the actual ranking of haze degrees is the highest. Pearson and Spearman are both 0.9862, and Kendall is 0.9770. Moreover, the proposed model conducts a sensitivity analysis on the local patch size and the transmission parameter w. The results show that with w=1 and the local patch size is between 39×39 and 47×47, the model can capture haze information more effectively and thereby improve performance. Then, the CHIC image database with haze levels was used for practical verification to confirm that the proposed model can indeed detect different haze densities correctly. Finally, we also explore and analyze the results of nighttime haze image detection. In the future, this model can not only be applied to existing CCTV infrastructure for haze concentration monitoring, but can also be further coordinated with other ITS modules. [ABSTRACT FROM AUTHOR]
	Copyright of IAENG International Journal of Computer Science is the property of International Association of Engineers (IAENG) and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database:	Engineering Source

Description
Abstract:	Driving in hazy conditions with low visibility can easily lead to accidents due to delayed reactions. Consequently, integrating image processing techniques into existing traffic CCTV systems offers a cost-effective solution for real-time haze detection and early warning systems. This study develops a novel haze detection model by combining the dark channel prior dehazing algorithm with advanced image quality assessment metrics, ensuring accurate haze degree classification. Experimental results demonstrate that this model combined with the Structural SIMilarity (SSIM) image quality assessment method can effectively capture the changes in luminance, contrast, and structure of haze images to distinguish different haze degrees. Therefore, the performance Accuracy @1 reaches 96.55%. In addition, the correlation coefficients Pearson, Spearman, and Kendall also proved once again that the correlation between the ranking results calculated by SSIM and the actual ranking of haze degrees is the highest. Pearson and Spearman are both 0.9862, and Kendall is 0.9770. Moreover, the proposed model conducts a sensitivity analysis on the local patch size and the transmission parameter w. The results show that with w=1 and the local patch size is between 39×39 and 47×47, the model can capture haze information more effectively and thereby improve performance. Then, the CHIC image database with haze levels was used for practical verification to confirm that the proposed model can indeed detect different haze densities correctly. Finally, we also explore and analyze the results of nighttime haze image detection. In the future, this model can not only be applied to existing CCTV infrastructure for haze concentration monitoring, but can also be further coordinated with other ITS modules. [ABSTRACT FROM AUTHOR]
ISSN:	1819656X