We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Analysis and selection of haze-relevant features for haze detection.
- Authors
Kadian, Garima; Kumar, Rajiv
- Abstract
Due to smog, haze, fog, etc., the visibility of outdoor images degrades significantly. Limited visibility leads to the failure of many computer vision applications like tracking an object, intelligent transportation, etc. Many image dehazing methods have been developed to resolve this problem. But, most of the existing dehazing techniques are applied directly to the image regardless of the presence or absence of haze, which results in image deterioration. For real-world applications, it is vital to know whether the obtained image needs to be processed by dehazing methods. Hence, haze detection plays an essential role. Most of the existing techniques of haze detection used multiple features without considering their need. Thus, the proposed method presents a study analyzing different haze-relevant features. The main contributions of the proposed approach include: (i) by using haze-relevant features on RESIDE and NH-Haze datasets, a proposed dataset is prepared. (ii) analysis of features is done using multiple feature selection methods(iii) mapping between selected features and classification models. The results demonstrate that a set of features performs better when compared with another set of features.
- Publication
Multimedia Tools & Applications, 2023, Vol 82, Issue 25, p39057
- ISSN
1380-7501
- Publication type
Article
- DOI
10.1007/s11042-023-14771-w