期刊
ALEXANDRIA ENGINEERING JOURNAL
卷 77, 期 -, 页码 443-460出版社
ELSEVIER
DOI: 10.1016/j.aej.2023.06.046
关键词
Shadow detection; Principal component; Threshold; High -resolution; Water detection
With the continuous increase in the resolution of high-resolution remote sensing satellite images, shadows have increasingly impacted feature target recognition and classification in images, urban reconstruction, and image interpretation. We propose a novel mixed shadow detection index (MSDI) for extracting urban feature shadows, which achieved excellent shadow detection results and accurately recognized waters, easily missed small shadowed areas, and easily confused non-shadowed areas.
With the continuous increase in the resolution of high-resolution remote sensing satellite images, shadows have increasingly impacted feature target recognition and classification in images, urban reconstruction, and image interpretation. Although various shadow detection methods exist for different features, undetected small shadowed areas, misclassified dark areas, and highlighted non-shadowed areas remain. Moreover, it is still difficult to distinguish waters from shadowed regions owing to their similar characteristics. To further solve these problems, we propose a novel mixed shadow detection index (MSDI) for extracting urban feature shadows, using the original information of the first principal component that contains the differences between most of the different objects in the image, as well as the difference features in the shadowed areas and water bodies. We evaluated the effectiveness and robustness of our method by conducting comparison experiments using Gaofen-1, Gaofen-2, and WorldView-3 images from different scenes, times, and acquisition locations. Through visual analysis and data analysis of our method, we found that the method achieved excellent shadow detection results, with an average total accuracy 94% for shadow detection. The proposed shadow detection algorithm could accurately recognize waters, and it could also accurately recognize easily missed small shadowed areas and easily confused non-shadowed areas.& COPY; 2023 THE AUTHORS. Published by Elsevier BV on behalf of Faculty of Engineering, Alexandria University. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/ licenses/by-nc-nd/4.0/).
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据