Title: Shadow removal based on separated illumination correction for urban aerial remote sensing images

Author: Luo, S (Luo, Shuang); Shen, HF (Shen, Huanfeng); Li, HF (Li, Huifang); Chen, YM (Chen, Yumin)

Source: SIGNAL PROCESSING Volume: 165 Pages: 197-208 DOI: 10.1016/j.sigpro.2019.06.039 Published: DEC 2019

Abstract: The presence of shadows in urban aerial images degrades the image quality and reduces the application accuracy. Removing shadows and recovering the ground information is therefore a crucial issue. The existing shadow removal methods can correct the shadow information, but the inconsistency between the corrected shadow and non-shadow areas is still obvious. A novel shadow removal method based on separated illumination correction is proposed in this paper, in which the shadow removal is only performed on the shadow-related illumination. A spatially adaptive weighted total variation model is constructed to obtain the shadow-related illumination and the shadow-free reflectance. The objects in the shadows are detected based on the reflectance, and object-oriented illumination correction is then implemented to compensate the shadow regions. The shadow removal results can be obtained by combining the corrected illumination and the reflectance. Three aerial remote sensing images were selected for the experiments, and two quantitative evaluation methods are introduced: the shadow standard deviation index and classification analysis. The results are shown and compared with four existing methods by visual and quantitative assessments, which indicate that the proposed method can yield more visually natural shadow-free images and show a better performance in the quantitative indices. (C) 2019 Elsevier B.V. All rights reserved.

WOS: 000485855000019

Language: English

Document Type: Article

Key words of author: Shadow removal; Spatially adaptive; Illumination correction; Aerial images

Keywords Plus: RESOLUTION; RECONSTRUCTION; INFORMATION; MODEL

Addresses: [Luo, Shuang; Shen, Huanfeng; Li, Huifang; Chen, Yumin] Wuhan Univ, Sch Resource & Environm Sci, Wuhan, Hubei, Peoples R China.

[Shen, Huanfeng] Collaborat Innovat Ctr Geospatial Technol, Wuhan, Hubei, Peoples R China.

[Shen, Huanfeng] Wuhan Univ, Key Lab Geog Informat Syst, Minist Educ, Wuhan, Hubei, Peoples R China.

Addresses of reprint authors: Li, HF，Wuhan Univ, Sch Resource & Environm Sci, Wuhan, Hubei, Peoples R China.

Email: huifangli@whu.edu.cn

Impact factor：4.068