Abstract
Shadows can hinder image interpretation in aerial remote sensing images. The existing shadow detection methods focus on all shadow regions and detect the shadow regions directly, but they ignore the fact that salient shadows have a more significant effect. In this work, a novel edge-aware spatial pyramid fusion network (ESPFNet) under a multitask learning framework is proposed for salient shadow detection in aerial remote sensing images. ESPFNet has three components: a parallel spatial pyramid (PSP) structure; an edge detection module (EDM); and an edge-aware multibranch integration (EMI). The PSP structure is constructed to extract multiscale features from the input image and fuse them gradually. The EDM then integrates the shallow features and deep features to detect the shadow edges. Finally, the EMI incorporates the edge features with multibranch features, and then concatenates them with the shallow features to generate the salient shadow detection result. The experimental analyses confirm the effectiveness of the ESPFNet method in both the qualitative and quantitative performance, compared to the existing methods, with the F-score reaching 92.04% in the salient shadow test set.
Highlights
S HADOW is a widespread phenomenon in high-resolution aerial images, especially in urban regions, due to the numerous high-altitude land covers, such as buildings, bridges and trees [1], [2]
It should be noted that, for the results of spectral ratioing segmentation (SRS) and extended random walker based shadow detection (ERWSD) shown in Figs. 15(a) and (b) and 16(a) and (b), the falsely detected vegetation regions are not removed through this postprocessing, because some of the vegetation regions are too large
For the results of U-Net shown in Figs. 15(c) and 16(c), most of the non-salient shadow regions are removed, but some small regions are still connected with the building shadows, which are difficult to exclude with the area threshold
Summary
S HADOW is a widespread phenomenon in high-resolution aerial images, especially in urban regions, due to the numerous high-altitude land covers, such as buildings, bridges and trees [1], [2]. Shadows can provide additional geometric information for object location and altitude estimation, but lead to radiometric information loss, making the image interpretation more difficult [3], [4], [5]. For both the geometric and radiometric applications of remote sensing images, shadow detection is of great importance. Manuscript received January 24, 2021; revised March 1, 2021; accepted March 12, 2021.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
