D-ViShaDeRec: Double Intensity of Video Shadow Detection, Removal, and Re-coloring in Autonomous Vehicle

Abstract

Shadows are the obstacle parts of a vision-based during the driving an autonomous vehicle (AV). The shadow also can be detected as an object. It is feared the AV’s system doesn’t well-work or hits another object. Many methods of the images and video shadow detection and removal in the AVs. However, to handle the various backgrounds on the video shadow and objects, we need to exploit many scenarios of single video shadow detection, removal, and semantic segmentation. We propose a new novelty method in the shadow video detection and removal which is combined with the re-coloring method in the segmentation process. It’s namely the D-ViShaDeRec. We also propose a new modified Sobel algorithm which has a kernel size of 5 × 5. The D-ViShaDeRec’s outcomes show , , , , , and of the recall, selectivity, precision, NPV, accuracy, and F1-Score, respectively.