Shadow detection and removal of road images with maximum extraction of road marker using multi-colour-space

Abstract

Road markers provide vital information ensuring safe vehicle manoeuvring on the road. Illumination condition, especially shadow, is the major contributor to the false detection of road markers. The shadow that falls on the road marker should be removed to reduce the error in abstracting road marker information. To tackle the adverse effect caused by shadows on the road marker, this paper attempts to detect shadows with image averaging from three different colour spaces, Saturation (S), Blue-difference (Cb) and Blue/Yellow Value (B), respectively, from HSV, YCbCr and LAB colour spaces. The combined data increases the accuracy of detecting shadow pixels for removal. The accuracy of the proposed algorithm reached 98.69%, with good feasibility and adaptability to perform in various road environments.