Road Image Shadow Removal Method Based on Retinex Algorithm
Zhang Chong, Yang Liu · 2016
There were some shadows, such as cars and trees on road images, which make great interference to extract and recognize the image features. A novel algorithm was proposed to remove the single road image shadows based on Retinex model. The algorithm consists of three steps: first, the study converted RGB space into HSV space and abstracted saturation information to detect the shadow regions in road scene. Second, set up light scale factor to shadow regions to fight back a certain illuminant. Finally, The Multi Scale Retinex algorithm was adopted in shadow and non-shadow regions separately to eliminate the effect of the illuminant. Images had high-quality results that shadow regions restored the illuminance, color and texture. Experimental results demonstrate the effectiveness of this algorithm.