Enhancement and license plate detection in nighttime scenes using LDR images fusion from a single input image

The incidence of traffic accidents during nighttime is higher than daytime; hence the clues from nighttime images are urgently needed. However, the light source from a camera works only for certain distance, resulting in the defect of underexposure. In this paper, we present a high dynamic range imaging method for vehicle images under a nighttime environment. The idea consists of three steps based on a single input image: (i) the license plate detection, (ii) generation of differently exposed images with license stroke thickening, and (iii) integrating them into a synthesized HDR image, in which the license plate regions and the backgrounds are better visualized. Therefore, the proposed method can be applied to mobile imaging devices to provide the HDR function. Experimental results show that the proposed approach effectively enhance contrast, especially for nighttime vehicle scenes, while still preserving image details, color, and information.