EUIEF: Enhancement of Underexposed Images via Exposure Fusion

摘要

Images captured in inappropriate light conditions can often result in flat color saturation, bad contrast and weakly illuminated images. In this paper, we propose a framework named as Enhancement of underexposed images via exposure fusion (EUIEF), to solve the problems associated with underexposed images. First, the enhanced input image is obtained by proposed Luminous Transform (LT) which enhances the low light images on the basis of their luminous value. Secondly, an enhanced input image is generated by Invert-Image enhancement method, which works by inverting a weakly illuminated image and then applying optimized enhancement algorithm on the inverted low light image. The enhancement of inverted low light image is carried out with the help of global atmospheric light, intensity of original scene objects and the amount of light reflected by the objects. The final enhanced low light image is the result of fusion technique which blends both enhanced images. Our work was evaluated on publicly available LOL dataset. The results suggest that the images enhanced by proposed EUIEF method possess better quality in terms of contrast enhancement, naturalness level and detail enhancement compared to existing methods. To validate the EUIEF, both qualitative and quantitative comparisons are conducted.