Fast efficient algorithm for enhancement of low lighting video

Abstract

We describe a novel and effective video enhancement algorithm for low lighting video. The algorithm works by first inverting the input low-lighting video and then applying an image de-haze algorithm on the inverted input. To facilitate faster computation and improve temporal consistency, correlations between temporally neighboring frames are utilized. Simulations using naive implementations of the algorithm show good enhancement results and 2x speed-up as compared with frame-wise enhancement algorithms, with further improvements in both quality and speed possible.