Fast Low Lighting Video Enhancement Algorithm Based On Dark Channel Prior

Multi-media technology and video monitor technology have already become thecommon communication and record tools. But when shooted in low illuminationenvironment such as conference room, cinema and the dark, the video will bedegenerated with low contrast, low signal-to-noise ratio and poor visual effect. Underthis circumstance, it is difficult to show the details of images, which makessubsequent image feature extraction, image analysis and recognition restricted. So it isvitally important and practical to enhance and denoise the low lighting video. Thiskind of technology can be applied in monitoring and control system, patternrecognition, medical image analysis, etc.This paper introduces traditional low lighting enhancement and denoisingtheories, illustrates their different processes and points out the defects of abovemethods in solving the local details brightness, noise, real-time processing, etc.Compared to traditional enhancement algorithm, existing “low lighting videoenhancement algorithm using dark channel prior” has higher fidelity with more details.But because of the underexposure of low lighting images, the noise in images afterenhancement will badly degenerate images. Therefore, this paper proposes animproved efficient method based on the dark channel prior to enhance low lightingvideo. Based on existing “low lighting video enhancement algorithm using darkchannel prior”, this paper takes advantage of both the information of time domain andspace domain and bilateral filter to denoise the low lighting video. This algorithm cannot only denoise images, but also remain more motion information. Meanwhile, thealgorithm is optimized to adapt to GPU CUDA programming by partitioning andparalleling the data. By using GPU CUDA parallel programming, the algorithm canachieve real-time processing performance.Experimental results show that the algorithm which achieves the goal of gainingclearer and more natural high quality video can effectively suppress the noise in thevideo, improve the contrast of videos and highlight the details.