基于多GPU的多路低照度视频实时增强系统-信号与信息处理专业论文.docxVIP

万方数据 万方数据 摘要 ABSTRA(了Abstract ABSTRA(了 Abstract The illumination enhancement of the low light image is an important research content in current video image processing field． Video imaging conditions in poor light environment，especially in the weak fight environment，call easily make the images or videos fuzzy and dark and hard to extract useful key information,especially detml information，which will affect the video viewing experience and the extraction of video information． Although the monitoring and photographic equipment becoming more and more powerful，many single channel video processing systems Can’t well deal with the eaton phenomena of high definition video，which greatly affect the actual use．It remains to improve in the practical application that has higher requirements on efficiency and effectiveness． For this，this paper designs a system based on image decomposition method and uses double GPU with good stability to handle multi-channel low illumination video，which meet the common application scenarios in the life．The paper proposes an enhancement algorithm based on the image decomposition after comparing and analyzing the classic low light color image enhancement algorithm．The paper also implements the real time system that processes video concurrently under the multi—gpu．The method and the system in the paper have an important theoretical and practical application value． In this paper，the main research contents and results are as follows： We make use of the image decomposition enhancement algorithm on the basis of improving and optimizing the traditional low light color image enhancement algorithm，changing to use fast guide filter that better efficiency，SO that increase processing speed and improve the efficiency at the same time． Compared with the traditional common enhancement algorithm，the new one is closer to actual appellations The paper sets up a multi—GPU system to process multichannel video in real time collaboratively，using multi-GPU coll