基于CUDA平台的实时去雾  

Haze removal in real-time based on CUDA

作　　者：黄飞虎[1] 兰时勇[1,2] 吴健[1] 刘东辉[1] 

机构地区：[1]四川大学计算机学院,成都610065 [2]视觉合成图形图像技术国防重点学科实验室,成都610065

出　　处：《计算机应用》2013年第A02期183-186,共4页journal of Computer Applications

基　　金：国家863计划项目(2012AA011804;2012AA011801)

摘　　要：在雾、霾等天气条件下,几乎不能采集到理想质量的图片。图像去雾技术的任务就是最大限度地消除不利天气对图像采集的影响。统一计算架构(CUDA)是NVIDIA公司所推出的充分利用图形处理单元(GPU)的计算能力进行大规模计算的一种构架。基于何恺明暗通道的观点,通过CUDA平台实现了对含雾图片的实时去雾。实验结果表明,在保证去雾质量和去雾自然度的前提下,对于1280×720的实时场景,可以达到每秒20帧的处理速度,基本满足实时去雾的要求。It almost could not get an ideal image, under certain circumstance, such as haze, fog. And the technique, haze removal, is used to dispel the bad effort by weather. Common Unified Device Architecture （CUDA） is a kind of architecture from NVIDIA, which makes full use of the ability of the Graphics Processing Unit （GPU） to meet large-scale computation. It makes the haze removal in real-time come true at the footstone of He Kaiming＇s opinion, the dark channel, by CUDA. According to the result of experiment, it could do at the speed of 20 frames per second, to the image at the size of 1 280 ＊ 720 pixels. It meets real-time haze removal requirements.

关 键 词：统一计算架构 图形处理器 并行计算 图像处理 图像去雾 实时 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]