YUV图像实时去雾算法的优化与改进  

Optimization and improvement of real-time dehazing algorithm for YUV image

作　　者：李博伦[1] 凌强[1] 朱学俊[1] 徐骏[1] 

机构地区：[1]中国科学技术大学信息科学技术学院自动化系,安徽合肥230027

出　　处：《微型机与应用》2016年第1期77-80,共4页Microcomputer & Its Applications

基　　金：国家自然科学基金(61273112)

摘　　要：针对已有的基于DSP的实时去雾算法出现的问题进行优化和改进。在降采样倍数过大时,去雾效果会出现一些副作用,包括纹理细节丢失、块效应和亮度偏暗。针对这三个问题,本文采用均值降采样、双线性插值升采样、透射率补偿等方式进行优化。最终针对实际YUV图像进行了优化方法的实验,表明可以达到较好的去雾效果。In this paper, a method is presented to optimize and improve existing DSP-based real-time dehazing algorithms, which has some problems. When the down-sampling ratio is too large, there are some negative effects on dehazed image, including the loss of texture details, block artifacts and being too dark. To resolve these three issues, the proposed method adopts average downsampling, bilinear interpolation upsampling and transmittance compensation to do optimization. Finally experimental results confirm that this method can achieve good dehazing performance.

关 键 词：去雾 优化 降采样 升采样 

分 类 号：TP368[自动化与计算机技术—计算机系统结构]