摘要
针对暗原色先验图像去雾算法存在的缺点,本文提出了一种利用均值滤波计算透射率的图像快速去雾算法。该算法首先对图像进行白平衡处理,达到将沙尘、偏色雾霾图像转化为白色雾天图像的目的,然后利用均值滤波计算图像暗通道,并对相关结果进行适当调整进而求得透射图,再通过Canny算子求取图像的边缘信息并划分出天空区域、计算大气光,最后根据相关参数还原无雾图像。通过实验对比验证,该算法计算复杂度低,复原结果清晰自然,具有较好的实时性。
Aiming at the disadvantages of the algorithm of single image haze removal using dark channel prior,this paper propose a fast image dehazing algorithm based on mean filter. First of all,the white balance is performed to ensure the picture degraded by fog or dust being pure white,then getting the minimum value of three color channel in each pixel and the transmission values by modified mean filter.Then use the canny edge detector to obtain the edges and expand the edges to eliminate the negative effect of small bright zone and get the accurate atmospheric light value. Finally,restore the foggy image with the atmosphere attenuation model. Experiments and comparisons show that this method generates good results with low computation complexity.
作者
唐鉴波
赵波
陈国樑
佟帅
习立坡
TANG Jian⁃bo;ZHAO Bo;CHEN Guo⁃liang;TONG Shuai;XI Li⁃po(Army Engineering University Communications officer School,Chongqing 400035,China;32369 Troops,Beijing 100043,China;32178 Troops,Beijing 100012,China)
出处
《电子设计工程》
2020年第1期189-193,共5页
Electronic Design Engineering
