摘要
提出了一种基于单幅图像的去雾新算法.首先把图像归一化后从RGB彩色空间转换到HSI彩色空间,对色调分量运用四叉树分割法进行分割图像;分割后图像的每一局部小方块可以认为具有相同的场景深度,从而可以对每一局部小方块估计出空气光.然后再对亮度分量运用雾天图像光学模型,从雾的物理特性上去除雾对图像的影响.最后再对图像的饱和度分量进行校正,得到复原后的图像.该算法的主要优点是速度快,且不仅可以应用于彩色图像,也可以适用于灰度图像.通过该算法与其它几种算法的实验结果进行分析和比较,表明该算法能有效恢复出清晰图像.
A novel algorithm for single image dehazing was proposed. Firstly,the color space is converted from RGB to HIS after the image normalization. The quadtree-segmentation is applied in hue component divided the image into elementary square units which had the same scene depth. The local airlight could be estimated from eaeh square unit. To employ the characteristics of fog, the effects of fog on images are erased by using the optical model of foggy weather. Then the image of the restoration is obtained alter the saturation component correction. The main advantage of this algorithm is reat time performance that is can be used in defogging both for color images and gray images. Compared with other algorithms, it is shown that this algorithm is able to restore foggy images effectively.
出处
《光子学报》
EI
CAS
CSCD
北大核心
2011年第9期1419-1422,共4页
Acta Photonica Sinica
基金
福建省教育厅科技研究项目(No.JB10142)资助
关键词
单幅图像去雾
物理模型
图像复原
四叉树分割
Single image dehazing
Physics model
Image restoration
Quadtree-segmentation
