中文摘要：

基于暗通道先验的去雾算法被认为是最具代表性的图像复原算法.对透射图边缘进行平滑优化可以有效抑制复原图像边缘出现的块状效应,但仍然存在复原图像颜色失真的问题.针对薄雾图像复原色调偏暗,尤其是图像中浅白色景物色彩失真的问题,提出一种色彩保真的暗通道去雾算法,在利用Prewitt算子检测和平滑边缘以优化透射图的基础上,对透射控制参数进行分段限定.实验结果表明,该算法在改善图像边缘块状效应的同时减少了薄雾图像颜色失真的现象,对不同雾霾程度的图片都能获得较好的去雾效果。

英文摘要：

Defog algorithms based on dark channel prior are the most popular image restoration algorithms. The edge block effect in the restored image can be effectively inhibited by the smoothing of the transmission. However, color distortion is still observed. For correcting the color distortion of mist images especially it contains light white objects, an color fidelity algorithm based on dark channel prior is proposed. The transmission is optimized by Prewitt detection and smoothing of image edge and the transmission bound is defined by a piecewise function. The experimental results show that this algorithm is effective in both decreasing the edge block effect and correcting the color distortion for images with different haze degrees.