结合暗原色图像参考的单幅图像去雾*

针对在暗原色先验理论下对天空区域的透射率估计总是过于偏小的问题，提出了一种基于暗原色图像的透射率纠正方法。以暗原色图像为参考，提取其天空等明亮区域的灰度归一化值替换生成新的透射率图。该方式提高了天空区域的透射率值，使其更接近于真实值。实验结果表明，对于各类雾化图像，在无需调整各参数值的情况下，该算法均能很好地避免天空区域出现色差失真问题；同时对于不含天空区域的图像，去雾仍然有效。此外，算法执行时无需进行复杂的天空区域识别及提取等操作。算法在实际去雾应用中实现了无参化，自动化程度高；对于分辨率为640?480彩色图像，速度可达1.5帧/s。

Single image dehazing based on dark colors image reference

Based on the dark colors map, this paper proposed a correction method of transmission maps to work out the estimated transmission value of sky area in too small under the dark color prior theory. The normalized gray values extracted form the sky and other bright areas generated a new transmission map in dark colors map as a reference image. It greatly improves the transmission values of the sky area, which brings them closer to the true values. Experimental results show that without the need to adjust each parameter value the algorithm can well avoid the color distortion problem in the area of the sky, and for the images without the sky area, the dehazing result is also good. Especially, the algorithm can avoid executing the complex operation of the sky area identification and extraction. It achieves parameterlessly in the application; and for a resolution of 640?480 color images, the speed is up to 1.5 frame/s.