Abstract: To address the problem that the single image dehazing algorithm based on dark channel prior can make the bright areas’ estimated transmission too small in haze images so that lead to color distortion, the paper proposes a novel revaluating algorithm of transmission based on ratio. It sets the threshold value for the difference between the dark channel image and the atmosphere light intensity in order to improve the bright areas’ small transmission, utilizing the ratio between estimative transmission in advance and the overall transmission in the difference. The experimental results show that comparing with dark channel and tolerance mechanism dehazing algorithms, the improved approach can provide better results.

Key words: dark channel prior, ratio, transmission, dehazing

摘要： 针对暗原色先验去雾算法中雾霾图像明亮区域透射率估计过小，造成图像色彩失真的问题，提出一种新的基于比值重估透射率去雾算法。设定暗通道图与大气光强的差值阈值，利用预估透射率和全局透射率的比值重新估计透射率，从而改善明亮区域过小的透射率。实验结果表明，比值重估透射率去雾算法跟暗通道和容差机制去雾算法相比，该算法恢复的图像更接近于真实图像。

关键词: 暗通道, 比值, 透射率, 去雾