暗原色先验图像去雾改进算法

摘要

This paper presents a new method for image dehazing by combining with fractional differential and dark channel prior. This method firstly uses dark channel prior theory to process the fog of the foggy image. In order to save the storage and running time, the mathematical morphology method is used to optimize the rough estimate. The tolerance mechanism is introduced to achieve the recorrection of the transmittance image and get the accurate picture transmission. Finally, the fog image is used to obtain the fog image formation model. Then, the fractional order differential template is applied to carry out the secondary dehazing and sharpening. Experimental results show that the proposed method can get good dehazing effect, highlight the details and shorten the operation time of the algorithm.%文章提出了一种结合分数阶微分及暗原色先验的图像去雾新方法。此方法首先应用暗原色先验理论对有雾图像进行去雾处理，为节省存储及运行时间，利用数学形态学方法对粗略估计得到的透过率图进行优化，引入容差机制，实现对透射率图像的再次修正，得到准确透射率图像，最后借助有雾图像形成模型获得去雾图像；然后应用分数阶微分模版对图像进行二次去雾及锐化。实验表明，文章提出的方法能得到较好的去雾效果，且突出了细节，缩短了算法的运行时间。