基于图像分割的去雾算法

针对雾霾条件下，所得的单幅图像出现降质现象，因而对视觉效果要求高的图像进行复原具有必要性。基于图像分割的去雾算法以暗通道先验模型为基础对大气光矢量A值的求取和透射率t(x,y)的处理方法实现改进。首先对单幅图像进行阈值分割找到天空区域，在所获取的天空区域部分结合skyline算法，可以找到精确的大气光矢量A值；进而对初始透射率t(x,y)采用改进的约束最小二乘方滤波进行优化，得到优化透射率t1(x,y)，最后将所得的大气光矢量A值和优化透射率t1(x,y)利用大气光传输物理模型复原。改进算法的去雾结果具备保留细致的边缘细节，同时具有高效的去除图像噪声能力。实验结果表明，与he方法相对比，改进去雾算法的处理时间大程度缩短的同时，图像效果得到了提升。

RapidSSDefogging Algorithm Based on Image SegmentationFu Hui,Li Linfei

Directed at the necessity of single image restoration under the haze, this paper improves the acquisition methods of global atmospheric light A and transmittance based on the prior model of dark channel; adopts the threshold segmentation on single image to find the sky area and the accurate value of A with the skyline algorithm; and optimizes the transmittance with the constrained least-square filtering which reserves intensive fringe details with strong de-noising ability. According to the experiment results, it is indicated that the image quality has been improved compared with He method as the processing time of defogging is drastically shortened.