基于改进二值分割的图像去雾算法

Dark channel image dehazing based on the improved binary image segmentation

为改善暗通道先验理论得出的去雾图像天空区域出现光晕现象、整体亮度偏低等问题,提出了一种基于二值图像分割的去雾算法.该算法利用二值图像分割法将引导滤波平滑后的有雾图像分割成两个区域,分别求取两个区域的大气光值,并将二者的平均值作为最终大气光值,来估计全局透射率;再次利用二值图像分割法将有雾图像分割成明亮与非明亮区域,并引入容差值优化明亮区域的透射率;最后对相应区域图像进行去雾,去雾后的图像通过YUV颜色空间调整亮度和图像加权融合实现进一步优化.实验结果表明,与其他算法相比,该算法处理后的去雾图像在天空处的光晕现象得到了明显的改善,图像亮度适中,清晰度也得到了提高.

In order to improve the problem of halo phenomenon and low overall brightness in the sky area of clear images derived from the dark channel prior theory,the algorithm first smoothed the image through guided filtering,and then optimizes the selection of atmospheric light valued with improved binary segmentation,and calculated the global atmospheric light value,estimated the global transmittance;then used the improved binary image segmentation method to divide the original foggy image into bright and non-bright areas,and optimized the inaccurate transmittance with the tolerance value to achieve defogging;Finally,the defogging image was further optimized through YUV color space and image weighted fusion.Experimental results showed that compared with other algorithms,the clear image processed by this algorithm has a significant improvement in the halo phenomenon in the sky,and the overall brightness was improved.