针对传统的暗通道先验算法在处理带有大面积天空区域的有雾图像时出现明显的块效应、色彩失真和亮度偏低等问题,提出了一种结合区域生长与容差机制的去雾算法。首先通过灰度图腐蚀求出暗通道；接着利用种子区域生长法分割出天空区域,并把天空区域的平均灰度值作为大气光值估计；然后结合大气散射模型得到粗略的透视率,并采用改良的容差机制和引导滤波对透视率进行修正和细化；最后,引入Retinex法对图像进行后处理,进一步调整色彩和亮度。实验结果表明,本文提出的去雾算法对带有天空区域的图像去雾效果明显,天空区域的色彩有了显著改善,图像整体清晰明亮。

Aiming at the obvious block effect,color distortion and low brightness of the traditional dark channel prior algorithm in processing foggy images with large areas of sky,a dehazing algorithm combining regional growth and tolerance mechanism is proposed.Firstly,the dark channel is obtained by gray scale image etching;Secondly,the sky region is segmented by the seed region growth method,and the average gray value of the sky region is estimated as the atmospheric light value;Then,combined with the atmospheric scattering model,the rough transmittance is obtained,and the improved tolerance mechanism and guided filtering are used to modify and refine the transmittance;Finally,Retinex method is introduced to post-process the image to further adjust the color and brightness.The experimental results show that the dehazing algorithm proposed in this paper has obvious dehazing effect when processing the image with sky region,the color of the sky region has been significantly improved,and the image is clear and bright as a whole.

国家自然科学基金(61862029)资助项目