【摘要】 针对暗通道先验算法在天空区域失效和易产生光晕的问题，提出一种基于容差机制与超像素的阈值暗通道去雾改进算法。首先，使用超像素算法将图像分割为景深一致的局部块，用局部块替换暗通道先验算法的固定方块；然后，通过计算局部块内元素的方差来判断离散程度，根据离散程度选择不同的暗通道计算策略；最后，在HSV空间中定位高亮区域，并用容差机制对高亮区域的透射率进行修正。实验结果表明，该算法能有效去除天空区域浓雾，并能改善光晕效应。更多还原

【Abstract】 Aiming at the problem that the dark channel prior algorithm fails in the sky area and is prone to halo, an improved algorithm for threshold dark channel defogging based on tolerance mechanism and superpixels is proposed.Firstly, the superpixel algorithm is employed to segment the image into local blocks with the same depth of field, which is applied to replace the fixed blocks of the dark channel prior algorithm.Then, by calculating the variance of the elements in the local block, the degree of dispersion is judged, and based on the degree of dispersion different options of dark channel calculation strategy are selected.Finally, the highlighted area is located in the HSV space, and the tolerance mechanism is utilized to correct the transmittance of the highlighted area.Experimental results show that the algorithm can effectively remove the dense fog in the sky area and improve the halo effect.更多还原