本文主要研究了图像去雾技术以提高由于雾霾影响而导致的失真图像的主客观质量。对基于暗通道先验和大气散射模型的去雾算法进行了改进, 设计了新的天空区域计算和透射率阈值确定方法, 通过增强结果图的对比度和调整亮度, 进一步降低了算法的失真。还通过对图像加雾, 分别从主观质量评价和以结构相似度作为客观标准的客观评价两方面, 用数理统计方法比较了各个相关算法的性能。实验表明, 所提算法有效地降低了图像的失真(评价分最高), 使去雾图像有较高的主观质量(总体评价分并列最高)。

In this paper, a study was conducted in the image haze removal to improve the subjective quality of images distorted due to haze. First, the haze removal algorithm based on dark channel prior and atmospheric scattering model was improved, and a new method for calculation of the sky area and determination of the transmittance threshold was designed; then the distortion of such algorithm was further lowered by enhancing the image contrast and adjusting the brightness; finally, mathematical statistic method was used to compare the performance of each relevant algorithm by adding haze to the image, with subjective quality assessment and structural similarity as the objective criterion. The experiment in the existing haze images and the actually shot images indicates that the proposed algorithm can effectively reduce the distortion of images (with the highest assessment scores), thus keeping the haze removed images with relatively high subjective quality (tying first in the overall assessment).