Abstract: To solve the problem of the slow speed convergence and the distortion in the sky region, a dark channel prior haze removal based on threshold segmentation is proposed. Firstly, the sky region is divided by threshold segmentation. Secondly, utilizing parameter [K], fundamentally calculating the transmissivity to solve the distortion in the sky region. Meanwhile, the fast bilateral filtering algorithm reduces time consumption, and improves the convergence speed. Finally, the Gray World algorithm is adopted to improve image quality.

Key words: threshold segmentation, transmissivity, fast bilateral filter, Gray World algorithm

摘要： 为解决传统暗原色先验去雾算法运算速度慢，处理结果中天空区域易失真的问题，提出了一种基于阈值分割的暗原色先验图像去雾方法。首先利用阈值分割将天空区域初步分割，然后引入参数[K]对天空区域再分割，精细透射率，解决了天空区域失真问题，同时利用快速双边滤波降低算法复杂度，提高了运算速度，最后通过Gray World假设进行图像增强，提高了复原图像的清晰度和色彩保真度，得到更为清晰的复原图像。

关键词: 阈值分割, 透射率, 快速双边滤波, Gray World假设