基于暗通道理论的快速单幅图像去雾算法

Fast haze removal algorithm using dark channel prior

为了恢复因大气粒子散射和吸收作用而降质的退化图像,传统的基于暗通道先验规律的图像去雾算法去雾效果较好,但因图像抠图计算复杂度高而不具有实效性。对去雾算法进行了研究,分析了现有算法对雾霾图像恢复的优缺点,新算法利用暗元先验,将白平衡理论作用于大气光照,利用三边滤波对暗通道图像进行边缘细化,进而恢复场景反照率。实验结果表明,该算法复杂度低,去雾时间短,能较快恢复场景信息。

To recover degraded images due to the atmosphere particles absorbing and scattering the light,the dark channel prior algorithm received good results in image removal,but soft matting algorithm was computational complex and quite time-consuming.The advantages and disadvantages of the existing defogging algorithms were analysed,based on which a new fast defogging algorithm using dark channel prior was proposed which firstly used the white balance and then used trilateral filter to obtain an edge-preserving dark channel image,thus the scene albedo was recovered.Experimental results proved that this algorithm was low in complexity and could effectively remove haze from a foggy image with a little time.