一种改进的基于暗通道先验的快速去雾算法

An improved fast defogging algorithm based on dark channel prior

针对暗通道先验去雾算法复杂度过高,存在大片天空区域不适用的情况,提出一种改进的基于暗通道先验的快速去雾算法。首先,通过对灰度图像迭代出最佳阈值将天空区域分割出来;然后采用文中提出的方法分别对天空和非天空部分求解大气光强度,以获得精确的透射率,为使图像边缘信息能够保留完整,采用梯度强化的方法对图像进行边缘锐化;最后通过大气散射模型恢复出无雾图像,采用自动色阶算法平衡图像亮度。实验结果表明,经过与4种算法对比后,能够较好的适用于存在大片天空区域的有雾图像,图像具有较好的视觉效果,能够保留图像的边缘信息、细节纹理,降低了处理时间,提高了运行效率,同时为后续的目标识别等领域奠定基础。

Since the dark channel prior defogging algorithm has high complexity and is not applicable to the image with large sky region,an improved fast defogging algorithm based on dark channel prior is proposed.The sky region is segmented by iterating the optimal threshold value of gray images,and then the atmospheric light intensity of the sky region and the non sky region is solved respectively by means of the method proposed in this paper,so as to obtain accurate transmissivity.The gradient strengthening method is used to sharpen the edge of an image to retain the integrity information of image edge.The defogged im age is recovered by the atmospheric scattering model,and the image brightness is balanced by means of the automatic color lev el algorithm.The experimental results show that,in comparison with the four algorithms,this algorithm can be better applied to the foggy images with large sky region,and the images have better visual effect.It can retain the edge information and detail tex ture of the image,reduce the processing time and improve the operation efficiency,which may lay the foundation for subsequent target recognition and other fields.