为了提高图像去雾效果, 采用三角形循环二分算法处理图像。首先分析基本暗通道算法图像去雾; 接着基于三角形循环二分算法对大气光值精确计算, 将图像沿对角线分成4个三角形区域, 只对雾浓度最较高的区域进行二分图形划分, 选取三角形最长边的中点, 连接到对应三角形最长边的顶点, 循环执行直到满足预先设定的面积阈值; 然后自适应调节透射率, 避免色彩失真; 最后给出了算法流程。实验仿真显示, 本文算法去雾后的图像在视觉上更加清晰, 可见边集合数目比、平均梯度比值、直方图相似度系数、综合评价指标相比其他算法较好, 能够满足图像去雾对清晰度、信息量的要求。

In order to improve the quality of image dehazing, the triangle loop dichotomy algorithm is presented. Firstly, the dark channel prior is used to analyze the image dehazing. Secondly, using triangle loop dichotomy algorithm ,the atmospheric light value is calculated accurately, the hazing image is divided into four triangular areas along the diagonal line, and the supreme hazing area is dichotomized only, which is selected the midpoint and connected the vertex of the longest side. The cycle execution is ended until satisfying the preset threshold. Thirdly, the adaptive transmission is made to avoid the color distortion. Finally, the algorithm process was given. The simulation results show that the triangle loop dichotomy algorithm is more clear in visual perception after image dehazing,number of edge sets, average gradient, the histogram similarity coefficient and the comprehensive evaluation index are better than other algorithms. It can satisfy the requirements of image clearness and information.