基于改进双直方图均衡化的舰载图像去雾方法

Shipborne Image Defogging Method Based on Improved Double Histogram Equalization

舰载图像具有大面积同色高亮区域,现有直方图均衡化在舰载图像去雾时存在亮区过度增强现象。针对这一问题,论文提出一种基于改进双直方图均衡化的舰载图像去雾算法,首先对归一化直方图中小于1的部分gamma校正,并设置过度增强阈值T对较大的直方图数据限制,然后以中位数像素作为分割点,计算左、右子直方图的灰度分布概率直方图并获得映射表;最后将两映射表合并,获得去雾后图像。实验结果表明,以论文自制舰载图像数据集为测试集,相比于基于灰度图像均值分割的直方图均衡化和等面积的双直方图均衡化,论文算法平均亮度误差分别缩短了46.27和7.10,结构相似性分别提高了0.25和0.16,可以有效改善现有直方图均衡化在对舰载图像去雾时的亮区过度增强问题。

Shipborne images have large areas of bright areas with the same color.The existing histogram equalization has the phenomenon of over enhancement of bright areas when ship borne images are defogged.In order to solve this problem,this paper proposes a shipborne image defogging algorithm based on improved double histogram equalization.First,part of the normalized his⁃togram that is less than 1 is corrected for gamma,and the excessive enhancement threshold T is set to limit the larger histogram da⁃ta.Then,the median pixel is used as the segmentation point to calculate the gray distribution probability histogram of the left and right sub histograms and obtain the mapping table.Finally,the two mapping tables are merged to obtain the defogger image.The ex⁃perimental results show that the average brightness error of the algorithm in this paper is shortened by 46.27 and 7.10 respectively,and the structural similarity is improved by 0.25 and 0.16 respectively,compared with the brightness preserving bi-histogram equal⁃ization and dualistic sub-image histogram equalization method,using the self-made shipboard image dataset as the test set,which can effectively improve the over enhancement of the bright area of the existing histogram equalization when defogging shipboard im⁃ages.