基于迭代直方图均衡化的常规光源下水下成像增强算法

Underwater Image Enhancement Algorithm Based on Iterative Histogram Equalization with Conventional Light Source

为了增强常规光源下水下图像的视觉对比度,提高水下图像的图像质量,提出一种基于迭代直方图均衡化水下图像增强算法.首先通过Retinex模型将水下图像分解为细节层和光照层图像.然后推导出一个图像增强模型,该模型能够在保证韦伯对比度的前提下完成图像增强工作.接着提出一种基于迭代直方图的直方图均衡化算法对光照层图像进行对比度增强,并通过S形状函数对细节层图像进行对比度拉伸.最后,合并拉伸后的细节层图像和增强后的光照层图像,进而获得较佳的图像增强效果.实验结果表明,该算法能够有效地提升水下图像的视觉对比度,图像信息熵值及均值结构相似度高于其他算法,图像的视觉效果得到显著提高.

To enhance visual contrast and image quality for underwater image with conventional light source,an image enhancement algorithm based on iterative histogram equalization is proposed.Firstly,underwater image is decomposed into detail layer image and illumination layer image through Retinex model,and an image enhancement model is derived to improve the contrast effect of image on the basic of Weber contrast constant.Then,an image enhancement algorithm based on iterative histogram equalization is proposed to enhance the contrast of illumination layer image,and S-shade function is employed to stretch the contrast of detail layer image.Finally,the stretched detail layer image and enhanced illumination layer image are mixed to derive the preferable enhanced underwater image.Experimental results show that the visual contrast of underwater image is enhanced validly by the proposed algorithms,and the entropy and mean structural similarity index are higher than other image enhancement algorithms.Meanwhile the visual effect is ascended observably.