【作者】 王晶；

【导师】 李海滨；

【作者基本信息】 燕山大学 ， 控制理论与控制工程， 2018， 硕士

【摘要】 近年来,海洋资源逐渐成为世界各国开发的战略重心,海洋资源如果得到合理的开发利用,可以缓解目前陆地上资源匮乏的现状。在海洋军事、海底目标检测跟踪等应用中,水下光学图像的采集与增强是获得水下信息进行研究的前提,然而由于水下环境对光的吸收、前向散射、后向散射作用的影响,水下光学图像往往存在颜色失真,图像不清晰、对比度低、细节丢失等多种问题,因此水下图像处理具有非常重要的研究意义。本文根据水下图像退化的原因,针对水下图像增强与复原技术进行了研究,主要做了以下研究工作:(1)本文针对在水下环境中获得的图像存在严重色偏现象,提出了一种改进的直方图均衡化算法来进行图像颜色校正。该方法基于一定的先验知识对水下图像进行颜色校正,对水下图像的颜色平衡效果显著,相对于一些经典的白平衡算法,该方法更加适合水下图像。(2)针对水下降质图像模糊、细节丢失问题,本文提出了一种基于分数阶微分的图像细节显著增强的算法。该算法不仅可以增强图像细节信息,还可以减少水下图像噪声增加,所以本文通过构造合适的分数阶掩膜算子对水下图像进行分数阶微分。经过处理后的水下图像放大后可以明显发现相较处理前的图像像素点数更加密集。(3)本文受到引导滤波与图像融合的思想启发,提出了一种基于引导融合的水下图像复原算法。该算法把利用改进的直方图均衡化得到的色彩平衡图像进行对比度增强后作为输入图像,把利用分数阶微分细节增强的图像作为引导图像,进行引导融合。经过引导融合后的图像更加真实,自然。(4)目前关于水下图像质量评价方法的相关研究比较少,空气中去雾算法的质量评价算法又不能直接应用到水下图像中来,因此本文介绍了几种比较适合水下图像质量评价的方法并把它们用于本文算法的质量评价体系中。通过实验结果的验证与分析,本文提出的算法对大部分的水下图像有较好的增强效果。更多还原

【Abstract】 In recent years,marine resources have gradually become the strategic focus of the development of various countries in the world.If marine resources are reasonably exploited and utilized,it can alleviate the current situation of lack of resources on land.In the application of marine military,underwater target detection and tracking,the acquisition and enhancement of underwater optical image is the prerequisite for the study of underwater information.However,due to the absorption of light in underwater environment,forward scattering,back scattering,underwater imaging system is difficult to achieve satisfactory results.Underwater optical images often have many problems,such as color distortion,image blur,contrast reduction,detail loss and so on.Therefore,underwater image processing is of great significance.According to the causes of underwater image degradation,this paper studies the underwater image enhancement and restoration technology,mainly doing the following research work:(1)In this paper,an improved histogram equalization algorithm is proposed to correct the color of images in underwater environment.This method is based on a certain prior knowledge to correct the color of underwater image,and the effect of color balance of underwater image is remarkable.Compared with some classical white balance algorithms,this method is more suitable for underwater images.(2)Aiming at the problem of image blur and detail loss in water descent image,this paper presents an algorithm for image detail information enhancement based on fractional differential.This algorithm can not only enhance the details of the image,but also reduce the increase of underwater image noise.So we construct the proper fractional mask operator to perform fractional differential of underwater image.After the underwater image is magnified,it can be found that the pixel number is denser than that before processing.(3)This paper is inspired by the idea of guided filtering and image fusion.An underwater image restoration algorithm based on guided fusion is proposed,which uses the improved histogram equalization to enhance the contrast of the color balance image.The image enhanced by fractional differential details is used as the guide image to guide fusion.After the final guided fusion the image is more realistic and natural.(4)Because of the lack of research on underwater image quality evaluation methods,the quality evaluation algorithm of de-fogging algorithm in the air can not be directly applied to underwater images.Therefore,we introduce several methods suitable for underwater image quality evaluation,and apply them to the quality evaluation system of our algorithm.Through the verification and analysis of the experimental results,the algorithm proposed in this paper has a good enhancement effect on most underwater images.更多还原