【作者】 王浩；

【导师】 陈学东； 韩斌；

【作者基本信息】 华中科技大学 ， 机械工程， 2021， 硕士

【摘要】 图像处理在发展历程中已经取得了显著的研究成果,但是在实际的水下图像增强处理方面,仍存在许多不足与待改进之处。我国海洋领域广阔,淡水湖泊众多,不论是生态发展还是军事战略发展,都对水下资源有着重要需求。而目前的水下图像处理方法并不能满足复杂水域或者动态水域下的图像增强,当水体发生变化,或者浑浊度发生改变时,算法效果往往会大打折扣,为了能够解决多水域、多浑浊度、复杂情况下的水下图像增强问题,本文对水下成像物理模型进行了研究,基于浑浊度分级提出了一种有效解决复杂水域下的水下图像增强方法。光线在水下的传播过程决定了最终光学设备的成像质量,而水体介质极其复杂,导致光线的传播过程中会出现各种干扰。其中,最严重的两个影响因素是光的衰减与散射效应,他们是水下图像模糊、雾化、色偏、对比度低等问题的主要原因。本文对光的传播过程进行了研究与分析,对其衰减和散射过程进行了推导,建立了水下成像物理模型。为了能够适应我国水域种类繁多这种特性的水下图像处理,本文提出了基于浑浊度分级的水下图像增强处理,能够在满足不同浑浊度下的水下图像增强要求,同时能够在特殊水体下取得较好的增强效果。浑浊度分级基于水下图像直方图提出,根据不同图像的直方图特性,对水下图像的浑浊度、色偏程度、色偏通道进行检测,根据检测结果,对图像采用相应的增强路线并获得最终结果。本文最后采用实验对比的方法对本文方法的有效性进行了验证。基于本文的水下图像数据库,使用本文方法对水下图像进行了处理,将结果进行了横向和纵向的对比。对比分为主观与客观两部分。客观对比部分采用多种水下图像质量评价标准进行对比,并在最后进行了总结与展望。更多还原

【Abstract】 Image processing has achieved significant research results in the course of its development,but there are still many shortcomings and areas for improvement in the actual underwater image enhancement processing.There are huge marine area and numerous freshwater lakes in our country.Whether it is ecological development or military strategic development,there is an important demand for underwater resources.However,the current underwater image processing methods cannot meet the requirements of image enhancement in complex waters or dynamic waters.When the water changes or the turbidity changes,the algorithm effect is often greatly reduced.In order to solve the problem of multiple waters,multiple turbidity and other complex situations,this paper researches the physical model of underwater imaging,and proposes an effective underwater image enhancement method based on the turbidity classification which could solve problems in complex waters.The propagation process of light determines the underwater imaging quality.The propagation of light is interference by various factors as the water medium is extremely complex.Two most serious influencing factors are the attenuation of light and the scattering of light,which are the main reasons for the blurring,hazing,color shift,and low contrast of underwater images.After analyzing the propagation process of light,we derive its attenuation and scattering process,and establishes a physical model of underwater imaging.In order to be able to adapt to various waters of our country,this paper proposes an underwater image enhancement processing method based on turbidity classification,which can meet the requirements of different turbidity and special waters.The turbidity classification is proposed based on the histogram of underwater images,according to the characteristics of which the turbidity,degree of color shift,and color shift channel of underwater images are detected.The corresponding enhancement route is adopted for the image based on the detection result.At the end of this article,experiments are carried out to verify the effectiveness of the method.Underwater images of dataset are processed,and the results are compared horizontally and vertically.The comparison is divided into subjective and objective parts.In objective part,a variety of underwater image quality evaluation standards are used for comparison.Finally,we summarize our work.更多还原