分数阶微积分在图像处理中的研究-计算机技术专业论文.docx

华北电力大学硕士学位论文 华北电力大学硕士学位论文 摘 要 在信息社会中，随着人们生活水平的不断提高，人们常常需要通过图像来获 取信息，并且对图像高分辨率、多细节的需求是非常迫切的。但是我们一般并不 是很容易能够得到有丰富的纹理与微弱细节对比明显的图像，于是我们非常需要 对图像细节进行增强来提高图像的分辨率质量与微弱特征。本文将进一步探讨和 研究图像增强的理论和方法。 图像增强是数字图像处理中最简单且最具吸引力的领域。近些年来，研究人 员已把分数阶微积分理论从纯粹的数学理论研究领域引入到了数字图像处理的 应用中，从而使分数阶微积分的研究取得了突破性的进展，这在世界上都是一个 研究很少的新兴领域。在图像处理中，对比于以前的基于整数阶微分的处理图像 的方法，分数阶微分拥有显著的优势。 本文首先对已有的用分数阶微分增强图像的方法进行了研究。由于分数阶微 分掩模处理图像存在方向性，将分数阶微分与泰勒公式和方向导数相结合，从而 充分利用了所处理像素的邻近像素点的信息。论文中对此改进的分数阶微分的方 法进行了分析，并重点推导了分数阶微分掩模的构造与变换过程，并将其用于图 像增强。 最后为了验证以上改进的图像增强方法的效果，在 Matlab 平台上进行仿真 实验，并计算出从实验得到的图像的质量参数，然后和现有的典型图像增强方法 的处理结果进行了比较，从结果可以得出，本文所提出改进的图像增强方法能有 效地提高灰度图像的纹理特征。 关键词：图像增强；分数阶微分；泰勒公式；方向导数 I Abstract In the information society, with the continuous improvement of living standards, people often need to access information through the image, and the demand of image with high resolution and rich detail is urgent. But it is difficult for people to get rich texture and apparent contrast about weak details of image in general. Then, we need to enhance the image details to improve the resolution quality and the weak characteristics of the image. This thesis will further explore and study the theory and methods of image enhancement. Image enhancement is the most simple and the most attractive areas in the digital image processing. In recent years, researchers have already introduced the theory of fractional calculus mathematics from purely theoretical research into the application of digital image processing, so that the study of fractional calculus has made breakthrough progress. It is emerging field with little research in the World until now. Compared with traditional image processing approaches based on integer order differential, the application of fractional differential has remarkable advantages in image processing. This paper firstly studied the existing methods. Because of the directional characteristic of different fractional differential mask processing image, we combined fractional differential with Taylor formula and