基于数学形态学的边缘检测研究-检测技术与自动化装置专业论文.docx

- - I - 摘 要 图像的边缘是图像的基本特征之一，它包含了图像的位置、轮廓等重要信息， 因此对图像的边缘进行检测是图像处理和分析中重要的过程。由于图像测量、图像 压缩、图像分割、以及图像识别都是以图像检测为基础的，因此对图像进行边缘检 测有极其重要的意义。一些经典的边缘检测方法大都是基于微分运算，其中包 含 Sobel 算子、Prewitt 算子、Robert 算子、LOG 算子、Canny 算子等。目前也有一些 新的算法，例如：统计判决法、边缘连接法、小波边缘检测法、基于模糊学的边缘 检测等。但这些方法都其局限性及适用范围。针对上述方法的缺点，本文提出了一 种多尺度、多结构的结构元素形态学的边缘检测方法。 本文首先介绍了图像边缘、边缘检测的定义，以及边缘检测的意义。然后进一 步介绍了各种经典的边缘检测算法，以及其他新的边缘检测方法。并且通过仿真实 验讨论了其中部分算法的优缺点。接着介绍了数学形态学中二值形态学与灰度形态 学的基本知识与性质以及数学形态学在图像处理中的应用，介绍了现有的基于形态 学边缘检测的方法。最后在现有的算法基础上提出一种新的边缘检测算法。该方法 是在传统的数学形态学边缘检测算子上加以改进，能有效的提取图像的边缘，对椒 盐噪声和高斯噪声都有很好的抑制作用，而且简单实用。该方法是一种多尺度和多 结构的结构元素结合的形态学边缘检测法，在边缘检测问题上取得了较好的效果。 关键词：图像处理；边缘检测；数学形态学；多尺度；多结构 - - II - Abstract Edge is one of basic traits of image, it contains some important information, for example place of image, figure and so on. Thence, edge detection is an important process of image processing and analyze. Edge detection is the foundation technology of image processing that contains image measuring, image segmentation, image compressing, and image recognition, so it is very important. There are many classic methods about edge detection that based on differential coefficient, for example, Sobel operator based grads, Prewitt operator, Robert operator, Laplace operator, LOG operator, Canny operator and so on. Now there are some new methods, such as statistical, edge connecting, edge detection based on micro-wavelet. However, these methods only use in some limitation area. Through analyze merits and drawbacks of these methods，and then create a new edge detection which based on mathematical morphology. Firstly，present definition of digital image，image detection and its functions and significance. Secondly, present some classic and new edge detection methods, and then summarize their merits and drawbacks by experiments. In succession, summarize basic knowledge and properties of binary morphology and gray morphology, and clarify purposes of these morphology operators in image processing. Finally, create a new edg