摘要：图像边缘是图像的一个重要特征，它包含着有价值的目的界限信息，能够用于图像 分析、目标辨识以及边缘滤波。数字图像的边缘检测分滤波、增强边缘强度、检测、边缘 的精准定位。本设计主要分析数字图像处理中经典的图像边缘检测算子，如Sobel算子、 Roberts算子、Prewitt算子、Laplacian算子、Canny算子等，然后通过Matlab进行仿真，并 对仿真结果分析比较。Roberts算子定位精度比较高，但对噪声敏感，检测出的边缘较细。 Sobel算子产生的边缘效果较好，但是检测到的边缘比较粗。Prewitt算子对噪声有平滑作用， 检测出的边缘也比较粗，定位精度低。Log算子克服了拉普拉斯算子抗噪声能力差和容易 丢失边缘信息的的弊端。Canny算子具有较强的噪声克制能力，边缘的连续性较好，但是会导致丢失部分边缘。改进Sobel算子适用于检测含高斯噪声的图像的边缘。仿真结果的分 析比较为图像边缘检测的实际应用提供对照和参考。75428

毕业论文关键词：Matlab,数字图像处理,边缘检测,边缘检测算子

Abstract:Image edge is an important feature of the image, It contains valuable information about the purpose of the boundaries, it can be used in image analysis, target identification and edge filter。 Digital image edge detection is including the filter, enhancing edge strength, detection, positioning。 This design mainly analyzes the digital image processing in the classic image edge detection operator, such as Sobel operator, Roberts operator, Prewitt operator, Canny operator, etc。 Then we can simulate by the Matlab and analyze the results。 Positioning accuracy of Roberts operator is high, but it is sensitive to noise and edge that is detected by Roberts operator is fine。 Sobel operator has good edge effect and wide edge。 Prewitt operator can eliminate the noise of digital image, it’s edge is wide too and positioning accuracy low。 Log operator overcomes the shortcomings of Laplace operator and lost edge。 Canny operator has the ability of eliminating noise and its edge is continuous, but it leads to losing the part of edge。 Improved sobel operator is suitable for the digital Image with Gaussian noise。 The results can provide help for the practical application of image edge detection。

Keywords: Matlab，Digital image processing，Edge detection，Edge detection operator

目 录

1 绪论 4

1。1 选题背景及意义 4

1。2 发展现状及趋势 4

2 边缘检测的基本概述 4

2。1 边缘检测概述 4

2。2 边缘检测步骤 5

2。3 Matlab 简介 6

3 常用的图像边缘检测算子 8

3。1 微分算子法 9

3。2 高斯-拉普拉斯算子法 12

3。3 坎尼算子法 14

3。4 改进的 Sobel 边缘检测方法 16

4 仿真与分析 17