《图像边缘检测方法的研究》-毕业论文（设计）.doc

PAGE 摘 要 数字图像边缘检测是图像分割、目标区域识别和区域形状提取等图像分析领域十分重要的基础，是图像识别中提取图像特征的一个重要方法。 本文对边缘检测理论和算法作了深入的研究，具体分析了各类传统的边缘检测算法如：Roberts，Sobel，Prewitt，Kirsch，Laplace等，他们基本上都是对原始图像中像素的小邻域构造边缘检测算子，进行一阶微分或二阶微分运算，求得梯度最大值或二阶导数的过零点，最后选取适当的阈值提取边界。但这些算法均存在对噪声敏感、不能自适应选择阈值、检测效果不太理想等缺点。针对这些问题，本文引出了基于小波变换的边缘检测算法。小波变换具有良好的时频局部化特性及多分辨率特性，能够有效地分析信号的奇异性。最后用MATLAB实现该算法，实验结果表明，小波边缘检测比传统的边缘检测算法有更好的边缘检测效果。 关键词： 图像处理，边缘检测，小波变换，自适应阈值 ABSTRACT Edge detecting of the digital image is the base of the image analysis area,which includes the image segmentation,the target recognition and the region shape extracting and so on, and it also is an important method for image recognition to extract the image’s characteristics. This paper,I deeply research the theory and the algorithm of the edge detecting and analyze ,the traditional algorithms of the edge detecting,for example,Roberts,Sobel,Prewitt, Kirsch and Laplaee.All of them construct an edge detection operator for the pixels of the primitive image,do first order or second order differential,find the max gradient and the zero of the second derivative,and at last ,choose the proper value to extract the border value.But all of them have some disadvantages,such as,be sensitive for the noise,not choose the proper threshold by itself,and the detecting result is not ideal.Because of these problems,this paper introduces an edge detecting algorithm based on wavelet transform.The time-frequency localization of wavelet transform and the muliresolution wavelet transform can be used to analyze the singularity of the signal.In matlab to realize it,from the result,the edge detection on wavelet transform is better than the traditional algorithm. KEY WORDS:image processing,edge detecting,Wavelet transform,self-adaptive threhold 目 录 TOC \o 1-3 \h \z \u HYPERLINK \l _Toc296323096 第一章 绪论 PAGEREF _Toc296323096 \h 1 HYPERLINK \l _Toc296323097 1.1课题研究的目的和意义 PAGEREF _Toc296323097 \h 1 HYPERLINK \l _Toc296323098 1.2图像边缘检测的发展与现状 PAGEREF _Toc296323098 \h 2 H