数字图像边缘检测算法的研究

【作者】 杨洋；

【导师】 杨先凤； 陈汶滨；

【作者基本信息】 西南石油大学 ， 计算机软件与理论， 2011， 硕士

【摘要】 随着电子计算机和数字图像处理技术的快速发展,图像边缘检测已成为图像处理的重要内容,它是图像分析的基础,也是图像分割、特征提取和图像识别的前提。本文对传统的图像边缘检测算法作了深入的分析与研究,在具体分析各类传统的边缘检测算法的基础上,重点研究了柔性形态学边缘检测算法和数学形态学边缘检测方法,提出了基于自适应柔性形态学图像边缘检测算法和基于多角度结构元素图像边缘检测算法。最后,用Matlab实现该算法,仿真实验结果表明,本文提出的基于自适应柔性形态学图像边缘检测算法和基于多角度结构元素图像边缘检测算法可以较好的检测出图像边缘特征的同时,并且也能较好的保持其图像的连通性,具有较好的鲁棒性。基于自适应柔性形态学图像边缘检测算法是在柔性形态学基础上提出的一种对图像边缘有效的边缘检测方法。与传统的柔性形态学边缘检测区别在于通过比较移动结构元素窗口的信噪比和整个图像的信噪比,从而实现自适应的选择柔性形态学K值的方法,因此就可以根据图像本身的性质来进行图像边缘检测。基于多角度结构元素图像边缘检测算法是在研究传统数学形态学边缘方法的基础上提出来的。因为传统的数学形态学边缘检测方法是用单一的对称的结构元素,这种结构元素方向相同,很难检测图像复杂的边缘。本文提出的基于多角度结构元素图像边缘检测算法使用的是从八个不同角度的结构元素分别进行形态学操作来进行图像的边缘检测,这样就可以很好的检测复杂的图像边缘。更多还原

【Abstract】 As the computer and digital image processing technology’s rapid development, the image edge detection has become an important part of image processing that is the basis of image analysis and is also the premise of image segmentation, feature extraction and image recognition. In this paper, the traditional algorithms of image edge detection for the in-depth analysis and research,is the base of the detailed analysis of various traditional edge detection algorithm,mainly studies algorithm of soft morphology edge detection and method of mathematical morphology edge detection, and two novel algorithms are proposed that is called adaptive soft morphological edge detection algorithm and morphological edge detector by multidirectional structuring element. Finally, Matlab simulation algorithm, experimental results show that this paper is propose two algorithms are better than the traditional edge detection in edge detection and noise suppression.Adaptive soft morphological edge detection algorithm is proposed based on the flexible morphological an effective edge of the image edge detection method, which has different between the traditional distinction and soft morphological edge detection is Comparing the SNR of image and SNR of moving structuring window,which can achieve the characteristics of adaptive algorithm. Therefore, according to the nature of the image itself it can make the image edge detection.Morphological edge detector by multidirectional structuring element is proposed based on the study of traditional methods of mathematical morphology. The standard morphological edge detector uses single and symmetric structuring element, so it is hard to detect the complex edge of images. This paper presents a morphologic edge detection methods based on multidirectional structuring element. Eight orientations structuring element was chosen for detecting complex edges which can make good detect the complex edge of images.更多还原