摘要：图像分割是图像处理与模式识别中一项重要而基础的技术手段，其目的是把图像分割成一些有意义的或者应用中感兴趣的区域，这些区域与现实中的各类目标相对应形成模型，便于以后在实际应用中对比与检测。课题讨论了bmp彩色图像的结构，并设计bmp图像类；在针对bmp图像的分割处理中运用了K-均值聚类分割算法和Hough变换算法；同时研究了图像分割的原理与算法进行了分析，Hough是图像处理中识别几何形状的基本方法之一，K-均值聚类法可以将一幅图像分成k个区域。设计采用C#语言编写运行程序，程序首先将图像像素导入，分别获取各类像素序列，然后经算法的计算，最后将处理后的新像素重组。在实验运行中，采用K-均值聚类算法和Hough算法处理的图像结果达到了预期的效果。
关键词： 图像分割，K-均值聚类法，彩色图像，欧氏距离，Hough变换,4413
The design process of color image segmentation
Abstract: Image segmentation, purposed on dividing image into some meaningful or interesting areas, is an important and basic technology in image processing and pattern recognition . Corresponding to these areas and targets in reality, models can be formatted and used in contrasting and testing in the practical application later. This subject is to discuss the structure of BMP color images, and design the BMP image classes. In BMP image segmentation, k-means clustering segmentation algorithm and the Hough transform algorithm are used; Meanwhile the principle and algorithm of image segmentation are analyzed, Hough is one of the basic methods in the geometry recognition of the image processing , K - average clustering method can divide an image into K areas. C # language is used to run the program , which will firstly imports the image pixels, then achieves different types of pixel sequence. With the algorithm of calculation, new pixel reorganization finally after processed. In the experiment, the k-means clustering algorithm and Hough algorithm are adopted and reached the expected effect.
Keywords: Segmentation, K- means clustering method, color image, Euclidean distance, Hough transform
目录
摘要 i
Abstract i
目录 ii
1 数字图像处理概述 1
1.1 图像技术概述 1
1.1.1 图像技术 1
1.1.2 彩色图像 3
1.2 C#图像处理基础 3
1.2.1 Bmp图像结构 3
1.2.2 C#语言概述 7
2 彩色图像分割处理原理及算法 9
2.1 图像分割概述 9
2.2 图像的分割原理 11
2.2.1 灰度图像的K-均值聚类法原理 11
2.2.2 彩色图像的K-均值聚类法原理 12
2.2.3 Hough变换原理 13
3 彩色图像分割处理的程序设计 14
3.1 K-均值聚类 14
3.2 HOUGH变换 16
4 彩色图像的分割处理 18
4.1 图像分割评价标准 18
4.2 彩色图像的分割程序处理 19
4.2.1 彩色图像K-均值聚类分割 20
4.2.2 Hough变换 24
5 总结 29
5.1 小结 29 彩色图像的分割处理程序设计+Hough算法:http://www.751com.cn/jisuanji/lunwen_1067.html