|
|
基于云模型和FCM聚类的遥感图像分割方法
Remote Sensing Image Segmentation Based on Cloud Model and FCM
查看参考文献7篇
|
文摘
|
模糊C均值算法由于具有良好的聚类性能而被广泛应用于图像分割领域，但聚类中心的初始化问题一直影响着该算法的运行效率。好的初始聚类中心，可以使算法很快收敛于最优解，而不合适的初始聚类中心，不仅需要更多的迭代次数，而且还可能使算法最终收敛于局部最优解。文章结合云模型和FCM（模糊c均值）聚类算法，提出了一种遥感图像分割的新方法。利用云变换解决模糊C均值聚类算法的初始化中心选择问题，可以根据样本特性自动确定聚类中心值及个数，并以较少的迭代次数收敛到全局最优解，提高了模糊C均值遥感图像分割方法的效率，具有较好的稳定性和鲁棒性。文章选取三幅TM遥感图像作为样本，分别利用云模型的FCM方法和传统的FCM方法对样本进行分割实验，实验表明采用云模型的FCM方法不仅能够取得较好的分割效果，而且大大减少了使算法收敛的迭代次数，提高了分割的效率。 |
|
其他语种文摘
|
FCM （Fuzzy C-Means） algorithm has good clustering efficiency, therefore it is widely used in the field of image segmentation. But it has the problem of clustering center initialization. Good initial clustering centers will constrain the value function to the overall situation optimal solution rapidly. However, inappropriate initial clustering centers, not only need more iterative times, but also may possibly caused the algorithm finally restrained to the partial optimal solution. Aim to solve the problem of clustering center initialization, the paper proposes a new approach of FCM based on cloud model which is a conversion model with uncertainty between a quality concept ex- pressed by natural language and its quantity number expression. This method in this paper combines cloud model with FCM clustering algorithm, and is applicable in the field of remote sensing image segmentation. It uses the method of cloud transformation to determine the value and number of clustering centers automatically according to the characteristics of the samples, and with less iteration times to reach the optimal solution of the overall situation. The method improves the efficiency of traditional FCM image segmentation of remote sensing image, with excellent stability and intelligence. The paper also chose three remote sensing images of TM to carry out image segmentation experiments using the methods of cloud model based FCM, and compared it with the traditional FCM method. Experimental results prove that by using our method we can reduce the iterative times and enhance the efficiency of Fuzzy C-Mean clustering algorithm. |
|
来源
|
地球信息科学
,2008,10(3):302-307 【扩展库】
|
|
关键词
|
模糊C均值
;
聚类
;
图像分割
;
云模型
|
|
地址
|
武汉大学遥感信息工程学院, 武汉, 430079
|
|
语种
|
中文 |
|
文献类型
|
研究性论文 |
|
ISSN
|
1560-8999 |
|
学科
|
自动化技术、计算机技术;航空 |
|
基金
|
国家973计划
;
中国科学院资源与环境信息系统国家重点实验室开放研究基金
|
|
文献收藏号
|
CSCD:3291327
|
参考文献 共
7
共1页
|
|
1.
Pham DL. An adaptive fuzzy C-means algorithm for image segmentation in the presence of intensity inhomogeneities.
Pattern Recognition Letters,1999(20):57-68
|
CSCD被引
1
次
|
|
|
|
|
2.
Ma L. A modified fuzzy C-means image segmentation algorithm for use with uneven illumination patterns.
Pattern Recognition,2007(40):3005-3011
|
CSCD被引
19
次
|
|
|
|
|
3.
Yager RR. Approximate clustering via the mountain method.
IEEE Transactions on Systems,1994,24(8):1274-1284
|
CSCD被引
1
次
|
|
|
|
|
4.
Chiu SL. Fuzzy model identification based on cluster estimation.
Journal of Intelligent and Fuzzy Systems,1994(2):267-278
|
CSCD被引
141
次
|
|
|
|
|
5.
裴继红. 聚类中心的初始化方法.
电子科学学刊,1999,21(3):320-325
|
CSCD被引
20
次
|
|
|
|
|
6.
李德毅.
不确定性人工智能,2005
|
CSCD被引
293
次
|
|
|
|
|
7.
骆剑承. 遥感影像生理认知概念模型和方法体系.
遥感技术与应用,2001,16(2):103-108
|
CSCD被引
4
次
|
|
|
|
|
|
了解气象卫星更多信息，请访问