Medoidshift
所属分类：matlab编程
开发工具：matlab
文件大小：36KB
下载次数：362
上传日期：2008-01-21 15:09:00
上 传 者：
walkking
说明： 中心点漂移是一种非监督聚类算法（与k-means算法相似，但应用范围更广些),可用于图像分割，基于Matlab实现的源码。
MedoidShift is a unsupervised clustering algorithm(similar to k-means algorithm, but can be used in border application fields), can be used for image segmentation. Included is the Matlab implementation source code.
(Center drift is a non-supervised clustering algorithm (k-means algorithm with the similar, but more a wider range of applications), can be used for image segmentation, based on the realization of the source Matlab. MedoidShift is a unsupervised clustering algorithm (similar to k-means algorithm, but can be used in border application fields), can be used for image segmentation. Included is the Matlab implementation source code.)
文件列表:
classify.m (420, 2007-08-19)
classify_slow.m (420, 2007-08-19)
dijkstra.dll (57344, 2007-04-07)
dijkstra.m (1822, 2007-04-07)
Example_TwoBivariateGaussians.m (879, 2007-08-19)
Example_TwoSpirals.m (993, 2007-08-19)
Figure1_MeanshiftVsMedoidshift.m (1366, 2007-08-19)
Figure7a_FiveCrescents.m (1006, 2007-08-19)
Figure8_FourSpirals.m (942, 2007-08-19)
IsomapIID.m (11332, 2007-04-08)
medoidshift.m (1152, 2007-08-19)
medoidshiftIterative.m (404, 2007-08-19)
randcrescent.m (429, 2007-08-19)
visualizeClustering.m (1078, 2007-08-19)
These files implement the medoidshift algorithm described in:
Yaser Ajmal Sheikh, Erum Arif Khan, Takeo Kanade, "Mode-seeking via Medoidshifts", IEEE International Conference on Computer Vision, 2007.
For further information contact: yaser@cs.cmu.edu
They have been tested on MATLAB Version 7.0.0.19920 (R14).
**
Examples are included:
Example_TwoBivariateGaussians.m: Clustering two bivariate Gaussian distributions
Example_TwoSpirals.m: Clustering using the ISOMAP distance matrix on 3D data
**
Code producing figures in the paper:
Figure1_MeanshiftVsMedoidshift.m
Figure7a_FiveCrescents.m
Figure8_FourSpirals.m
**
The basic code:
medoidshift.m: Code that takes in a distance matrix and a bandwidth parameter and returns the modes for each index
medoidshiftIterative.m: Implements one iteration of medoidshift
IsomapIID.m: Modified ISOMAP code from http://isomap.stanford.edu
dijkstra.m: For IsomapIID.m from http://isomap.stanford.edu
classify.m: Tree traversal algorithm (Step 2 in paper)
classify_slow.m: Slower version
近期下载者：
相关文件：
收藏者：