MATLAB: K-means Clustering

image processingk means clustering

I am doing color based segmentation using kmeans clustering.I am using inbuilt function of matlab(kmeans).My input image has object and background where I need to segment the object. so I am using cluster value as 2.But while doing this, for few images, the background has a cluster index of 2 and object as 1 while in some other images, background has cluster index 1 while object 2. How does K-means clustering decide the index for clusters? If there are to be two clusters, will the background be labeled with number 2 and object with 1 or vice-versa?

Best Answer

kmeans chooses seeds by a random process if you do not specify initial locations using the 'start' option. Information on how the algorithm works is in the 'More about' section of the documentation

doc kmeans

There is no way for the algorithm to know what you call 'object' and what you call 'background'. If you give it two start seeds, one in the object and one in the background then these will always remain in this order, but otherwise they can change around with each run of the algorithm unless you fix the random seed, but that would only keep the class centres the same for the same data, not across different data sets obviously.

Related Solutions

MATLAB: How to segment the hand part in the image using K-means clustering

See my attached kmeans demos. Adapt as needed. You'd be wise to do background masking first though, to set the unchanging background to black.

MATLAB: How to find out the content of one cluster

Not really, no. Clusters, especially as determined by k-means, do not have "features". The k-means documentation indicates,

This iterative partitioning minimizes the sum, over all clusters, of the within-cluster sums of point-to-cluster-centroid distances. Rows of X correspond to points, columns correspond to variables.

If you were working with two variables (two columns) then a geometric interpretation would be that any given point P is considered to be associated with a centroid C if the circle of radius |PQ| around P does not (strictly) include any other centroid.

Likewise you can extend to three variables (three columns) and extend the geometric interpretation to be "sphere" instead of "circle".

You can see, then, that the cluster associated with C does not have any particular value or even value range for the variables. The boundary between clusters would be the intersection of hyperspheres around the centroids.

If you were using SVM, and you had only two clusters, then you could find a hyperplane that divided the two clusters -- but a hyperplane is not what one would typically think of as being a "feature".

Best Answer

Related Solutions

MATLAB: How to segment the hand part in the image using K-means clustering

MATLAB: How to find out the content of one cluster

Related Question