find the infected cluster

i segmented a diseased plant image using K means... i got 4 clusters... now i want to find the infected cluster... is it possible thru code.. in different images the infected cluster is different... so how to find the infected cluster??... please do reply....

Answer 1

[Walter Roberson]

One of these things is not like the others.
One of these things just doesn't belong.
One of these things is not like the others.
Can you guess which one before I end my song?

We are not going to be able to assist you much until you can describe what distinguishes an infected cluster from the other clusters.

[Image Analyst]

If help is needed in determining that, an image would help, http://www.mathworks.com/matlabcentral/answers/7924-where-can-i-upload-images-and-files-for-use-on-matlab-answers, though no guarantees since I'm not a botanist.

[Elysi Cochin]

sir i couldnt upload image here sir.... so i'm showing an image from google that is the output of kmeans clustering algorithm.....

in that the cluster 4 is the infected cluster.... but in some images it might be cluster 2& 3 or only cluster 1 and so....

sir how to detect the infected cluster thru coding...

is it possible to detect thru coding?? please do reply sir.... the link of the image is below....

http://www.google.co.in/imgres?q=output+of+K-Means+clustering+for+a+leaf+that+is+infected+with+early+scorch+disease.&um=1&hl=en&tbo=d&biw=1366&bih=594&tbm=isch&tbnid=D7mSbFh4YJmkNM:&imgrefurl=http://www.scialert.net/fulltext/%3Fdoi%3Ditj.2011.267.275%26org%3D11&docid=iNwDJePOvnnXTM&imgurl=http://docsdrive.com/images/ansinet/itj/2011/fig2-2k11-267-275.gif&w=335&h=337&ei=B0W0UIniBI2KrgfGuYCgCA&zoom=1&iact=hc&vpx=123&vpy=138&dur=567&hovh=225&hovw=224&tx=120&ty=154&sig=116683586742086246913&page=1&tbnh=143&tbnw=142&start=0&ndsp=20&ved=1t:429,r:1,s:0,i:85

[Walter Roberson]

http://www.mathworks.com/matlabcentral/answers/7924-where-can-i-upload-images-and-files-for-use-on-matlab-answers

[Walter Roberson]

That paper indicates that Euclidean distances are used for the clustering, but does not describe what objects are whose Euclidean distance is being measured. Without that information, it is not possible to proceed.

It is not obvious from that paper that it is possible to automatically distinguish which cluster would be infected. There is also no justification given there for choosing 4 as the number of clusters.