I have an image data set and I want to cluster them to K clusters useing k-means clustering but I didn't know how to start.
I know that matlab has a function that do this work but for simple data, how does it work for images?
You are right. Matlab provides a kmeans function that does what you want to have.
As far as I know, inputs should be vectorized, which can be done by, e.g.
1. vectorize the image (stacking of column of row vectors). This approach induces a high-dimensional feature space with lost of redundancy in it. Alternatively, you could
2. get a low-dimensional vector representation of your images. For instance via the bag of visual words idea, which is as well based on a (e.g. kmeans) clustering step.
You are right. Matlab provides a kmeans function that does what you want to have.
As far as I know, inputs should be vectorized, which can be done by, e.g.
1. vectorize the image (stacking of column of row vectors). This approach induces a high-dimensional feature space with lost of redundancy in it. Alternatively, you could
2. get a low-dimensional vector representation of your images. For instance via the bag of visual words idea, which is as well based on a (e.g. kmeans) clustering step.
Hi, just use the Matlab example:
http://www.mathworks.com/help/images/examples/color-based-segmentation-using-k-means-clustering.html?nocookie=true
:-)
Take a look at vl_feat toolbox at
http://www.vlfeat.org/
Or for your need
http://www.vlfeat.org/api/kmeans.html
Thank you all for your answers they was very usefull for me
thanks again
A good example to start with
http://www.mathworks.com/help/stats/kmeans.html
Matlab segmentation program with source code
http://www.codeforge.com/article/211076
- Badges
- Science topic