How to obtain number of clusters in k-means clustering for images

I used the example in Colour-Based segmentation (<http://www.mathworks.com.au/help/images/examples/color-based-segmentation-using-k-means-clustering.html)>, and it works for the images that I am analyzing. However, I wonder if it is possible to derive computationally, based on the data in the image, the number of clusters (instead of me arbitrarily setting a number). Thanks!