There are two answers to your question:
The simple answer is that from the Mahout point of view, the classification of images is no different from the classification of any other data type. You can find the right set of features to describe your data, and then: train, validate, test, and deploy.
The second answer is a bit more complicated, and I'm going to summarize. In the case of images, the step at which you calculate a suitable set of functions covers the entire field of research (called computer vision). There are many methods: DHOG, gradient direction, SURF, SIFT, etc. Depending on the images and your expectations, you can only get reasonable results using the existing method or maybe not. It would be impossible to say without looking at your images, and you tell us about your goals.