Learning outdoor color classification from just one training image

Abstract

We present an algorithm for color classification with explicit illuminant estimation and compensation. A Gaussian classifier is trained with color samples from just one training image. Then, using a simple diagonal illumination model, the illuminants in a new scene that contains some of the same surface classes are estimated in a Maximum Likelihood framework using the Expectation Maximization algorithm. We also show how to impose priors on the illuminants, effectively computing a Maximum-A-Posteriori estimation. Experimental results show the excellent performances of our classification algorithm for outdoor images. © Springer-Verlag Berlin Heidelberg 2004.