Using neural network in color classification

The artificial neural network (ANN) has widely been used in the field of pattern classification. The main task of image segmentation is to extract interesting objects placed at different locations in images, so it is a sort of pattern classification problem. It can be treated as a maximum likelihood estimation problem in a color image when represented in a color histogram. In order to improve the flexibility of the classification result in a changed environment we propose the method of training the color pattern in a neural network using the EM algorithm which is a general method for the maximum likelihood problem. An experiment proved that it is applicable and significant.