Automated feature detection using evolutionary learning processes

The design and implementation of a software environment for the study of learning strategies applied to task in image processing are described. This environment facilitates systematic exploration of evolutionary learning processes and embedded adaptive control mechanisms that modify the process of feature extraction. Selected principles of mathematical morphology define the nucleus of this image processing and learning system. A preliminary experiment with a two-class recognition system is described. Initial observations are discussed for a recognition task that requires the classification of English capital letters into two categories: target or nontarget