On the Existence of Complete Invariant Feature Spaces in Pattern Recognition

This paper presents a general and model independent analysis of the problem of feature extraction in pattern recognition. We will derive two criteria which ensure the existence of a complete feature space. This is a space which contains exactly the information relevant for the classiication process following feature extraction. We discuss several possibilities for the construction of such a complete feature space and present experimental results which indicate the potential of the proposed methods for practical applications.