A FIELD-THEORY OF PATTERN IDENTIFICATION USING THE CONCEPT OF GAUGE-FIELDS

A held theory for geometrical pattern identification is developed base d on the postulate that various modified patterns are identified via i nvariant characteristics of pattern transformations. The invariant cha racteristics of geometrical patterns are written as the functional of the light intensity distribution of pattern, its spatial gradient, and also its spatial curvature. Some definite expressions of the invarian t characteristic functional for two dimensional linear transformation are derived, and their invariant and Feature extracting property are e xamined numerically. It is also shown that the invariant property is c onserved even when patterns are deformed locally by introducing a ''ga uge field'' as new degree of freedom in the functional in form of a co variant derivative. Based on this idea. we discuss a field theoretical model for pattern identification performed in biological systems.