A computational approach for the emulation of visual receptive fields

摘要

We introduce a generalized theory of discriminant functions as a mathematical framework for the development of engineered receptive fields. These functions replicate mathematically some of the functional properties of the biological receptive fields, and are implemented into computer algorithms for the enhancement of transition information in signals and images. Two methods for the generation of discriminant functions are presented which can be successfully employed for the design of artificial receptive fields.