Two-Dimensional Textural Edge Detection Using Ω-Gaussian Filter

Abstract

A textural edge detection frequency domain filter is proposed to locate the boundary between two textural domains modulated by different spatial frequencies. This filter is derived from a zero-order prolate spheroidal function which can be approximated by a _-Gaussian shape. The extension to a 2-D rotational symmetric filter capable of locating textural edges independently of the relative orientation of the textural frequencies is demonstrated. Results show that the proposed filter is more effective at concentrating the output energy around the edge than the filter proposed by Shanmugan et al., which was derived with an additional constraint of evenness in the design of the filter function.