Textural edge extraction with an optical heterodyne scanning system

Abstract

In many classes of images, some of the important features can be recognized by repetitive structures over a large region. Such a spatial feature provides useful information in image segmentation and classification. Various digital image processing schemes to extract textural information have been reported based on local operators manipulating image pixels, often to extract local statistics from images [1]. Other schemes exploiting the textural information in the frequency domain have also been investigated along with some successful demonstrations [2, 3]. In this paper, textural edge detection by bandpass filtering is considered. A simple technique for realizing the bandpass filter in a heterodyne optical scanning system is described and illustrated experimentally. Section II discusses textural edge detection by bandpass filtering in the context of textural image modeling and frequency domain filter function construction. Section III describes the optical scanning system capable of extracting the edge in the texture and presents some experimental results.