Local ordered grey levels as an aid to corner detection

Abstract

This paper presents a novel method for detecting corner points in a scene. It is based on features extracted from the distribution of ordered greylevel values within a local window. The algorithm is effectively orientation independent. Results are presented for an image containing corners of different angles and orientations. A comparison is made between the positions obtained for the corners detected using this method and those obtained by direct measurement. The signal strengths of similar corners at different orientations are also compared. The performance of the algorithms with a noisy image is also discussed. Its domain of application and usefulness is considered. © 1984.