University of Western Australia

Department of Computer Science

Technical Report 90/14

EARLY JUMP-OUT CORNER

DETECTORS

J. Cooper, S. Venkatesh

and

L. Kitchen

[email protected]

Abstract: We present two new corner detectors, one of which works by using dissimilarity along the contour direction to detect curves in the image contour, and the other of which estimates image curvature along the contour direction. These operators are fast, robust to noise and almost selfthresholding. The standard deviation of the image noise must be specified, but this value is easily measured and the explicit modeling of image noise contributes to the robustnes of the operators to noise. We also present a new interpretation of the Kitchen-Rosenfeld corner operator in which we show that this operator can also be viewed as the second derivative of the image function along the edge direction.

Keywords: corner detection, similarity measures, thresholding, image noise modeling, sequential analysis

December 1990