A highly selective HT based algorithm for detecting extended, almost rectilinear shapes

Abstract: The paper presents an highly selective algorithm for detecting extended and almost rectilinear shapes in digital images, in presence of structured and unstructured noise; it exploits the Gradient-based Hough Transform, followed by a special purpose correlation process in the parameter space. The paper discusses the algorithm and its application in a quality inspection task for detecting fabrication defects in mechanical pieces.

 not available