Measuring conic properties and shape orientations

New methods for computing a shape s orientation and
several shape measures for elongation, linearity,
circularity, ellipticity, hyperbolicity, and
parabolicity of 2D point sets are outlined. These
measures are invariant to rotation, scaling, and
translation. They are calculated very quickly as
well. Among other things, we discover that the
definition of elongation highly correlates with the
definition of linearity. All of the shape measures
are tested on digital curves and compared with
existing methods. All of the methods work in real
time. The goal was to find a way of identifying basic
shapes in images. The motivation was to be able to
use these basic shape descriptors as features in a
computer vision system, where more complicated shapes
can be seen as collections of basic ones. This way, a
computer should one day be able to identify the
objects in even the most complicated scenes we deal
with in our daily lives.

Autorentext

The author completed his PhD at the University of Ottawa in thefall of 2008, in the field of Computer Science. His researchinterests lie in Computer Vision, Image Processing, and PatternRecognition. His Masters was done at Carleton University, in2005. Applications of his work are found in surveillance andsecurity, and medical imaging.

Klappentext

New methods for computing a shape's orientation andseveral shape measures for elongation, linearity,circularity, ellipticity, hyperbolicity, andparabolicity of 2D point sets are outlined. Thesemeasures are invariant to rotation, scaling, andtranslation. They are calculated very quickly aswell. Among other things, we discover that thedefinition of elongation highly correlates with thedefinition of linearity. All of the shape measuresare tested on digital curves and compared withexisting methods. All of the methods work in realtime. The goal was to find a way of identifying basicshapes in images. The motivation was to be able touse these basic shape descriptors as features in acomputer vision system, where more complicated shapescan be seen as collections of basic ones. This way, acomputer should one day be able to identify theobjects in even the most complicated scenes we dealwith in our daily lives.