Use of Dynamic Time Warping for Object Shape Classification through Signature

Abstract

This paper expresses an application of similarity matching of the signatures through DTW.
Fundamental aspect of classification is template matching. The classification is robust to
noise, scaling, and rotation. Feature includes radius plus angle along the boundary points with
respect to center of gravity. The classification automatically and confidently discloses the
shape of every object at once throughout page from top to bottom. The paper expresses its
promising results within an average of a few seconds (cheaper classification) for an object. A
series of tests is done with all possible configurations of geometrical shapes.