Planar Shapes Descriptors Based on the Turning Angle Scalogram

In a previous paper, we have presented an original approach for 2-D shapes representation. Based on a multi-scale analysis of closed contours, this method deals with the scalogram of the differential turning angle. We then showed that this representation is rotation, translation and scale change invariant and that it is also shearing and noise resistant. In this paper, we propose some features extracted from this scalogram. This enables us to evaluate the turning angle scalogram representation of planar objects in the context of pattern recognition. When applied to shape retrieval from a database and for various transformations (deformations), experimental results confirm the efficiency of this new description approach.

[1]  Farzin Mokhtarian,et al.  A Theory of Multiscale, Curvature-Based Shape Representation for Planar Curves , 1992, IEEE Trans. Pattern Anal. Mach. Intell..

[2]  Kidiyo Kpalma,et al.  Contours smoothing for non-occluded planar shapes description , 2005, IEEE International Conference on Image Processing 2005.

[3]  Kidiyo Kpalma,et al.  A Novel Multi-scale Representation for 2-D Shapes , 2007, ICIAR.

[4]  Kidiyo Kpalma,et al.  Multiscale contour description for pattern recognition , 2006, Pattern Recognit. Lett..

[5]  Mohamed S. Kamel,et al.  Image Analysis and Recognition , 2014, Lecture Notes in Computer Science.

[6]  Tony Lindeberg,et al.  Scale-Space Theory in Computer Vision , 1993, Lecture Notes in Computer Science.