Un Nuevo Conjunto de Descriptores de Forma basados en Simetría Bilateral y Radial

REVOLLO, NATALIA; CLAUDIO DELRIEUX; ROLANDO GONZALEZ-JOSÉ; MOLAS, LEONARDO

Congreso; 6ta Escuela y Workshop de Ciencias de las Imágenes; 2014

Computer vision, Axis symmetry. 11 1. ABSTRACT 12 Form and shape descriptors are among the most useful features for object identification and 13 recognition. Even though that there exists a widely used set of shape descriptors, and the 14 underlying computational methods for their evaluation, they are usually unable to distinguish 15 among very similar, previously segmented objects, which are notwithstanding very different to 16 the human eye. In this paper we propose a new set of shape descriptors, based on a finer 17 determination of the principal axes of the objects, and a more accurate estimation of their 18 bilateral and radial symmetry. These descriptors are thoroughly tested using several synthetic 19 objects with varying degrees of symmetry. The methods for axes estimation, and symmetry 20 descriptors extraction outperform the widespread shape descriptors in recognising and identifying 21 among very similar objects. Also, our proposal is more robust in recognition under scaling and 22 rotation.