Symmetry Detection via Particle Normal Form

Abstract

Image normalization is a common method in pattern recognition. Seeing corners of plane graph as graph features is a way to effectively reduce the computational complexity of the image processing. This paper presents an image normalization method according to geometrical features of plane graph. The new method is called PNF (Particle Normal Form) which is applicable to any corner-pointed plane graphs. The PNF coordinate sequences of symmetrical graphs have strong regularities. The ability of achieving symmetry detection via PNF is mainly analyzed. Experiments show that PNF has broad application potential in symmetry detection.