Dipole and Quadrupole Moments in Image Processing

This paper proposes an algorithm for image processing, obtained by adapting to image maps the definitions of two well-known physical quantities. These quantities are the dipole and quadrupole moments of a charge distribution. We will see how it is possible to define dipole and quadrupole moments for the gray-tone maps and apply them in the development of algorithms for edge detection. 1. Introduction A well-known concept used in condensed matter physics and optics, that of coherence length, has been recently adapted for texture analysis of images. According to this physical concept, we developed an algorithm able to compare local and global behaviors of image maps [1-4]. By means of this algorithm we characterized irregular textures, where Fourier or wavelets approaches are less useful and statistical evaluations are preferred [5,6]. In this paper again, we propose a new approach for image processing, developed by adaptation of two other physical quantities 1. We will show how the dipole and quadrupole moments of charge distribution turn out to be very interesting quantities for image pattern recognition. After defining the dipole and quadrupole moments for image maps, we will use them in the edge detection.