Image clustering based on hermetian positive definite matrix and radial Jacobi moments

Abstract

The main purpose of this work is to present a new 3D image clustering technique based on the radial Jacobi moments to extract the descriptor vectors. We use a distance linked to a Hermitian definite positive matrix to minimize the objective function obtained. The results show that the positive definite matrix to minimize the objective function and significant improvement in terms of recognition accuracy and invariability.