A novel algorithm for image representation using discrete spectrum of the Schrödinger operator

This paper extends the recent signal analysis method based on the spectral analysis of the semi-classical Schrödinger operator to two dimensions. An algorithm based on the tensor product approach when writing the eigenfunctions of the semi-classical Schrödinger operator is proposed. The algorithm is described and the effect of some parameters on the convergence of this method are numerically studied. The performance of the algorithm is illustrated through some examples.