Source localization problem consists of an ensemble of techniques that are used to obtain spatial information of present radiation in given medium of propagation, with a constraint of the antenna geometry and the characteristics of radiating sources. This condition gives multitude of cases to study, hence several methods were proposed in the literature. In this paper, a new algorithm for estimating the Direction of Arrival (DoA) of narrowband and far field punctual sources is introduced. By exploiting the spectrum of covariance matrix of received data, the Lorentzian function on spectral matrix to filter the eigenvalues is applied. This filtering process eliminates the eigenvalues belonging to signal subspace. Parameters of Lorentz function are adjusted using first and second statistics of eigenvalues. The algorithm requires the knowledge of minimum eigenvalue and is performing when the dimension of antenna is relatively large which is confirmed by several Monte Carlo simulations.