Radio astronomical data are increasingly corrupted by human telecommunication activities. Therefore, Radio Frequency Interferences (RFI) mitigation becomes an important step in the data processing flow, in particular for phased radio telescope array. In this framework, the preliminary step is to retrieve the RFI spatial information. This article presents three new techniques in radio astronomy allowing the estimation of the RFI spatial signature. These techniques are based on subspace decomposition of time-lagged correlation matrices or cyclic correlation matrices, and on a Blind Source Separation approach. Compared to classical methods, these approaches improve the quality of the spatial filtering obtained on the raw uncalibrated sky map.