A new method for estimating the thermal properties of composite materials is proposed. It uses a previously developed thermal characterization method that is based on Karhunen–Loève decomposition (KLD) techniques in association with infrared thermography experiments or any other kind of experimental device providing dense data in spatial coordinates. The novelty of this work lies in the introduction of two techniques based on two phase-wise defined test functions that extend the previously developed method to cases where the morphology of the composite material is not straightforward. Thanks to the orthogonal properties of KLD, only a few eigenelements are needed for an accurate estimation, which allows for a significant amplification of the signal/noise ratios. Furthermore, the proposed methods represent an attractive combination of parsimony and robustness to noise thanks to spatially uncorrelated noise being entirely reported on states. The effectiveness and accuracy of both techniques are proven with numerical tests.