Abstract: Porosity is one of the important factors to evaluate the related properties of porous materials and an important index for material quality inspection. To solve the problem that the porosity value is greatly affected by the resolution, under the condition that the spectrum and material information are unknown, this paper, based on the multi-spectrum CT and NC-POCS(non-convex projection onto convex sets) algorithm and combined with the non-negative matrix factorization method, realizes decomposition image reconstruction. Taking the porous material aluminum foam as an example, we analyze the porosity change of the porous material under different resolutions. The experimental results show that the proposed algorithm can suppress the hardening artifacts, and the measured porosity is less affected by the resolution.