Face hallucination based on independent component analysis

In this paper a novel independent component analysis (ICA) based face super-resolution algorithm is proposed. In this scheme, the independent components (ICs) are obtained by offline training high-resolution face images. Given a low-resolution image, the high-resolution image is reconstructed by the linear combination of the ICs where the weight coefficients are obtained by the method of maximum a posteriori (MAP). The prior of ICA coefficients are estimated by performing PCA on training images. Meanwhile, a structure-tensor based filter is proposed to refine the intermediate SR result which is estimated by inverse ICA. Experimental results demonstrate that the proposed algorithm is robust to various pose, expressions and lighting conditions. The hallucination results preserve both the global structure and the high spatial-frequency information better such as sharp edges and high contrast.