Face hallucination with pose variation

Face hallucination is used to synthesize a high-resolution facial image from a low-resolution input. In this paper, we present a framework for face hallucination with pose variation. We derive a texture model consisting of a set of linear mappings between the Gabor wavelet features of the facial images of every two possible poses. Given a low-resolution facial image, its pose is first estimated using a SVM classifier. Then, the Gabor wavelet feature corresponding to the frontal face is computed by our texture model and the low-resolution frontal facial image is reconstructed from its Gabor wavelet feature by a novel algorithm we propose. Finally, the high-resolution face can be hallucinated by one of the hallucination approaches for frontal faces. Our framework is demonstrated by extensive experiments with high-quality hallucinated results.