Hallucinating multiple occluded face images of different resolutions

Learning-based super-resolution has recently been proposed for enhancing human face images, known as "face hallucination". In this paper, we propose a novel algorithm to super-resolve face images given multiple partially occluded inputs at different lower resolutions. By integrating hierarchical patch-wise alignment and inter-frame constraints into a Bayesian framework, we can probabilistically align multiple input images at different resolutions and recursively infer the high-resolution face image. We address the problem of fusing partial imagery information through multiple frames and discuss the new algorithm's effectiveness when encountering occluded low-resolution face images. We show promising results compared to those of existing face hallucination methods from both simulated facial database and live video sequences. (c) 2006 Elsevier B.V. All rights reserved.