Hallucinating Faces by Interpolation and Principal Component Analysis
Hua Shen, Shutao Li · 2009
In this paper, we propose a novel face hallucination method which mainly focuses on recovering the lost high-frequency information of low resolution face images. Both the global information and local detailed features are taken into account in the proposed method. Firstly, the resolution of the low-resolution face image is enhanced by an edge-protecting interpolation algorithm. Then, the structural information of the interpolated face is extracted by principal component analysis. Utilizing the training database, we synthesize the local high-frequency details according to its corresponding structural information. At last, we can get a final high-resolution result by integrating the interpolated face with its corresponding local high-frequency details. Experiments demonstrate that this method can achieve face hallucination with good visual quality.