This study developed a face hallucination system based on a novel two-dimensional direct combined model (2DDCM) approach that employs a large collection of low-resolution/high-resolution facial pairwise training examples. The proposed 2DDCM approach achieves face hallucination by addressing three key issues. First, we directly combine each low-resolution and high-resolution pairwise image in a concatenated form in order to completely preserve their relationship. Second, images are formed as two–dimensional matrices instead of vectors in order to preserve the facial geometry. Third, both the vertical and the horizontal facial-geometry features are considered in 2DDCM approach. Experiments demonstrate our approach can synthesize high quality reconstructed facial images from given low-resolution images.
