Hallucinating Facial Image Based on Adaptive Neighbourhood Selection
Liu Fang, Yu Fu Deng · Advances in engineering research/Advances in Engineering Research · 2015
In most manifold learning based face hallucination algorithms, the nearest neighbourhood metric is often adopted to describe the face subspace, which could not accurately capture the local geometrical structures of the samples.In this paper, a novel face superresolution approach based on adaptive neighbourhood selection is presented, which can adaptively select the nearest neighbours for each sample point.The corresponding neighbours of sample points well reflect the local geometrical structure of the face manifold, so that the linear subspace determined by the optimal linear fitting can approximate the local geometry well.Experimental results show that our method is more effective than other manifold learning based strategies for super-resolving face images.