Image Enlargement with Lost High-Frequency Components Estimation Using Clustered Eigenspace-BPLP

In this paper, in order to realize the image enlargement with high performance, we propose a new enlargement method, which restores high-frequency components lost in the interpolation-based methods, by using high-frequency estimation. The estimation is based on the eigenspace method proposed by Amano et al.. The proposed enlargement method can generate lost high-frequency components by using pairs of low- and high-frequency components of the original image, and is referred as back projection for lost high-frequency components. Experimental results show the proposed method can achieve promising enlargement.