Color Face Image Super-Resolution Reconstruction Based on Learning

Proposes a super-resolution algorithm on color face image based on quaternion, combining with the ideas of principal component analysis(PCA) and interpolation. By using singular value decomposition to de-noise the color image, improves the quality of the result, and improves the algorithms when the result is not so faithful to the original face image. Quaternion is used to deal with the three components of <em<R,G,B</em< at the same time to increase the efficiency of the operation and used fully the correction of the three components.