Super-Resolution for HD to 4K using Analysis K-SVD dictionary and Adaptive Elastic-Net

This paper, we present Super-Resolution algorithm for increase resolution of High Definition resolution video to 4K resolution video. Our proposed method combined K-means clustering via a Singular Value Decomposition with Analysis pursuit (Analysis K-SVD) algorithm and Adaptive Elastic-Net algorithm. Analysis K-SVD algorithm improves the learning process so that better dictionary coefficients are generated such that the optimal path between low resolution and high resolution patch datasets is determined. The Adaptive Elastic-Net algorithm is applied to estimate high resolution patch as a solution to sparse representation problem. Experimental results show that our proposed method improves the PSNR around 1 dB compared with bicubic interpolation and sparse representation method.