Super Resolution of Quality Images through Sparse Representation

  • A. Bhaskara Rao
  • J. Vasudeva Rao
Part of the Advances in Intelligent Systems and Computing book series (AISC, volume 249)


This paper addresses the problem of generating the super resolution (SR) image from a single low resolution input image. Image patches can be represented as a sparse linear combination of elements from an over-complete dictionary. The low resolution image is viewed as down sampled version of a high resolution image. We look for a sparse representation for each patch of the low resolution image, and then use the coefficients of this representation to generate high resolution. Theoretically the sparse representation can be correctly recovered from the down sampled signals. The low and high resolution image patches are mutually training two dictionaries. We can look for the similarity between low and high resolution image patch pair of sparse representations with respect to their own dictionaries. Hence the high resolution image patch is applied to sparse representation of a low resolution image patch. This approach is more compact representation of the patch pairs compared to previous approaches. The earlier approaches simply sample a large amount of image patch pairs. The effectiveness of sparsity prior is demonstrated for general image super resolution. In this case, our algorithm generates high resolution images that are competitive or even superior in quality to images produced by other similar SR methods. This algorithm is practically developed and tested and it is generating high resolution image patches. The results are compared and analyzed with other similar methods.


SR sparse representation dictionary noise sparsity 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Yang, J., Wright, J., Huang, T., Ma, Y.: Image super-resolution as sparse representation of raw image patches. In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 1–8 (2008)Google Scholar
  2. 2.
    Tipping, M.E., Bishop, C.M.: Bayesian image super-resolution. In: Proc. Advances in Neural Information and Processing Systems 16, NIPS (2003)Google Scholar
  3. 3.
    Yang, J., Wright, J., Huang, T., Ma, Y.: Image Super-Resolution via Sparse Representation. IEEE Transactions on Image Processing 19(11) (2010)Google Scholar
  4. 4.
    Donoho, D.L.: For most large underdetermined systems of linear equations, the minimal ℓ1-norm solution is also the sparsest solution. Comm. on Pure and Applied Math 59(6) (2006)Google Scholar
  5. 5.
    Donoho, D.L.: For most large underdetermined systems of linear equations, the minimal ℓ1-norm near-solution approximates the sparsest near-solution (2004) (Preprint)Google Scholar
  6. 6.
    Freeman, W.T., Jones, T.R., Pasztor, E.C.: Example based super-resolution. IEEE Computer Graphics and Applications 22(2) (2002)Google Scholar
  7. 7.
    Elad, M., Aharon, M.: Image denoising via sparse and redundant representations over learned dictionaries. IEEE TIP 15(12) (2006)Google Scholar
  8. 8.
    Mairal, J., Sapiro, G., Elad, M.: Learning multiscale sparse representations for image and video restoration. SIAM Multiscale Modeling and Simulation (2008)Google Scholar
  9. 9.
    Aharon, M., Elad, M., Bruckstein, A.: K-SVD: An algorithm for designing over complete dictionaries for sparse representation. IEEE Transactions on Signal Processing 54(11), 4311–4322 (2006)CrossRefGoogle Scholar
  10. 10.
    Murray, J.F., Kreutz-Delgado, K.: Learning sparse over complete codes for images. The Journal of VLSI Signal Processing Systems for Signal, Image, and Video Technology 46(1), 1–13 (2007)CrossRefGoogle Scholar

Copyright information

© Springer International Publishing Switzerland 2014

Authors and Affiliations

  1. 1.Department of Computer Science and EngineeringGMR Institute of TechnologyRajamIndia

Personalised recommendations