Super Resolution of satellite images using wavelet based edge extraction through sparse representation

Author Name(s): *M. HemaLatha, S.Varadarajan
Author Email: maddihemalatha@gmail.com

Abstract

In this research paper a super-resolution (SR) image is generated from low resolution satellite image. By using high frequency (HF) as intermediate stage, sharper image can be generated. Interpolation is performed between satellite portrait and HF sub bands through the discrete wavelet transform which preserves edge. Sparse signal is obtained by applying sparse mixing weights to low resolution (LR) image. Lanczos interpolation is used to avoid artifacts generated in an image. High resolution (HR) image is obtained by using all bands and inverse DWT is performed. Experimental results proves that proposed method has superior performance over existing methods in terms of intension criteria and skewed  perception improving the image resolution. The quantitative PSNR (peak signal to noise ratio), RMSE (root mean square value), and CC (correlation coefficient) are calculated for satellite images. The proposed technique i.e., super resolution using wavelet domain interpolation with edge extraction and sparse representation (SR-WDIEE-SR) has got better values compared to DWT (Discrete wavelet transform).

Introduction

The images obtained from medical, optical, and radar is seen on electron microscopy and high definition television, obtained from electronic sensors. Image enhancement techniques are applied to low resolution images to get high resolution images. This algorithm is preprocessing technique used in many applications such as satellite image processing, radar, and optical. Whenever image enhancement is done, the drawback is some noise is added. This noise is to be eliminated by using some denoising algorithms. SR algorithms cost is less compared to other sensors used to increase the resolution of images [1], [2].  Recently many SR algorithms have been proposed [3]. The wavelet based sparse mixing plays very important role in image enhancement of satellite images. Spatial and spectral information of satellite image is enhanced by this proposed algorithm [4].

The major difference between existing SR algorithms and proposed novel SR algorithm exists in mutual interpolation by lanczos [5], [6] and nearest neighbor interpolation (NNI) to wavelet transform high frequency sub bands. Out of these interpolation techniques lanczos interpolation is used because it has got many merits compared to others. It reduces ringing, sharpness and aliasing. It also detects edges and linear features.  Edge extraction is done by DWT and adaptive LR image is computed by sparse representation of low frequency sub bands. NLM  is used to avoid artifacts. Loss of low frequency results noise. The noise present in LL sub band is eliminated by NLM (non local mean) filter [7]. Finally, all sub bands are combined via inverse DWT. The output is super resolution image in terms of objective and subjective perception.

To give reason for that the new algorithm called super resolution using wavelet domain interpolation with edge extraction and sparse representation (SR-WDIEE-SR) has real compensation we have compared the proposed SR algorithm with other similar algorithms, Discrete Wavelet Transform-based Satellite Image Resolution Enhancement [8].

Conclusion

In this research paper, a new resolution-enhancement method based on the sparse based interpolation of the HF sub band images in the discrete wavelet domain has been implemented. This proposed technique (SR-WDIEE-SR) has superior performance in terms of peak signal to noise ratio, root mean square value and correlation coefficient when compared to DWT method. Experimental results highlight the superior performance of the proposed algorithm in terms of objective criteria, as well as in the subjective perception via the human visual system, in comparison with other conventional methods.

Acknowledgement

The authors would like to thank Google Earth from where satellite images are downloaded for research purpose.

References

  1. V. Shkvarko, J. Tuxpan, and S. R. Santos, “l2 − l1 structured descriptive experiment design regularization based enhancement of fractional SAR imagery,” Signal Process., vol. 93, no. 12, pp. 3553–3566, Dec. 2013.
  2. M. Lillesand, R. W. Kiefer, and J. W. Chipman, Remote Sensing and Image Interpretation. Hoboken, NJ, USA: Wiley, 2004, xiv 763 pp.
  3. Temizel and T. Vlachos, “Image resolution up scaling in the wavelet domain using directional cycle spinning,” J. Electron. Imaging, vol. 14, no. 4, p. 040501, 2005.
  4. Elad, M. A. T. Figueiredo, and Y. Ma, On the role of sparse and redundant representations in image processing, Proc. IEEE, vol. 98, no. 6, pp. 972–982, Jun. 2010.
  5. Unser, Splines : A perfect fit for signal and image processing, IEEE Signal Process. Mag., vol. 16, no. 6, pp. 22–38, Nov. 1999.
  6. Herminio Chavez-Roman and Volodymyr Ponomaryov, super resolution image generation using wavelet domain interpolation with  edge extraction via sparse representation , IEEE geoscience and remote sensing letters, vol. 11, no. 10, pp. 1777–1781, 2014.
  7. Protter, M. Elad, H. Takeda, and P. Milanfar, Generalizing the nonlocal-means to super-resolution reconstruction, IEEE Trans. Image Process, vol. 18, no. 1, pp. 36–51, Jan. 2009.
  8. Demirel and G. Anbarjafari, Discrete wavelet transform-based satellite image resolution enhancement, IEEE Trans. Geosci. Remote Sens., vol. 49, no. 6, pp. 1997–2004, Jun. 2011.
  9. Hema Latha and Dr.S.Varadarajan, A Survey on Satellite Images Resolution Enhancement Using DWT, DWT-SWT and SVD    International Journal of Innovations in Engineering and Technology ,Vol.7 ,Issue 3 , pp. 461-465,October 2016.
  10. Hemalatha, Dr.S.Varadarajan and A.V.Kiranmai, Contrast Enhancement of Satellite  Images using  Dual-Tree  Complex Wavelet Transform Technique ,International Journal of Innovations in Engineering and Technology Journal, Vol.3 ,Issue 10 ,pp. 86-92, October 2016.

701 total views, no views today

Download PDF File

About the author: admin