A Robust Approach for Low Resolution Medical Image Enhancement using DTCWT and SVD

Author Name(s): M.Mahesh, T. Murali Krishna, B.Naresh Babu
Author Email:

Abstract

Low Resolution Image Enhancement is one of the most important features in Image processing domain whose main target is to improve with the help of which contrast and the visual appearance of an image can be improved that makes the input image more suitable than the original one for specific application. Low Resolution Medical Image enhancement is the active research field. A novel approach to enhance Low resolution medical images is combination of Dual tree complex wavelet transform(DTCWT), A Non-Local Mean(NLM)  filter and singular value decomposition (SVD).Here, SVD is proposed to enhance the contrast of an image while high frequency sub bands are produced using dual-tree complex wavelet transform. A NLM Filter is used to remove the artifacts produced by dual-tree complex wavelet transform. Interpolated filtered high frequency sub-bands and contrast enhanced low resolution image are combined using inverse dual-tree complex wavelet transform to obtain contrast enhanced image of super resolution. Qualitative and quantitative measures such as Peak Signal to Noise Ratio (PSNR), Mean Square Error and Mean Absolute Error are used to justify the importance of proposed technique.

Introduction

Low Resolution enhancement of medical images has been a matter of concern and a very attractive field for the researchers as well. As analysis of medical images mainly depends on the size and quality of acquired image like noise immunity, illumination conditions etc [1]. For proper diagnosis the interpretation of medical images and its resolution enhancement are very much desirable. Therefore many different techniques has been proposed for medical image resolution enhancement (MIRE)[2]. There has been a lot of research mainly on frequency domain methods. After the fast decomposition method given by Mallat [3], Wavelet Transform emerged as a new time –frequency analysis tool which accomplished desired results in image processing field. Many image enhancement methods were developed based on wavelet transform such as Lu et al[4], Yang and Hansell [5], Fang and Qi[6], Zhou et al [7] etc. General approach of image enhancement lies in getting detail information of image through wavelet transform by decomposing high frequency sub-images to get enhanced image of better visual quality.

As our visual system is very much sensitive to contrast, for surgeons and physicians in addition to RE (to assist interpolation and diagnosis) contrast enhancement (CE) is also equally important and desirable. In image processing applications generally most effective and simplest. In [8] using un-decimated decomposition was used in stationary wavelet transform approach which outperforms other conventional wavelet techniques.

Discrete wavelet transform (DWT) in combination with SWT and nonlocal means filter is used in [9]. As visual system of human being is more sensitive to contrast therefore enhancement is also important for physician and surgeons in addition to other image processing techniques to for diagnosis purpose. The local histogram equalization and generalized histogram equalization methods are also used for contrast enhancement in many image processing applications. Intensity information of an image is contained in singular values [11] and good contrast image can be obtained by changing the singular value. In [12, 13] Image enhancement has been       done using Singular value decomposition which has proved better than conventional techniques i.e. histogram based approaches. In [14] Discrete wavelet transform is used for contrast enhancement. Least squares trained filters are used in [15] for image resolution enhancement. The method proposed in [16] shows impressive results but this technique uses only training process not temporal which is also one of the important factors for image quality enhancement.

Conclusion

The proposed algorithm is a hybrid approach using DTCWT, NLM filter and SVD and has been tested on a set of medical images. The medical input image is decomposed using DTCWT. Less artifacts are generated with the help of DTCWT compared to that of DWT because of nearly shift-invariance characteristic of DTCWT. Further image quality is improved using NLM filtering approach and SVD is used for to get originality of image and obtain a better quality image both quantitatively and qualitatively. Simulation results showed that proposed technique outperforms other conventional techniques for improving visual quality of medical images for proper manual interpretation and Computer based diagnosis. This can be considered as an efficient method which will be very useful in some important areas for image enhancement in satellite imaging, remote imaging, medical imaging etc

References

  1. Gagandeep Kaur ,Shruti Vashist,”A Robust Approach for Medical image Enhancement using DTCWT” , International Journal of Computer Applications (0975 – 8887) Volume 167 – No.6, June 2017
  2. R C Gonzalez, R. E. Woods ,”Digital Image Processing Using MATLAB”,  2nd ed, Prentice Hall
  3. Mallat, S.G.: ‘Multifrequency channel decompositions of image and wavelet  models’,  IEEE Trans.        Speech Signal Process., 1989, 37, (12), pp. 2091–2110.
  4. Lu, J., Healy, D.M., and Weaver, J.B.: „Contrast enhancement of medical images using multi-scale edge repressentation‟, Opt. Eng., 1994, 33, (7), pp. 2151– 2161.
  5. Yang, G. , and Hansell, D. M. : ‘CT image enhancement with wavelet analysis for the detection of small airways disease’, IEEE Trans. Med. Imaging, 1997, 16, (6), pp. 953–961.
  6. Zhou, X., Zhou, S., Huang, F., and Zhou, X.T.: „New algorithm of image enhancement based on wavelet transform‟, Comput. Appl., 2005, 25, (3),pp. 606–608.
  7. L. Lee, C.C. Yang, H.T. Wu, M.J. Chen, Wavelet-based interpolation scheme for resolution enhancement of medical images, J. Signal Process. Syst. 55 (2009) pp.251–265.
  8. Z. Iqbal, A. Ghafoor, A.M. Siddiqui, Wavelet transform and non local means based super resolution, in: International Symposium.
  9. Demirel, C. Ozcinar, G. Anbarjafari, Satellite image contrast enhancement using discrete wavelet transform and singular value decomposition, IEEE Geosci. Remote Sens. Lett. 7 (April (2)) (2010).
  10. Tian, T. Tan, Y. Wang, Y. Fang, Do singular values contain adequate information for face recognition, Pattern Recognit. 36 (3) (2003) pp. 649–655.
  11. Demirel, G. Anbarjafari, M.N.S. Jahromi, Image equalization based on singular value decomposition, in: IEEE International Symposium on Computer and Information Sciences, October 27–29, 2008, pp. 1–5.
  12. Fang, Y., and Qi, F.: „A method of wavelet image enhancement based on soft threshold‟, Comput. Eng. Appl., 2002, 23, pp. 16–19.
  13. Demirel, G. Anbarjafari, M.N.S. Jahromi, Image equalization based on singular value decomposition, in: IEEE International Symposium on Computer and Information Sciences, October 27–29, 2008, pp. 1–5.
  14. Yang, Z. Su, L. Sun, Medical image enhancement algorithm based on wavelet transform, Electron. Lett. 46 (January (2)) (2010).
  15. L. Starck, F. Murtagh, J.M. Fadili, Sparse Image and Signal ProcesQ5 sing: Wavelets, Curvelets, Morphological Diversity, Cambridge University Press, 2010.
  16. Shao, R. Yan, X. Li, Y. Lin, From heuristic optimization to dictionary learning: a review and comprehensive comparison of image denoising algorithm, IEEE Trans. Cybern. (2013).
  17. Khare, A., Khare, M., Jeong, Y., Kim, H., and Jeon, M.: ‘Despeckling ofmedical ultrasound images using daubechies complex wavelet transform’, Signal Process., 2010, pp. 428–439.
  18. Nick Kingsbury,”Shift invariant properties of the dual tree Complex Wavelet Transform” IEEE Trans. On signal processing letters,1999.

608 total views, no views today

Download PDF File

About the author: admin