Performance Evaluation of Hybrid Dual Tree Complex Wavelet Transform based Signal Denoising Technique

Author Name(s): Srinivasulu Reddy, M. Sadasiva, K. Gopi
Author Email: cnudega@gmail.com

Abstract

In this paper, the performance of the signal and denoising techniques using different types of Wavelet Transform techniques is evaluated based on the different performance metrics such as Signal to Noise Ratio(SNR).  A new Complex Wavelet Transform technique called Hybrid Dual Tree Complex Wavelet Transform, has been proposed in this work, so as to achieve the better denoising performance over the existing Wavelet Transform based denoising techniques. As the proposed technique mixes all the features of the existing Wavelet Transform techniques, the best performance can be expected from this technique.

Introduction

It is quite common that any signal can be corrupted with noise during acquisition, transmission and other different sources of noise. To get the meaningful information from this noisy signal, it should be subjected to the pre-processing stage where the signal is made as suitable for further processing. The most significant step in the pre-processing stage is denoising of data which will greatly impact the processing stage performance in different applications viz., medical, defense etc.

As the performance of the time (spatial) domain based denoisingmethods viz., mean filter, median filter, Wiener filter etc. is not effective over the transform domain based denoising methods viz. Fourier Transform, Wavelet Transform etc., the transform domain based denoisingmethods have been opted always for reducing the noise present in the data. The different steps of the transform based denoising techniques are:

  1. Compute the forward transform of the data
  2. Treat the transform domain data using the optimum thresholding techniques
  3. Compute the inverse transform of the treated transform domain data.

The Fourier Transform based denoising technique is not suitable for the non-stationary signals, hence its denoising performance cannot be comparable with Wavelet Transform based denoising   methods which are suitable for non-stationary signals also.

In this work, a comparison between the performance of the different Wavelet Transform based denoising techniques ie evaluate. Moreover, Complex Wavelet Transform (CWT) based denoising techniques are emphasized over the Discrete Wavelet Transform (DWT) based denoising techniques because CWT can overcome the limitation of DWT such as  shift sensitive, poor directionality and absence of phase information.

Conclusion

In this work, the denoising efficiency of the various types of wavelet transform based denoising techniquesis evaluated and compared to one another with the help of the performance metrics SNR and Correlation Coefficient. From the comparison of the performance metrics, it is found that the performance of the proposed technique HDTCWT based signal denoising is better that that of the other techniques like DWT, AWT, DTCWT based signal denoising techniques. As the DWT suffers from the limitation such as shift sensitivity, lack of signal phase information, DWT based denoising technique is not as effective as complex wavelet transform based denoising techniques. AWT and DTCWT based denoising techniques are more efficient because they have overcome the limitations like shift sensitivity and absence signal phase information. Each of these techniques (AWT, DTCWT) perform denoising with their own important features, the performance of these techniques can be maximized by combing both the techniques into a single technique and the same is proved from the results.

HDTCWT based denoising technique can be applied to the different signal and image processing applications like, BiomedcalSigna Processing, Radar Signal Processing etc., Though the proposed technique performs better in signal denoising, there will be always room further improvement of the denoising capability of the technique. The further improved performance can be observed by designing wavelet filters which suits well to the given signal, and by choosing effective shrinkage technique to determine the threshold value for treating the detail coefficients of the wavelet decomposition.

References

  1. Kingsbury, “Complex Wavelets for Shift Invariant Analysis and Filtering of Signals”,Applied and Comp. Harm. Anal.10, 2001, 234-253.
  2. N G Kingsbury, “A Dual-Tree Complex Wavelet Transform with improved orthogonality and symmetry properties”, Proc. IEEE Conf. onImage Processing, Vancouver, September 11-13,           2000, paper 1429.
  3. Iona Adam et a, “A New Implementation of Hyperanalytic Wavelet Transform”, International Conference on Signals, Circuits and Systems, 2007, IEEE Explore,
  4. Firoiu, C. Nafornita, “Image Denoising using a New Implementation of HyperanalyticWavelet Transform”,IEEE 2009
  5. Leela Rani, S. Varadarajan, M. Sadasiva, “Atmospheric Radar Signal Processing using Complex Wavelet Transforms”, International Journal of Computer Applications, ISSN: 0975-8887, Vol 91, No 11, 2014, pp. 33-37
  6. Leela Rani, S.Varadarajan, M.Saasiva, “MST RadarSignal Processing using Analytic Wavelet Transform,”Proc., Second International Conference on AdvancedComputing Methodologies,(ISBN:9789351071495),GRIET, Hyderabad, pp. 353-355,21-23 November 2013
  7. Sreenivasulu Reddy et al, “A New Image Denoising Scheme using Hyperanalytic Wavelet Transfrom with an Adaptive Wiener Filter”, DIGITECH: An International Journal of Computer Science and Applications. (ISSN: 2229-4597), Vol. 3, No. 6, pp. 29-36
  8. Mallat, “A Wavelet Tour of Signal Processing”. Academic Press, 1999.
  9. Sadasiva et al, “A New Wiener Filter based Image Denoising Scheme using Hyperanalytic Wavelet Transform”, International Conference on Emerging Trends in Electrical, Communication and Information Technologies, (ICECT 2012)
  10. L. Donoho, I. M. Johnstone, Ideal spatial adaptation by wavelet shrinkage, Biometrika, 81(3) : 425-455, 1994

701 total views, no views today

Download PDF File

About the author: admin