Elsevier

Digital Signal Processing

Volume 17, Issue 1, January 2007, Pages 189-198
Digital Signal Processing

Medical ultrasound image compression using joint optimization of thresholding quantization and best-basis selection of wavelet packets

https://doi.org/10.1016/j.dsp.2006.05.008Get rights and content

Abstract

This paper introduces an efficient image-coding algorithm using wavelet packets. The algorithm combines the top-down search approach with an operational rate-distortion (R-D) cost function to select the best wavelet packet basis at low-computational cost. The proposed method jointly optimizes the best-basis selection, coefficient “thresholding” and quantizer selection within the minimum description length (MDL) framework to develop a wavelet packet image coder named as JTQ-WP. We present results to verify the usefulness and versatility of this adaptive image coder both on medical US-images and natural images. The experimental results show that the joint optimization has a dramatic effect on the compression performance of medical ultrasound images. To further demonstrate the potential performance of the proposed method in comparison with the current state-of-the-art image coding algorithms, the results on Barbara image are also presented. The results show a coding gain of 0.91 dB over the benchmark wavelet-coding algorithm, SPIHT, on the Barbara image at a bit-rate of 0.25 bpp.

Section snippets

L. Kaur received the M.E. degree from TIET, Patiala, Punjab, in 2000, in computer science and engineering. She has been in the teaching profession since 1992. Recently, she has submitted dissertation for Ph.D. degree at PTU, Jalandhar. Her research interests include image compression and denoising, and wavelets.

References (27)

  • B. Banjanin et al.

    Entropy-threshold method for best-basis selection

    Elsevier J. Image Vision Comput.

    (2001)
  • A.S. Tolba

    Wavelet packet compression of medical images

    Digital Signal Process.

    (2002)
  • N.M. Rajpoot et al.

    Adaptive wavelet packet basis selection for zerotree image coding

    IEEE Trans. Image Process.

    (2003)
  • M.B. Martin

    New image compression techniques using multiwavelets and multiwavelet packets

    IEEE Trans. Image Process.

    (2001)
  • R.R. Coifman et al.

    Orthonormal Wave Packet Bases

    (1990)
  • R.R. Coifman et al.

    Size Properties of Wavelet Packets

    (1992)
  • M.V. Wickerhauser, in: A.K. Peters (Ed.), Adaptive Wavelet Analysis from Theory to Software,...
  • R.R. Coifman et al.

    Entropy-based algorithms for best basis selection

    IEEE Trans. Inform. Theory

    (1992)
  • F.G. Meyer et al.

    Fast adaptive wavelet packet image compression

    IEEE Trans. Image Process.

    (2000)
  • K. Ramchandran et al.

    Best wavelet packet bases in a rate-distortion sense

    IEEE Trans. Image Process.

    (1993)
  • Y. Shoham et al.

    Efficient bit-allocation for an arbitrary set of quantizers

    IEEE Trans. Acoust. Speech Signal Process.

    (1998)
  • J.R. Goldschneider et al.

    Optimal bit allocation and best basis selection for wavelet packets and TSVQ

    IEEE Trans. Image Process.

    (1999)
  • M. Vetterli et al.

    Wavelets and Subband Coding

    (1995)
  • Cited by (29)

    • Simultaneous compression and speckle reduction of clinical breast and fetal ultrasound images using rate-fidelity optimized coding

      2021, Ultrasonics
      Citation Excerpt :

      Furthermore, medical image coding is inherently different from the coding of natural images in the sense that the diagnostic value of the image should be preserved, rather than its overall visual subjective appearance. Various techniques for ultrasound image compression have been suggested [25–32]. The impact of transform-like compression methods had been studied, and it was demonstrated that compression ratios of 20:1 to 40:1 are attainable with fair image quality.

    • Variance-constrained resilient H<inf>∞</inf> filtering for time-varying nonlinear networked systems subject to quantization effects

      2017, Neurocomputing
      Citation Excerpt :

      On another research frontier, it is now well recognized that in comparison to analog systems, digital systems have many advantages, such as small size, high precision and strong anti-interference. On the other hand, in today’s networked systems, the extensive adoption of digital components will inevitably give rise to certain unanticipated phenomena, for example, the quantization effects [8,12,13,29,38]. The error between the original value and quantized value cannot be negligible, and accordingly, quite a few techniques have been proposed to analyze and mitigate the quantization effects, among which the most common way is to treat the quantization errors as uncertainties.

    • Comparison of evolutionary and swarm based computational techniques for multilevel color image thresholding

      2014, Applied Soft Computing Journal
      Citation Excerpt :

      Distinguishing the objects in an image from the background by determining a single threshold value is called bilevel thresholding [1], while separating the image into numerous different regions regarding with their colors by setting multiple threshold values is known as multilevel color image thresholding [2]. Popular applications for image thresholding are video and image compression [3–5], geovideo and document processing [6], particle counting [7,8] and object recognition [9,10]. It is possible to determine the optimum threshold in an image by analyzing the image histogram characteristics or by optimizing an objective function [11].

    • Technology and research developments in carotid image registration

      2012, Biomedical Signal Processing and Control
      Citation Excerpt :

      Intensity-based registration methods can be automated because such geometric features as points and surfaces do not need to be segmented from the images. However, for some applications, intensity-based methods are susceptible to local optima in the cost function and thus need initial transformations that are close to the correct transformation in order to converge reliably [12,14,16,23]. A relatively new class of hybrid approaches combines advantages of both geometric and intensity approaches to achieve more accurate registration in difficult problems.

    • Ultrasound Contrast Imaging Based on a Novel Algorithm Combined Pulse Inversion with Wavelet Transform

      2011, Ultrasound in Medicine and Biology
      Citation Excerpt :

      Wavelet transform (WT) is a promising technique for time-scale (or time-frequency) analysis that decomposes a signal into components that appear at different scales. It can provide more flexible and powerful performance than common correlation analysis and has been extensively utilized in background removing, image compression as well as image denoising (Tan and Brown. 2002; Kaur et al. 2007; Zhang et al. 2001). Among various types of WT, continuous wavelet transform (CWT), because of its redundancy, is a good tool for analysis and feature determination.

    • A hybrid method for MRI brain image classification

      2011, Expert Systems with Applications
    View all citing articles on Scopus

    L. Kaur received the M.E. degree from TIET, Patiala, Punjab, in 2000, in computer science and engineering. She has been in the teaching profession since 1992. Recently, she has submitted dissertation for Ph.D. degree at PTU, Jalandhar. Her research interests include image compression and denoising, and wavelets.

    S. Gupta received the B.Tech. degree from TITS, Bhiwani, in 1992, and the M.E. degree from TIET, Patiala, Punjab, in 1998, both in computer science and engineering. She has been in the teaching profession since 1992. Recently, she has submitted dissertation for Ph.D. degree at PTU, Jalandhar. Her research interests include image processing, image compression and denoising, and wavelet applications.

    R.C. Chauhan received the B.Sc. (Eng.) degree from BHU, and the M.Tech. and Ph.D. degrees from IIT, Roorkee, all in electrical engineering. He is involved in teaching and research activities for the last 21 years. Presently, he is working as Director, DIT, Dehradun. He also worked in the power industry as chief engineer for 11 years. His research interests include signal processing and power engineering.

    S.C. Saxena received the B.E. degree from Allahabad University in 1970, and the M.E. and Ph.D. degrees from IIT, Roorkee, all in electrical engineering. He has been teaching and involved in research activities since 1973. Presently, he is working as Director, IIT, Roorkee. He has published over 150 research papers and guided a number of research scholars for their Ph.D. theses. His research interests include biomedical engineering, measurement and instrumentation, and signal processing.

    View full text