Waveletbased Image Compression

1
Wavelet-based Image Compression
By Heriniaina Andrianirina 1817662 Akakpo
Agbago 1817699 University of Ottawa March,12
2002 This report was prepared for Professor L.
Orozco-Barbosa in partial fulfilment of the
requirements for the course ELG4183
2
Overview

• Introduction.
• What is wavelet compression ?
• Compressing still images.
• Compressing video.
• Applications.
• Discussion.
• Trends.
• Conclusion.

3
Introduction

• During the past years, there has been tremendous
  increase in user demand for multimedia content.
• Multimedia content takes up a lot of storage
  space and bandwidth
• A truecolor 512x512 image would take 0.75 Mbyte
  of space.
• One second of NTSC colour video takes about 23
  Mbytes ! 1
• Hence, we need to compress images and video.

4
What is wavelet compression?

• Wavelet is a transform just like Fourier, DCT,
  Laplace, etc...
• It is the newest compression technology available
  on consumer market 2.
• It performs better than existing techniques in
  this area it achieves better quality and higher
  compression ratios.
• It has many applications in digital signal
  processing.

5
Compressing still images (1)

• Image compression techniques operate by removing
  redundancy and details not perceived by the human
  eye .
• Compression is achieved by keeping only the most
  significant wavelet coefficients (lossy
  compression).
• Compression algorithm allows progressive
  transmission most important information
  transmitted first.

6
Compressing still images (2)

• Wavelet compression is applied on an image as a
  whole3, unlike DCT based transforms.
• Wavelet gracefully degrades the image quality4
  and suppresses the blocky artefacts found in
  JPEG.
• This results in better performance than JPEG at
  high compression rates.
• Wavelet-based algorithms compress images up to
  200 times with no appreciable degradation in
  quality 5

7
Comparison with JPEG
Fig 1 Images compressed with DWT and JPEG

• Left image uses Discrete Wavelet Transform, right
  image uses JPEG. Both images have compression
  ratio of 1001 1

8
Compressing video

• Compression takes advantage of inter-frame
  redundancy successive frames have very little
  differences.
• Video compression and transmission over
  heterogeneous networks requires
• Scalability.
• Packet loss and error resilience.
• Low delay in real-time situation.
• Low Jitter.

9
Compression techniques

• Two ways to achieve compression
• Compute the difference between two successive
  frames and encode the residual frame similar to
  MPEG, results in peaks during transmission.
• Use 3-D wavelet transforms on a group of frame
  don't use motion compensation.
• The most popular algorithms are
• 3D-SPIHT Set Partitioning in Hierarchical Trees.
• 3D-IEZW Improved Embedded Zero-tree coding.
• Achieves better results than MPEG-2

10
Scalability

• These "embedded" codes achieve scalability and
  robustness.
• Coefficients transmitted progressively First
  bits provide a coarse description of a frame.
• The image is refined as coefficients are
  received.
• It allows precise bit rate control by sender in
  response to changing network conditions 6.
• Temporal and spatial scalability.

11
Comparison with MPEG-2 (1)
Table 1 Comparison of PSNR of video compressed
with MPEG and wavelet 1
12
Comparison with MPEG-2 (2)

• Fig 2 Left video uses 3D wavelet with a GOF of
  16 frames, right video uses MPEG-2. Both have bit
  rate of 0.2 bpp

13
Applications

• Storage of fingerprints by the FBI7.
• Medical imaging.
• Alternates for current image compression
  standards used on the internet.
• Video surveillance industry hardware solutions
• Real-time video applications video conferencing,
  video on demand. E.g. VDOLive.
• Video Broadcasting.
• Image restoration and denoising.

14
Discussion

• Advantages
• Scalability and error resilience.
• Better quality at high compression rates.
• Simple algorithm easy to implement with
  hardware.
• Precise rate control.
• Disadvantages
• 3D wavelet algorithms introduce additional delay
  and needs relatively large buffer.
• Software playback takes time.
• Currently all solutions are proprietary no
  standards.

15
Some trends

• Wavelet compression has been well accepted in the
  field of video and image compression.
• It is expected to become a widespread technique.
• Large companies are interested.
• JPEG 2000 is a big step towards standardization
  5.
• Wireless video transmission 8.

16
Conclusion

• Wavelet-based image compression is a new buzz
  technology.
• It outperforms current formats used on the
  internet (JPEG, MPEG-2).
• It is well suited for image and video
  transmission over the internet.
• It is expected to become the next generation
  image compression standard.

17
Questions (1)

• What is wavelet-based image compression?
• Wavelet compression uses wavelet transform to
  compress images. Wavelet is a transform similar
  to Fourier, Laplace and DCT transforms. It
  achieves better performance than existing
  image-compression standards.
• What is embedded code as explained here?
• Embedded code packs the most important data at
  the beginning of a bit stream, further data
  refines the image. It can be truncated at any
  point and the decoder will be able to construct
  an image from it.
• What is the most important advantage of
  wavelet-based image compression?
• It is its scalability. The use of embedded code
  allows the sender to control precisely the
  transmission rate and therefore the image quality
  in order to adapt to changing network conditions.

18
Questions (2)

• What are the applications of wavelet image
  compression?
• - Real-time video transmission over the
  internet video on demand, video conferencing.
• - Video surveillance.
• - Medical imaging.
• - Storage of fingerprints.
• - Video Broadcasting.
• What is the main obstacle for widespread use of
  wavelet now?
• The main obstacle is the lack of standards. All
  solutions are proprietary and incompatible with
  one another.

19
References

• 1 Lee, K. Park, S, Suh, W. Wavelet-based
  Image and Video Compression. TCOM 502, April,
  1997. http//www.seas.upenn.edu/ksl/Classes/TCOM5
  02/Wavelets/
• 2 Loronix Information Systems, Inc. Wavelet
  technology The next generation Digital video
  compression. 2002. http//www.loronix.com/solution
  s/whitepapers/wavelet.asp
• 3 Lee, T. Wavelet history. Wavelet Technology
  Marketing Ltd. December, 1995. http//www.wavestor
  e.com/wavelet-history.html
• 4 Ögren, M. Wavelet methods for image and video
  transmission. Telia Research AB.
  http//www.itm.se/NTM/Annual_Report/ar_1997/GEP/Ex
  Pr/MatsOgren.html
• 5 Johnson, R. C. JPEG2000 wavelet compression
  spec approved. December 29, 1999.
  http//www.eetimes.com/story/OEG19991228S0028
• 6 Kim, B. Pearlman, W. A.. An embedded wavelet
  Video Coder Using Three-dimensional set
  partitioning in hierarchical trees (SPIHT).
  Department of electrical, computer and systems
  engineering Rensselaer Polytechnic Institute,
  Troy, NY 12180.
• 7 Brislawn, K. The FBI Fingerprint Image
  Compression Standard. July 12, 1996.
  http//www.c3.lanl.gov/brislawn/FBI/FBI.html
• 8 Pedagog integrates LuraTechs wavelet
  compression to enable worlds first video
  surveillance over mobile phone networks. Pedagog
  Ltd. http//www.luratech.com/company/press/pressre
  leases/000807LuraTechPedagog.doc