JPEG 2000 CS 525 Research Project Spring 2008

1. JPEG 2000CS 525 Research ProjectSpring 2008 Presented By - Ankur Chattopadhyay University Of Colorado At Colorado Springs

2. Overview • What is JPEG 2000? • Technology Description (The Overall Technique) • JPEG2000 Vs JPEG (Analysis) • Tools Used For Testing Technology • Results From The Tests Conducted • Conclusions : Why and where should we use JPEG2K?

3. So what is JPEG2K? • Image compression standard for the next millennium • Uses DWT (Discrete Wavelet Transform) instead of DCT • Intended to compliment, not replace, the current JPEG standards • Addresses areas where JPEG (or previous compression techniques) have failed to give best quality performance

4. Main Criteria • Progressive recovery of an image by resolution or fidelity • Region Of Interest (ROI) coding (i.e. specific regions can have different resolution) • Random access to any region without fully decoding image stream • File format allows opacity information and image sequences (animation) • Good error resilience

5. Standard Codec Process

6. JPEG2K Codec Components • Source Image Model: RGB, Grayscale ‘Sheets’ (Varying size; maximum no. 2^14) • Reference Grid: Square Samples; Maximum Dimension 2^32 -1 • Tiling: Divide each ‘sheet’ component into 9 smaller rectangles • Codec Structure: - Intercomponent Transform: Irreversible Component Transform (ICT) – real lossy , Reversible Color Transform (RCT) – integer lossy / lossless • Intracomponent Transform: 1D DWT (2 sub bands); 2D DWT (4 sub bands); variable component resolution R

7. Intra Component Transform Subbands

8. JPEG2K Codec Components • Quantization/Dequantization: same as JPEG • Tier-1 Coding: Bit-Plane Coding (coefficients sent into code-blocks and then packaged into “planes” with symbols) • Tier-2 Coding: Code blocks are placed into ‘data-packets’; bit-errors can occur but at individual packet level • ROI Encoding: Focus on area ofpriority • Code Stream: Packetizeddata in integer byte size • File Format: Allows opacity, ownership, origin, even animation

9. Why not JPEG? Shortcomings of JPEG:- • Lossy by nature • Based on DCT blocks (fixed length size) • Correlation across the block boundaries lead to noticeable and annoying ‘blocking artifacts' particularly at low bit rates • No provision/support for:- • Multi resolution • Progressive transmission • Robust error handling • Advanced file format

10. Why JPEG2K? Wavelet based scheme (DWT) outperforms DCT & other coding schemes:- • Basic components have variable length • Flexible sub-band structures • At higher compression successfully avoids ‘blocking artifacts’ • More robust under transmission and decoding errors • Facilitates progressive transmission of images • Better matched to the HVS characteristics because of inherent multi resolution nature • Suitable for applications where scalability and tolerable degradation are critical

11. What Are Blocking Artifacts At High Compression?

12. Analyzing JPEG2K Strengths of JPEG2000 Weaknesses of JPEG2000 • For “massive compressions”, it wins hands down • Provides uniformity • Accommodates both lossy and lossless modes • ROI encoding feature • Scaled image reconstruction (progressive display) • Images can appear out-of-focus and less detailed after being compressed as the wavelet transform uniformly compresses the entire image (as opposed to DCT block encoding in JPEG)

13. LuraWave SmartCompress Tool • Software product by LuraTech; Free download (demo version) at www.luratech.com in different forms of packages • Successfully implements JPEG 2000 wavelet compression • Advantages: - Fast compression and decompression • Program can be run from the command line or from other programs enabling the program to act as a viewer for custom application • LuraWave.jp2 images are full fledged JPEG2000 images • Standard and can be viewed with any JPEG2000 compliant software • Both lossy & lossless compression at a selectable, user-defined compression rate, image (file) size or visual quality • Supported Image Formats: TIFF, BMP, PPM, PGM,JPEG, .JP2 and LWF (LuraWaveFormat) • Browser plug-ins: MS compatible, Netscape and ActiveX available; also PhotoShop, QuickTime, etc.

14. JPEG 2000 Test Tool • Memory optimized compression (Baseline) • Compression/Decompression supporting user-specified regions of higher quality within the image (ROI) • Integrated with image password protection • Another suggested tool: CompressIt by WaveL Software

15. Test Results Conversion of JPEG to Lossless .JP2 takes up more memory. Original: JPEG 2300 KB Lossless: 815 KB

16. 1:168 LOSSY COMPRESSION Recall: original file was 2300 KB JPEG image .jp2 13.7 K JPEG 13.7 K

17. Summary – JPEG2K Vs JPEG

18. JPEG2K Vs JPEG In Pics Original 969 KB Bitmap .jp2 7.29 KB JPEG 7.29 KB

19. References • ISO/IEC JTC 1/SC 29/WG 1 N 1814, “A study of JPEG 2000 still image coding versus other standards,” July 2000 • ISO/IEC JTC 1/SC 29/WG 1 N 2412, “The JPEG-2000 Still Image Compression,” September 1, 2001 • Richard Clark, “An introduction to JPEG 2000 and watermarking” (http://elysium.ltd.uk ) • R. Colin Johnson, “JPEG2000 wavelet compression spec approved,” December 29, 1999 (http://www.eetimes.com/story/OEG19991228S0028 ) • http://www.amara.com/IEEEwave/IEEEwavelet.html (Publications & Other Material By Amara Graps) • http://perso.orange.fr/polyvalens/clemens/wavelets/wavelets.html (Publications & Other Material By Clemens Valens) • http://www.bearcave.com/misl/misl_tech/wavelets/index.html (Publications & Other Material By Ian Kaplan) • http://users.rowan.edu/~polikar/WAVELETS/WTtutorial.html (Publications & Other Material By Robi Polikar)

20. Questions???