Proposal for ECE734 Final Project
DWT Core of JPEG2000
Zhenyu Tang and Liang Yin
1. Background of JPEG2000
The JPEG2000 is intended to provide a new image coding/decoding system using state of the art compression techniques, based on the use of wavelet technology. It’s wide range of usage includes from portable digital cameras through pre-press, medical imaging. Right now, JPEG2000 is composed of 6 main parts:
- Part 1, JPEG 2000 Image Coding System
- Part 2, Extensions (adds more features and sophistication to the core)
- Part 3, Motion JPEG 2000
- Part 4, Conformance
- Part 5, Reference software (currently Java and C implementations are scoped)
- Part 6, Compound Image file format (for pre-press and fax like applications)
The part 1 (core part) is to be agreed and delivered as a full ISO international standard by the end of 2000. The main procedure for the image coding system is shown below.
2. Our project goal
Since JPEG2000 includes so many complicated parts and has tons of documentation, it is impractical for us to work on the full version for this final project. The two important parts in the coding/decoding processes are Transform and Arithmetic Coding. Since the later is of very mature, we want to focus on the hardware implementation of discrete wavelet transform (both FDWT and IDWT), which will provide the transform coefficients for later stage and is one key part to JPEG2000 implementation.
In JPEG2000, the still image is divided into tiles. These tiles are rectangular arrays that include the same relative portion of all the components that make up the image. Therefore, these tile components can be decoded and reconstructed independently. Each tile component is transformed into a set of two-dimensional sub-band signals (called sub-bands), each representing the activity of the signal in various frequency bands, at various spatial resolution. The different number of levels of spatial resolution NL is called the number of decomosition levels.
To perform the forward DWT (FDWT) on one tile component, the JPEG2000 uses a one-dimensional sub-band decomposition of a one-dimensional set of samples into low-pass coefficients, representing a downsampled low-resolution version of the original set, and high-pass coefficients, representing a downsampled residual version of the original set, needed to perfectly reconstruct the original set from the low-pass set. On the other hand, the inverse DWT (IDWT) utilizes a one-dimensional sub-band recomposition of a one-dimensional set of samples from low-pass and high-pass coefficients. The full description of FDWT and IDWT procedures will be included in our final report.
We plan to implement IDWT with Verilog first, and try to finish FDWT depending on the time limit. All simulation results for each component will be included in the final report. If possible, we will also try to synthesis, implement and download our design to the FPGA Virtex800. In this way, we may demo some results in the hardware.
3. References
JPEG 2000 Part I Final Committee Draft Version 1.0