Date of Award

1999

Document Type

Thesis

Publisher

Edith Cowan University

Degree Name

Bachelor of Engineering Honours

Faculty

Faculty of Communications, Health and Science

First Supervisor

Dr Ganesh Kothapalli

Abstract

This thesis discusses the findings of the final year project involving the VHDL (V= Very High Speed Integrated Circuit, Hardware Description Language) design and simulation of an EZT (Embedded Zero Tree) codec. The basis of image compression and the various image compression techniques that are available today have been explored. This provided a clear understanding of image compression as a whole. An in depth understanding of wavelet transform theory was vital to the understanding of the edge that this transform provides over other transforms for image compression. Both the mathematics of it and how it is implemented using sets of high pass and low pass filters have been studied and presented. At the heart of the EZT codec is the EZW (Embedded Zerotree Wavelet) algorithm, as this is the algorithm that has been implemented in the codec. This required a thorough study and understanding of the algorithm and the various terms used in it. A generic single processor codec capable of handling any size of zerotree coefficients of images was designed. Once the coding and decoding strategy of this single processor had been figured out, it was easily extended to a codec with three parallel processors. This parallel architecture uses the same coding and decoding methods as in the single processor except that each processor in the parallel processing now handles only a third of the coefficients, thus promising a much speedier codec as compared to the first one. Both designs were then translated into VHDL behavioral level codes. The codes were then simulated and the results were verified. Once the simulations were completed the next aim for the project, namely synthesizing the design, was embarked upon. Of the two logical parts of the encoder, only the significance map generator has been synthesized.

Share

 
COinS