A Study on JPEG2000 Data Compression and Coding

碩士 === 國立高雄第一科技大學 === 電腦與通訊工程所 === 92 ===   In the changeable ages, how to communicate the messages efficiently is a significant topic about our life. However, the information includes texts, images, … etc. As the texts don’t occupy much memory space, it’s important to record and store the images ob...

Full description

Bibliographic Details
Main Authors: Zih-Ren Lai, 賴梓仁
Other Authors: I-Chang Jou
Format: Others
Language:zh-TW
Published: 2004
Online Access:http://ndltd.ncl.edu.tw/handle/77560778576214758765
Description
Summary:碩士 === 國立高雄第一科技大學 === 電腦與通訊工程所 === 92 ===   In the changeable ages, how to communicate the messages efficiently is a significant topic about our life. However, the information includes texts, images, … etc. As the texts don’t occupy much memory space, it’s important to record and store the images obviously.   When the existing image compression standard couldn’t satisfy the users’ requirement, we should hurry and pay more and more attention on researching and developing a new one to instead of. In 2000, JPEG2000 was anticipated to be announced. It provides not only the better subjective image quality, but also the better objective compression ratio, and supports numerous unprecedented features that are not available by the other still image compression standards.   In this thesis, we’ll discuss the techniques of transformation, quantization, and entropy coding in JPEG2000 Part I. And then, try to program the main functionality of JPEG2000 encoding/decoding modules by Matlab and use two criteria, compression ratio and peak signal-to-noise ratio, to evaluate the compression efficiency and reconstructed image quality of JPEG2000 respectively. JPEG2000 adopts discrete wavelet transformation and provides two mode:9/7 lossy wavelet transformation and 5/3 lossless wavelet transformation. The 5/3 mode use fixed-point wavelet coefficients and support lossy and lossless compression; the 9/7 mode use floating-point wavelet coefficients and only support lossy compression. In quantization, JPEG2000 adopts scalar quantization to reduce the valid image data. Finally, pass the valid image data into the embedded block coding with optimized truncation algorithm. It consists of context formation, binary arithmetic coding, and bit-stream packing.   In hardware design, we used HDL Verilog to describe the architecture of arithmetic encoder and decoder. Furthermore, Use Synopsys’ Design Compiler to convert the Register-Transistor-Level code into Gate-Level netlist-file and verify the result of arithmetic encoding/decoding by hardware simulation are the same as the software’s or not.