Date of Award
Electrical and Computer Engineering
Kwan, H. K.
Engineering, Electronics and Electrical.
CC BY-NC-ND 4.0
The objective of this thesis is to investigate the lattice modeling technique on the application of image compression. Auto-Regressive (AR) modeling is a classical model for linear prediction. To obtain the model parameters based on the Least Square Error (LSE) criterion, lattice algorithms provide an efficient and robust approach. With prediction coding as the framework an accurate model or prediction can yield a better reduction in entropy or redundancy. As a result, coding on the less well correlated prediction residue will result in a compression. The characteristics of the prediction residue have also benefited the quantization process. For this purpose, Vector Quantization and various alternative approaches have been studied. Different from time domain, a 2-D image has already contained all the information when we intend to process it, and therefore, it allows a noncausal support plane. To avoid the synthesis problem that encountered for noncausal support, this thesis presents two alternative methods: Binary Decomposition and Matrix Method. Various aforementioned methods have been implemented and tested using Matlab and C++. Comparisons of results show the competitiveness and effectiveness of the noncausal lattice modeling for image compression. Paper copy at Leddy Library: Theses & Major Papers - Basement, West Bldg. / Call Number: Thesis1997 .C424. Source: Masters Abstracts International, Volume: 39-02, page: 0560. Adviser: H. K. Kwan. Thesis (M.A.Sc.)--University of Windsor (Canada), 1998.
Chan, Raymond Chi Ho., "Noncausal predictive lattice modeling for image compression." (1998). Electronic Theses and Dissertations. 2862.