Off-campus UNL users: To download campus access dissertations, please use the following link to log into our proxy server with your NU ID and password. When you are done browsing please remember to return to this page and log out.
Non-UNL users: Please talk to your librarian about requesting this dissertation through interlibrary loan.
Vector quantization of non-stationary sources
Abstract
Vector quantization (VQ) is one of the more popular compression techniques to appear in the last twenty years. There are several drawbacks to the use of the common vector quantizers. These include search complexity and memory requirements, especially at higher rates, and a mismatch between the codebook and the inputs. The latter mainly stems from the fact that the VQ is generally designed for a specific rate and a specific class of inputs. When the bandwidth constraints and/or the source statistics change, this can result in severe degradation in the quality of the reconstructed output. We propose an adaptive technique for vector quantization of non-stationary sequences. The technique is an extension of the recursively indexed scalar quantization algorithm (RISQ). This approach involves the use of a small codebook, reducing the computational complexity. The codebook adapts to the input statistics. The use of a recursively indexed VQ (RIVQ) results in distortion limited outputs which can be used in adaptive algorithms. The RIVQ is tested on image data and composite sources.
Subject Area
Electrical engineering
Recommended Citation
Al-Araj, Ali Ghanim, "Vector quantization of non-stationary sources" (1994). ETD collection for University of Nebraska-Lincoln. AAI9519523.
https://digitalcommons.unl.edu/dissertations/AAI9519523