Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.660304
Title: Improved algorithms for VQ codeword search, codebook design and codebook index assignment
Author: Pan, Jenq-Shyang
Awarding Body: University of Edinburgh
Current Institution: University of Edinburgh
Date of Award: 1996
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
Abstract:
This thesis investigates efficient codeword search algorithms and efficient clustering algorithms for vector quantization (VQ), improved codebook design algorithms and improved codebook index assignment for noisy channels. In the investigation of codeword search algorithms, several fast approaches are proposed, such as the improved absolute error inequality criterion, improved algorithms for partial distortion search, improved algorithms for extended partial distortion search and a fast approximate search algorithm. The bound for the Minkowski metric is derived as the generalised form of the partial distortion search algorithm, hypercube approach, absolute error inequality criterion and improved absolute error inequality criterion. This bound provides a better criterion than the absolute error inequality elimination rule on the Euclidean distortion measure. For the Minkowski metric of order n, this bound contributes the elimination criterion from the L1 metric to the Ln metric. This bound is also extended to the bound for the quadratic metric by using methods of metric transformation. The improved absolute error inequality criterion is also extended to the generalised form of the mean-distance-ordered search algorithm for VQ image coding. Several fast clustering algorithms for vector quantization based on the LBG algorithm are presented. Genetic algorithms are applied to codebook design to derive improved codevectors. The approach of stochastic relaxation is also applied to the mutation step of the genetic algorithm to further improve the codebook design algorithm. Vector quantization is very efficient for data compression of speech and images where the binary indices of the optimally chosen codevectors are used. The effect of channel errors is to cause errors in the received indices. A parallel genetic algorithm is applied to assign the codevector indices for noisy channels so as to minimize the distortion due to bit errors. The novel property of multiple global optima and the average distortion of the memoryless binary symmetric channel for any bit error in the assignment of codebook index are also introduced.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.660304  DOI: Not available
Share: