Index compressed vector quantisation

Vector quantisation (VQ) is a lossy data compression scheme that has gained tremendous attention because of its simpHcity and firm information-theoretic basis. This thesis introduces two groups of novel techniques to improve the performance of traditional VQ (basic VQ) schemes, such as full search or treestructured, at low bit rates. These new schemes have a simpler implementation, more computational efficiency, and lower memory requirement than other VQ schemes for low bit rate apphcations. The proposed schemes are capable of providing significant coding improvement over traditional VQ coders, and better results than predictive VQ, finite state VQ, and JPEG at low bit rates.