Title :
Multiple description lattice vector quantization using Coinciding similar lattices of A4
Author :
Akhtarkavan, Ehsan ; Salleh, M.F.M.
Author_Institution :
Electr. & Electron. Eng., Univ. Sains Malaysia (USM), Nibong Tebal, Malaysia
Abstract :
Lattice Vector Quantization (LVQ) has been very popular in image compression since it needs lower computations and offers high performance. This paper presents a new multiple A4 lattice based coinciding quantization scheme for multiple description coding (MDLVQ - A4). The Coinciding similar A4 sublattices are used to quantize the input stream for multiple description coding system. The use of multiple coinciding quantizers eliminates the need for labeling function as in the traditional schemes. The MDLVQ - A4 scheme is applied to image coding to evaluate the performance. Experimental results of the MDLVQ - A4 scheme show a higher performance in terms of PSNR of the side decoders as compared to the renowned techniques of wavelet image coding and optimized wavelet image coding for several test images.
Keywords :
image coding; vector quantisation; wavelet transforms; A4 lattice; coinciding quantization; coinciding similar lattices; image compression; multiple description coding; multiple description lattice vector quantization; wavelet image coding; Correlation; Decoding; Generators; Image coding; Labeling; Lattices; Vector quantization; A4; Multiple Description (MD) Coding; Multiple Description Lattice Vector Quantization (MDLVQ); Wavelet transforms;
Conference_Titel :
Industrial Electronics & Applications (ISIEA), 2010 IEEE Symposium on
Conference_Location :
Penang
Print_ISBN :
978-1-4244-7645-9
DOI :
10.1109/ISIEA.2010.5679371