Summary: Channel Optimized Quantization of Images
over Bursty Channels \Lambda
Julian Cheng Fady Alajaji
Mathematics and Engineering
Dept. of Mathematics and Statistics
Queen's University, Kingston, Ontario, Canada K7L 3N6
Abstract --- A DCT­based combined source­channel
coding system is proposed for transmitting grey­level
images over a binary channel with additive Markov
noise. It consists of a channel optimized quantiza­
tion scheme that exploits the channel memory by in­
corporating the characteristics of the correlated noise
in the quantizer design. Experimental results show
that this simple system -- which employs a fixed zonal
coding bit allocation technique -- provides significant
improvements over traditional tandem systems, espe­
cially during bad channel conditions. The loss of op­
timality due to the use of the zonal coding bit allo­
cation method is also examined. The loss is shown
to be small for various images; this suggests that a

Source: Alajaji, Fady - Department of Mathematics and Statistics, Queen's University (Kingston)
Linder, Tamás - Department of Mathematics and Statistics, Queen's University (Kingston)

Collections: Engineering