M.Sc Student | Shapiro David |
---|---|
Subject | Generalization and Efficient Realization of the HZ Universal Data Compression Algorithm |
Department | Department of Electrical Engineering | Supervisor | Research Professor E Jacob Ziv |
A compression algorithm
with limited memory was suggested by Prof. Jacob Ziv and Yehuda Hershkovits (HZ
algorithm), based on context matching. It was later shown by Ziv that the HZ
approaches general universal lower bound on universal compression with limited
training data.
The algorithm presents
efficient compression scheme for moderate as well as long sequences based on
context matching. An efficient implementation of this algorithm was performed
by Zeev Cohen using block coding length of one symbol.
This work deals with
further research regarding the implementation of HZ algorithm and relies on
work performed by Zeev Cohen. While prior works deal with single symbol coding,
this work investigates the influence of encoded block length on compression
results. These work also describes existing vastly used schemes emphasis on
algorithmic, functional and performance comparison of these schemes.
A new compression scheme
is suggested incorporating combined one symbol and two symbol block coding. The
data structure and simulation codes are based on Zeev Cohen work and allow
coding and decoding using HZ algorithm in relatively small number of
operations.