M.Sc Thesis

M.Sc StudentKostenko Oleg
SubjectSelf Adjusted Zero Trees
DepartmentDepartment of Applied Mathematics
Supervisors PROFESSOR EMERITUS Abraham Berman
PROF. Ilan Bar-on


In this thesis we generalize the classical Zero Tree algorithm such as EZW and SPIHT to work with all decompositions and dimensions. We have proposed three algorithms, that adjust their structure dynamically to the wavelet transform steps. The Basic Self Adjusted Zero Tree that adjoins the subbands of same type into the Zero Tree. The Directed Self Adjusted Zero Tree that disallows decompositions that are incompatible with the philosophy of the Zero Tree. Finally, the Nonrecurring Self Adjusted Zero Tree that builds up decompositions that are suitable to natural signals. Furthermore, we have shown that we can construct these Zero Trees, dynamically, by applying local changes concurrently with the wavelet transform.