Abstract
dc:description.abstract<p>The comparative performance of several Lapped Orthogonal Transforms(LOTs) [1][2][3][12] is studied in this thesis . Blocking effects of conventional block transforms at low bit rates are reduced with LOT, through basis functions that overlap in adjacent blocks. The block transforms of Discrete Cosine Transform (DCT), Discrete Sine Transform (DST)[4], Discrete Legendre Transform (DLT), and the Modified Hermite Transform (MHT)[16] are extended to LOT in a way that a fast algorithm for nearly optimal LOTs is derived. The Markov and two different exponential correlation models are used as the source models for the derivation of these LOTs. It is shown that the LOT theoretically outperforms the conventional block transforms for all the cases considered here. The implementations of the fast LOTs are also considered. Performance evaluation with quantization shows that the LOTs reduces blocking effect at low bit rates. The over all performance of the LOTs with the images considered here produced an improvement ip signal to noise ratio over the conventional block transform approach. It is also shown in this thesis that the LOT is very insensitive to the signal correlation models used and the initial basis functions employed. The 2-D basis functions of the LOTs is presented. The LOT is an alternative to improve the block transforms for signal coding applications. The price paid is about 30% increase in the computational complexity of the transform algorithm.</p>
Degree
thesis:*- Name thesis:degree_name
- Master of Science in Electrical Engineering - (M.S.)
- Discipline thesis:degree_discipline
- Electrical and Computer Engineering
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Wadas, Frank E.
- Contributors dc:contributor
-
- Ali N. Akansu
- Nirwan Ansari
- Zoran Siveski
Subjects
dc:subject × 4Identifiers
dc:identifier.*- Repository record dc:identifier
- https://digitalcommons.njit.edu/theses/2986