On Computational Aspects of Tchebichef Polynomials for Higher Polynomial Order

Tchebichef polynomials (TPs) and their moments are widely used in signal processing due to their remarkable performance in signal analysis, feature extraction, and compression capability. The common problem of the TP is that the coefficients computation is prone to numerical instabilities when the p...

Full description

Saved in:
Bibliographic Details
Published in:IEEE access Vol. 5; pp. 2470 - 2478
Main Authors: Abdulhussain, Sadiq H., Ramli, Abd Rahman, Al-Haddad, Syed Abdul Rahman, Mahmmod, Basheera M., Jassim, Wissam A.
Format: Journal Article
Language:English
Published: Piscataway IEEE 2017
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Subjects:
Algorithm design and analysis
Algorithms
Coefficients
Compression tests
Computation
Computers
Discrete cosine transform
Discrete cosine transforms
Electronic mail
Feature extraction
Image reconstruction
Indexes
numerical propagations errors
Orthogonal polynomials
Polynomials
recurrence algorithm
Signal analysis
Signal processing
Signal processing algorithms
Speech
Speech compression
Tchebichef polynomials
ISSN:2169-3536, 2169-3536
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Tchebichef polynomials (TPs) and their moments are widely used in signal processing due to their remarkable performance in signal analysis, feature extraction, and compression capability. The common problem of the TP is that the coefficients computation is prone to numerical instabilities when the polynomial order becomes large. In this paper, a new algorithm is proposed to compute the TP coefficients (TPCs) for higher polynomial order by combining two existing recurrence algorithms: the three-term recurrence relations in the n-direction and x-direction. First, the TPCs are computed for x, n = 0, 1, ... , (N/2) - 1 using the recurrence in the x-direction. Second, the TPCs for x = 0, 1, ... , (N/2) - 1 and n = (N/2), (N/2) + 1, ... , N - 1 based on n and x directions are calculated. Finally, the symmetry condition is applied to calculate the rest of the coefficients for x = (N/2), (N/2) + 1, ... , N - 1. In addition to the ability of the proposed algorithm to reduce the numerical propagation errors, it also accelerates the computational speed of the TPCs. The performance of the proposed algorithm was compared to that of existing algorithms for the reconstruction of speech and image signals taken from different databases. The performance of the TPCs computed by the proposed algorithm was also compared with the performance of the discrete cosine transform coefficients for speech compression systems. Different types of speech quality measures were used for evaluation. According to the results of the comparative analysis, the proposed algorithm makes the computation of the TP superior to that of conventional recurrence algorithms when the polynomial order is large.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:2169-3536
2169-3536
DOI:10.1109/ACCESS.2017.2669218