Please use this identifier to cite or link to this item:
http://idr.nitk.ac.in/jspui/handle/123456789/9977
Title: | 2D-spectral estimation based on DCT and modified magnitude group delay |
Authors: | Sandeep, P. Shreyamsha, Kumar, B.K. Narasimhan, S.V. |
Issue Date: | 2013 |
Citation: | Signal, Image and Video Processing, 2013, Vol.7, 5, pp.959-973 |
Abstract: | This paper proposes two new 2D-spectral estimation methods. The 2D-modified magnitude group delay (MMGD) is applied to 2D-discrete Fourier transform (2D-DFT) for the first and to the analytic 2D-discrete Cosine transform for the second. The analytic 2D-DCT preserves the desirable properties of the DCT (like, improved frequency resolution, leakage and detectability) and is realized by a 2D-discrete cosine transform (2D-DCT) and its Hilbert transform. The 2D-MMGD is an extension from 1D to 2D, and it reduces the variance preserving the original frequency resolution of 2D-DFT or 2D-analytic DCT, depending upon to which is applied. The first and the second methods are referred to as DFT-MMGD and DCT-MMGD, respectively. The proposed methods are applied to 2D sinusoids and 2D AR process, associated with Gaussian white noise. The performance of the DCT-MMGD is found to be superior to that of DFT-MMGD in terms of variance, frequency resolution and detectability. The performance of DFT-MMGD and DCT-MMGD is better than that of 2D-LP method even when the signal to noise ratio is low. 2012 Springer-Verlag London Limited. |
URI: | 10.1007/s11760-011-0286-9 http://idr.nitk.ac.in/jspui/handle/123456789/9977 |
Appears in Collections: | 1. Journal Articles |
Files in This Item:
There are no files associated with this item.
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.