PARAMETER ESTIMATION II

Chair: R. Kumaresan, The University of Rhode Island (USA)

Home


Parameter Selection for a Markovian Signal Reconstruction with Edge Detection

Authors:

Mila Nikolova, CNRS-Supelec-UPS (FRANCE)

Volume 3, Page 1804

Abstract:

We propose a method for the parameter selection for a Bayesian reconstruction of 1D or 2D signals, constituted by locally homogeneous regions, from incomplete and noisy projection data. A piecewise Gaussian Markov model (PG MM), defined as a sum of truncated quadratic potential functions, is used to regularise the reconstruction, which is otherwise ill-posed. This model is called the weak string in 1D and the weak membrane in 2D [1]. The posterior energy is highly non-convex and the MAP estimator is piecewise continuous; the model parameters play a particularly decisive role. The resolution of the reconstruction -- the finest recoverable features -- is determined jointly by the parameters and the observation model. On the other hand, we propose a method for the determination of the parameters in order to reach, or at least to approach as closely as possible, a desired resolution. This model needs the evaluation of several posterior edge detection thresholds.

300dpi TIFF Images of pages:

1804 1805 1806 1807

Acrobat PDF file of whole paper:

ic951804.pdf

TOP



Estimation of Phase for Noisy Linear Phase Signals

Authors:

James A. Cadzow, Vanderbilt University (USA)
Ramakrishna Kakarala, University of Auckland (NEW ZEALAND)

Volume 3, Page 1808

Abstract:

The concept of symmetry plays an important role in the theory and applications of digital signal processing. As is well known, the Fourier transform of any symmetric signal has a linear phase spectrum. To facilitate the analysis of symmetry, a procedure for estimating the parameters associated with a linear phase signal is developed. When the data being modelled is composed of a linear phase signal corrupted by additive Gaussian noise the approach taken results in maximum-likelihood estimates of the linear phase parameters. These estimates are useful for detecting and estimating the presence of symmetry in both one and two dimensions. The effectiveness of the estimates is tested on both synthetic and real images.

300dpi TIFF Images of pages:

1808 1809 1810 1811

Acrobat PDF file of whole paper:

ic951808.pdf

TOP



Three Techniques for State Order Estimation of Hidden Markov Models

Authors:

Predrag Pucar, Linkoping University (SWEDEN)
Mille Millnert, Linkoping University (SWEDEN)

Volume 3, Page 1812

Abstract:

In this contribution three examples of techniques that can be used for state order estimation of hidden Markov models are given. The methods are also exemplified using real laser range data, and the computational burden of the three methods is discussed. Two techniques, Maximum Description Length and Maximum a Posteriori Estimate, are shown to be very similar under certain circumstances. The third technique, Predictive Least Squares, is novel in this context.

300dpi TIFF Images of pages:

1812 1813 1814 1815

Acrobat PDF file of whole paper:

ic951812.pdf

TOP



On Quantization and Its Impact on the Exact Recovery of High Order Moments

Authors:

L. Cheded, King Fahd University of Petroleum & Minerals (SAUDI ARABIA)

Volume 3, Page 1816

Abstract:

This paper addresses the problem of the exact recovery of unquantized moments from their quantized counterparts. A brief review of amplitude quantization and its impact on the Exact Moment Recovery (EMR) problem is given. In particular, a special class of order p, called Lp, for which EMR is always achieved regardless of the quantization fineness used, is introduced together with some new results on its properties. Due to the tremendous practical gains that can accrue from the use of 1-bit quantized members of L1, it is shown how to force any signal to become a member of this class, hence naturally re-discovering the dithered quantization process. Two approaches to the EMR problem and some simulation results which are in very good agreement with the theory, are presented.

300dpi TIFF Images of pages:

1816 1817 1818 1819

Acrobat PDF file of whole paper:

ic951816.pdf

TOP



Radar Clutter Classification Using Autoregressive Modelling, K-Distribution, and Neural Networks

Authors:

C. Bouvier, DCN-STSN/CTSN
L. Martinet, DCN-STSN/CTSN
G. Favier, CNRS/UNSA
H. Sedano, (FRANCE)
M. Artaud, (FRANCE)

Volume 3, Page 1820

Abstract:

DCN-STSN/CTSN- Centre Technique des systemes navals- Lutte Surface Air- Servicd Traitement du signal et de l'information, B.P. 28, 83800 TOULON NAVAL FRANCE 13S Laboratory, CNRS/UNSA, Bat. SPI n4, 250 rue Albert Einstein, Sophia Antipolis, F-06560 VALBONNE This paper is concerned with the classification of radar returns including sea, ground and composite clutters. We first present an analysis of radar clutter recorded data allowing to validate the K amplitude distribution and the autoregressive modelling of the spectrum. Then, we briefly describe a classifier based on a multi-layer neural network. The inputs of which are the shape parameter of the K-distribution, the magnitude and the phase of the poles and the reflection coefficients calculated by means of the Burg's or multi-segment algorithm. Experimental results are presented to illustrate the performance of the proposed classifier.

300dpi TIFF Images of pages:

1820 1821 1822 1823

Acrobat PDF file of whole paper:

ic951820.pdf

TOP