Search

This paper presents new alternative derivations of the well‐known constrained Cramér–Rao bound for estimating parameters that satisfy a set of deterministic and differentiable equality constraints. Specifically, for unbiased parameter estimation, the equality constraints are treated as pseudo‐measurements corrupted by independent zero‐mean Gaussian noise with infinitely small covariance. In this way, the desired constrained Cramér–Rao bound can be established via utilising the property that the Fisher information matrices for independent measurements are additive. This paper also provides a new simple way for deriving the constrained Cramér–Rao bound when the parameter estimate has a specified bias through exploring that the estimation error lies in the null space of the gradient matrix of the constraints and invoking the Cauchy–Schwarz inequality. [ABSTRACT FROM AUTHOR]

Elimination of mixed noise consisting of white Gaussian noise and impulsive noise is one of the important problems in image processing. In this paper, a data-dependent α-trimmed mean filter using backpropagation networks is proposed. In order to determine an appropriate value of α for each processing point, the pattern classification capability of the backpropagation networks is used. Further, by processing taking into account the edge portions, both edge preservation and mixed noise elimination are attained simultaneously. Finally, the effectiveness of the proposed method is verified by simulation. © 2003 Wiley Periodicals, Inc. Electron Comm Jpn Pt 3, 86(7): 30–40, 2003; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ecjc.10061 [ABSTRACT FROM AUTHOR]

An important problem in image processing is noise removal that preserves edges in images degraded by white Gaussian noise. In this paper, we propose a data-dependent weighted mean filter using counter propagation networks. This method achieves high-level performance in the removal of white Gaussian noise while preserving the edges. Coefficients of the filter and features to change them are adjusted and decided by using a layered neural network and self-organizing network, respectively. This allows a robust filter to be obtained. Analysis of this network enables the derivation of a simpler filter structure. Using this fact, we show an example of analysis and simplification, and obtained a filter capable of processing at high speed by fewer computations. © 2000 Scripta Technica, Electron Comm Jpn Pt 3, 84(3): 42–51, 2001 [ABSTRACT FROM AUTHOR]

Joint detection and maximum-likelihood (ML) classification of linear modulations based on observations collected over an unknown flatfading additive Gaussian noise channel is considered. It is assumed that some of the observations are subject to data failures, in which case the receiver acquires only noise. Expectation-maximisation algorithm is employed to compute the ML estimates of the unknown channel parameters, which are then substituted into the corresponding likelihood expressions to perform hypothesis testing. Numerical simulations indicate that a suboptimal classifier, which is ignorant to data failures, exhibits extremely poor performance in the presence of high failure rates. On the other hand, the proposed classifier demonstrates comparable performance with that of the clairvoyant classifier which is assumed to have a priori knowledge of the channel parameters and data failures. [ABSTRACT FROM AUTHOR]