September 2000, Volume 9, Number 3
J. Pospisil, J. Brzobohaty, Z. Kolka, J. Horska
New Reference State Model of the Third-Order PWL Dynamical Systems
Starting from the first elementary canonical state model, the new simple state model of the third-order dynamical systems that belong to Class C is derived. A typical property of this new model is a very simple form of its partial transformation matrix in the conditions of linear topological conjugacy, which represents the unity matrix. Complete state equations and the corresponding integrator-based block diagram of this model are shown and relation to the first canonical form is graphically illustrated.
R. Lukac, S. Marchevsky
A Neural LUM Smoother
In this paper a design of neural LUM smoother is presented. The LUM smoother distinguishes by a number of smoothing characteristics done by the filter parameter. However, the tuning parameter for smoothing is fixed for whole image. The new method realizes adaptive control of the level of smoothing by neural networks. The well-known and very popular backpropagation algorithm is used. The analysis of the proposed methods is evaluated through subjective and objective criteria and compared with the traditional LUM smoother.
Modeling of Modern Active Devices for Simulations of Analog Circuits in PSpice
Suitable models of the modern active components and functional blocks, namely new types of current conveyors, operational and transimpedance amplifiers, in several appropriate levels, using analog behavioral modeling are given in this paper.
P. Pechac, M. Klepal, K. Novotny
Novel Approach to Indoor Propagation Modeling
An indoor propagation prediction for personal communication systems is demanded for modern wireless services. There are two main general approaches for indoor propagation modeling: empirical and deterministic. Both of them have their advantages and disadvantages. Novel semi-deterministic approach to modeling of propagation of electromagnetic waves inside buildings, which combines both deterministic and empirical approaches is introduced. It is based on ray-launching technique, Monte Carlo method and statistics. The model is very fast and requires only easy to obtain inputs. The model is capable of wide-band parameters prediction as well. A description of the new approach together with firs results of the model implementation and its evaluation by measurements are presented.
Comparison of Sensitivity Properties of Selected Models of Dynamical Systems
In this paper the piecewise-linear (PWL) autonomous dynamical systems are described. The sensitivity properties for all models are calculated (analytically and numerically). We will start with circuit model of 2nd-order system, which relative eigenvalue sensitivity of characteristic polynomials with respect to all circuit parameters change is calculated. Using the cascade models we can proceed to the third- and higher-order models and their relative sensitivity we can obtain easily from the lower-order models. In the last part of this paper the sensitivity of Chua's circuit is compared with the sensitivity of the third-order elementary canonical models.
R. Lukac, C. Stupak, S. Marchevsky, L. Macekova
Order-Statistic Filters in Dynamic Image Sequences
This paper is concerned with filtering methods in dynamic image sequences corrupted by impulsive noise. Spatial filters based on order statistics are well known. However, image sequences can be considered as spatiotemporal data that is a time sequence of two-dimensional (2D) images. Thus, in many applications temporal or spatiotemporal filters achieve well performance of noise suppression. Impulse detector can improve the obtained results. Two new spatiotemporal structures of SDV detector are introduced. The analysis of various filtering methods is presented. The performance of the proposed methods is evaluated through subjective and objective criteria, and a new objective criterion was developed.