September 1999, Volume 8, Number 3
Two new modified types of canonical state models simulating chaotic phenomena in piecewise-linear dynamical systems are derived. Both are topologically conjugate to Class C similarly as Chua's circuit family. Their state matrix equations and corresponding integrator-based circuit models are proposed including their relations with the first elementary canonical state model. As an example the phase portraits of typical chaotic attractor are shown.
A dielelectric-loaded horn with transverse strips is analysed theoretically. The method is based on a planar periodic strip structure and is used for a circular cylindrical and uniform waveguide model of the feed. The propagation constant and the total field can be determined by approximate solution for the "dominant" Floquet mode with n = 0. Radiation pattern of the feed with small flare angle is computed as a superposition of the field components radiated from center of the aperture and from its region filled with dielectric.
In this paper some modifications of fractal image coding are presented. Proposed methods are based on correlation coefficients computing as an alternative approach to searching of similarity between blocks. The convergence speed of decoding process is faster then convergence speed of standard method. The convergence process with modified start conditions of decoding process are analysed and verified on gray scale static images too.
New simple computer-aided design formulas for the rectangular microstrip patch antennas have been developed. The cavity model is used but the more accurate models for open-end effect of microstrip lines and the effective permittivity are used. That allows increasing the calculation resonant frequency accuracy. Calculations of several cases have been compared with the conventional cavity calculations, expressions generated by curve fitting to full wave solutions and experimental values.
This paper presents experimental results of extracting features in the Radar Target Classification process using the J frequency band pulse radar. The feature extraction is based on frequency analysis methods, the discrete-time Fourier Transform (DFT) and Multiple Signal Characterisation (MUSIC), based on the detection of Doppler effect. The analysis has turned to the preference of DFT with implemented Hanning windowing function. We assumed to classify targets-vehicles into two classes, the wheeled vehicle and tracked vehicle. The results show that it is possible to classify them only while moving. The feature of the class results from a movement of moving parts of the vehicle. However, we have not found any feature to classify the wheeled and tracked vehicles while non-moving, although their engines are on.
In this contribution we present transform and neural network approaches to the interpolation of images. From transform point of view, the principles from  are modified for 1st and 2nd order interpolation. We present several new interpolation discrete orthogonal transforms. From neural network point of view, we present interpolation possibilities of multilayer perceptrons. We use various configurations of neural networks for 1st and 2nd order interpolation. The results are compared by means of tables.
Wavelet algorithms allow considerably higher compression rates compared to Fourier transform based methods. The most important field of applications of wavelet transforms is that the image is captured in few wavelet coefficients. The successful applications in compression of image or in series of images in both the space and the time dimensions. Compression algorithms exploit the multi-scale nature of the wavelet transform.