September 1997, Volume 6, Number 3
20 years of radiosonde data from the aerological station Prague-Libus were analysed to evaluate integrated values of water vapour content and cloud water content. Both parameters are proportional to the attenuation in microwave frequency bands. The soundings from 00, 06, 12, and 18 UTC were taken into account to discuss diurnal as well as seasonal variations. The methods available to compute the cloud and the water vapour attenuation are presented.
Various cases of electromagnetic wave propagation have been solved such as two obstacles with the sharp edges or the spherical earth especially for low altitude propagation considering several combinations of antenna and observation point heights. Computation results and published solutions for individual special cases agree quite well. That demonstrates the usefulness of the method.
Spectrum estimation belongs to the most frequent problems solved by the digital stochastic signal processing. The application of the unconventional method of spectrum estimation based on maximum entropy approach introduced by J. P. Burg is presented. Drawbacks of classical spectrum estimation methods involving interpretation difficulties are avoided. Basic principles and underlying conceptions yielding the same results as autoregressive modelling and reasonableness of its application are discussed. Stochastic biosignal reflecting fetus movements obtained by the tocography is analyzed by the maximum entropy method and compared to the conventional method of Fourier analysis.
The paper is devoted to the development of the analog electronic switch for video signals. The simple linear equivalent diagram of the integrated circuit CMOS 4066 on the basis of measurements its transfer characteristics was designed. Consequently the connection of the double switch was designed and experimentally verified. Its properties (parameters) (especially the attenuation in the transmission and suppressed band, non-linear distortion, frequency band etc.) are quite suitable for video signals processing.
In this paper the dependence of PSNR on the size of both a codebook pool and range blocks is compared. The statistical properties of the values of transformation coefficients and distances between coded range blocks and optimal domain blocks in the fractal image block coding scheme are discussed.
An artificial neural network (NN) for displacement vectors (DV) determination is presented in this paper. DV are computed in areas which are essential for image analysis and computer vision, in areas where are edges, lines, corners etc. These special features are found by edges operators with the following filtration. The filtration is performed by a threshold function. The next step is DV computation by 2D Hamming artificial neural network. A method of DV computation is based on the full search block matching algorithms. The pre-processing (edges finding) is the reason why the correlation function is very simple, the process of DV determination needs less computation and the structure of the NN is simpler.