Home – Home – Jornals – Avtometriya 2012 number 6

A solution was found to the problem of recognizing a spatial three-dimensional group point object with coordinate noise, a priori uncertainty of angular parameters, and a shift of the initial vector component that defines the object. An iterative approach to matching angular parameters and vector descriptions of the recognized and reference groups of point objects is proposed.

Correlations for the lower limit of the estimate error variance of images distorted by additive and applicative noise are obtained. The potential and actual estimate error variances for linear and nonlinear filtering algorithms are analyzed. A combined filtering algorithm for images distorted by applicative noise is developed.

A new method of abandoned object detection based on analyzing a sequence of depth images is proposed, and algorithms of real-time determination of the quasi-stationary background and abandoned objects are developed. The efficiency of the proposed algorithms is compared with that of the algorithm based on analyzing brightness images.

A method of solving two-dimensional limited-angle tomography problems, based on the Cavalieri condition for the Radon transform, is investigated. Two algorithms realizing this approach are compared between themselves and also with the algorithm calculating missing projections from approximate tomograms. Numerical simulations show that all three algorithms allow the reconstruction accuracy to be improved in the case under consideration. Each of them, however, has certain specific features, which should be taken into account in solving practical tasks.

The emergence and development of algorithms for three-dimensional reconstruction in X-ray tomography are considered in historical sequence. The data completeness condition for precise reconstruction of an object at one point is studied. It is shown that some precise algorithms are used to reconstruct part of the object from data satisfying the completeness condition for this part. Other algorithms for solving the same problem require a full set of data for the entire object.

The construction of a reliability function from the results of accelerated failure time (AFT) models is considered. The constructed AFT models are verified by analyzing a sample of residuals. The fit of the residual sample to the baseline probability distribution is tested using modified nonparametric goodness-of-fit tests. In the absence of censoring in tests, it is proposed to use previously constructed models of distribution of statistics for testing composite hypotheses. In the case of censoring of type I or II, distributions of the goodness-of-fit test statistics are found by statistical modeling.

The problem of stabilization of an unstable plant by a reduced-order single-channel controller is considered. The possibility of compensating small deviations of a double inverted pendulum on a cart by acting on the cart is studied. An algebraic method for finding the extreme locations of the poles through the construction of root simplexes is proposed. Using this, the Hurwitz stability boundaries for various reduced-order controllers are calculated. It is shown that robust stabilization of the pendulum system in the class of reduced-order controllers is impossible.

The basis sources and methods of reducing the errors of incremental photoelectric digital converters for angular measurements are analyzed. A method of constructing hybrid photoelectric converters that increase the accuracy of angular measurements is proposed. The result is achieved by simultaneous measurements of the turning angle and current radial deflections of the converter shaft axis.

The influence of diffraction on the results of computer cytophotometry has been explored through experiments and theoretical analysis. The relative error introduced by diffraction is shown to depend significantly on the optical characteristics of the object: size and absorption. In computer cytophotometry, the choice of standards having a lower DNA concentration in nuclei with large sizes is found to be favorable for measurements of the length of the genomes of various organisms. It is proved theoretically and experimentally that the systematic error due to diffraction can be accounted for without the digital processing of each micro-image involving deconvolution procedures with the point spread function of the microscope.

A short-term memory structure (three-dimensional spheroidal retinoid) is proposed for stabilization of the wide-angle vision field and realization of the stereopsis process, as well as formation and storage of the binocular visual space in observer's consciousness.

It is found that damaged regions are formed around indium contacts to p-type CdHgTe {310} heteroepitaxail layers (HELs) on GaAs substrates and the sizes of these regions depend on the temperature and time of annealing in air. It is shown experimentally that at an annealing temperature of 90 ºC, the rate of expansion of the damaged regions is about 4 µm/h, and at temperatures of 120 ºC, it is more than 25 µm/h. After 488 hours of annealing of plates of CdHgTe HELs at 60 ºC in air, the formation of damaged regions around the indium contacts to the p regions was not observed. The studies were performed on plates of p-type CdHgTe HELs on GaAs substrates whose surface was covered with SiO₂ and Si₃N₄ dielectrics (with a total thickness of about 0.15 µm), with windows where p-n junctions were generated by ion implantation of boron.

Approaches to determining the unknown operator and the diagram of the neural network in the process of its analysis for diagnostics of digital systems are considered.