Search

The use of a machine (computer) vision system for remove probing of the Earth is considered. A survey of software and hardware methods used in computer vision systems in the processing of satellite images is presented. Methods of processing data with the use of a trained neural network are described. Examples of algorithmic processing of satellite images by means of artificial convolutional neural networks are presented. Methods of increasing the precision of recognition of satellite images are determined. Practical applications of convolutional neural networks onboard microsatellites used for remote probing of the Earth are presented.

The disperse media composed of non-spherical particles (say, dust aerosols layers, and ice crystal clouds) can appear both optically isotropic and optically anisotropic, depending on local optical characteristics of turbid medium in question and also on the orientation of particles.

The present paper is a retrospective review of the development of the approach to numerical modeling of radiation propagation in biological tissues, based on the use of finite difference methods for the solution of the radiative transfer equation (RTE) and its application to the problems of biomedical diagnostics. The advantage of finite difference methods is the possibility to obtain the solution of the direct problem of radiation propagation in turbid media, when the exact analytical solution is impossible. In turn, the possibility to solve the RTE opens wide perspectives for the solution of inverse problem and reconstruction of structural and functional characteristics of objects from the detected scattered radiation. We present a review of the finite difference methods applied to the solution of angiography problems and investigation of blood using optical biomedical diagnostic technologies.

In this paper we derive radiation balance equations for the cloudy atmosphere with account for horizontal inhomogeneity effects. It is assumed that the clear sky and cloud occupy two different half-spaces separated by a vertical plane. Numerical results for corresponding fluxes are derived using the 3D code RADUGA based upon the discrete ordinate method. It is shown how “jump” of optical properties forms shadowing or brightening at both the cloud edge and the neighboring clear sky.

The simplified M-1D algorithm (M stands for modified) for the calculation of horizontal distribution of reflected brightness coefficient in 2D regions with large homogeneous pixels is presented. This algorithm is based upon modified 3⁎N−1 1D-transport equations (where N is the number of large pixels) instead of one 2D-transport equation, usually used in such problems. The method does not rely on empiric assumptions on both optical properties of atmosphere or diffuse radiation intensity. Numerical results demonstrating the accuracy of the presented algorithm for simulating brightening and shadowing effects in a vicinity of the jump of optical properties given. The accuracy of the M-1D approximation strongly depends on the geometry and illumination conditions. However, it remains below 15% and can reach 1% for all cases studied and is strongly higher than the accuracy of usually employed independent pixel approximation. Time reduction via replacing 2D problem via modified 1D problem is about 17 times for all cases considered in this paper.

The problem of probing a nonhomogeneous medium imitating the biological tissue by an ultrashort light pulse is investigated. The pulse propagation in the medium is described by a transient transport equation. A grid algorithm for its solution based on the analytical presentation of the density of the unscattered photon flux is presented as well as a semianalytical algorithm for computing the density of the flux of singly scattered photons. Methodological calculations carried out by the grid method and by the alternative Monte Carlo method are given.

Several numerical and analytical solutions of the radiative transfer equation (RTE) were compared for plane albedo in a problem of solar light reflection by sea water. The study incorporated the simplest case—a semi-infinite one-dimensional plane—parallel absorbing and scattering homogeneous layer illuminated by a monodirectional light beam. Inelastic processes (such as Raman scattering and fluorescence), polarization and air–water surface refraction–reflection effects, were not considered. Algorithms were based on the invariant imbedding method and two different variants of the discrete ordinate method (DOM). Calculations were performed using parameters across all possible ranges (single-scattering albedo ω 0 and refracted solar zenith angle θ 1 ), but with a special emphasis on natural waters. All computations were made for two scattering phase functions, which included an almost isotropic Rayleigh phase function and strongly anisotropic double-peaked Fournier–Forand–Mobley phase function. Models were validated using quasi-single-scattering (QSSA) and exponential approximations, which represent the extreme cases of ω 0 →0 and ω 0 →1, respectively. All methods yielded relative differences within 1.8% for modeled natural waters. An analysis of plane albedo behavior resulted in the development of a new extended QSSA approximation, which when applied in conjunction with the extended Hapke approximation developed earlier, resulted in a maximum relative error of 2.7%. The study results demonstrated that for practical applications, the estimation of inherent optical properties from observed reflectance can best be achieved using an extended Hapke approximation.

The problem of calculating the brightness coefficient of sunlight reflected from the atmosphere in the presence of spatially local horizontal inhomogeneities (property discontinuity surfaces) of a medium (where one-dimensional mathematical models, traditionally used in such calculations, cannot be applied) is considered. A two-dimensional mathematical model, which makes it possible to obtain the detailed spatial distribution of the brightness coefficient in the vicinity of an inhomogeneity, has been proposed and used to obtain the characteristic distributions in a model problem. It has been proposed to use a quasione-dimensional model to effectively estimate the extension of a subregion including an inhomogeneity, outside of which multidimensional effects in the brightness coefficient can be neglected, and this model has been numerically tested.

Grid algorithms developed for parallel code RADUGA‐5.1(P) to calculate by discrete ordinate method radiation fields from sources, localized in 3‐D spatial regions, are outlined. Computation results confirming high calculation accuracy guaranteed by the algorithms are presented.

Radiative transfer through turbid media is usually modeled on the basis of the stationary radiative transfer equation (RTE). As a rule, in addition various approximations of the radiative transfer equation, such as the spherical harmonics equations or small angle approximations, are used. The spherical harmonics equations are relevant for transport problems in optically thick and weakly heterogeneous media, whereas small angle approximation works well for radiation transfer problems in media characterized by the phase functions peaked in the forward scattering direction.

The direct problem of noninvasive optical biomedical diagnostics based on laser probing is considered. Light propagation in tissue is described by the radiative transfer equation. A grid algorithm using an analytical representation of the non-scattered radiation and a semi-analytical method for calculating the single scattered radiation intensity is presented. The direct solution of the radiative transfer equation is compared with the experimental data, Monte-Carlo simulation results and the diffusion approximation solution. The proposed algorithm has advantages in terms of accuracy and lower calculation time.

Operational remote sensing of terrestrial atmosphere is heavily based on the 1-D radiative transfer equation. However, cloudy scenes are influenced by 3-D effects (e.g., illumination from cloud sides, etc.). This leads to biases in aerosol/cloud/land/ocean retrieval schemes for scenes with clouds. These biases can be understood and quantified only with the use of the 3-D radiative transfer theory, which allows to account for arbitrary spatial variation of atmospheric parameters. The task of this paper is twofold. First of all we introduce a novel technique for the solution of the 3-D radiative transfer equation based on the grid approximations and the straightforward iteration procedure realised on supercomputers with parallel architecture. We study the performance of our technique comparing with the solutions obtained by the Monte-Carlo code. A close correspondence is found. Secondly, we quantify the influence of neighbouring clouds on the clear sky reflection function at the nadir observation depending on the solar illumination conditions. We find that the influence of cloud on the clear sky reflectance function is not negligible (even outside the cloud geometrical shadow). Thus, the peculiar inner boundary layer arises in the sky reflectance function with shadowing and brightening effects.