Scientific and methodological foundations have been developed for the optimal identification of non-stationary objects transmitted by random time series in control systems of industrial and technological complexes, monitoring of environmental protection and ecology, palynology, medicine based on the use of statistical, dynamic, neural network models and mechanisms for extracting statistical, dynamic specific characteristics of information. The mechanisms for controlling the error in the identification of RTS, based on the use of statistical criteria - the rules of the sequential analysis, Bayesian estimation, ±3σ, confidence interval, have been investigated. The problem of control of the error in the identification of RTS by the confidence interval, the mechanisms of threshold control of the values of the random time series elements, their increments, the error of prediction by statistical, trend, dynamic models, algebraic polynomials, parabolic, cubic interpolation, extrapolation spline functions. Mechanisms for regulating the variables of the random time series identification models based on autoregressive, adaptive smoothing by R. Brown, Newton, Lagrange polynomials, orthogonal algebraic polynomials of 3, 5, 7 orders have been developed. The methodology for assessing the quality of the identification of RTS by mathematical expectation and variance and the criterion of correspondence between the relations of mathematical expectation and variance of two consecutive elements have been implemented. The experimental study was carried out according to the real data of the power grid enterprise. The effectiveness of the implemented mechanisms was studied according to the criteria of the minimum root-mean-square error, labor intensity, cost of information processing.

Optimization of identification of micro-objects with error control and statistical image filtering mechanisms

Constructive approaches, principles, and models for optimizing the identification of micro-objects have been developed based on the use of combined statistical, dynamic models and neural networks with mechanisms for filtering noise and foreign particles of images of medical objects and pollen grains. Algorithms for learning neural networks under conditions of a priori insufficiency, uncertainty of parameters, and low accuracy of data processing are investigated. The mechanisms of contour selection, segmentation, obtaining the boundaries of segments with hard and soft thresholds, filtering using the morphological features of the image have been developed [1]. Mechanisms for recognition and classification of images, adaptation of parameter values, tuning of the network structure, approximation and smoothing of random emissions, bursts in the image contour are proposed. A mechanism for suppressing impulse noise and noise is implemented based on various filtering methods, preserving the boundaries of objects and small-sized parts. Mathematical expressions are obtained for estimating the identification errors caused by nonstationarity, inadequacy of approximation, interpolation, and extrapolation of the image contour. A software package for the recognition and classification of micro-objects has been developed. The results were obtained for correct, incorrect recognition, as well as rejected pollen samples, which were synthesized with cubic, biquadratic, interpolation spline-functions and wavelet transforms.

/pdf/improving-the-quality-of-identification-and-filtering-of-4yd8uh3vc1.pdf

Improving the quality of identification and filtering of micro-object images based on neural networks

A methodology has been developed for optimizing the identification and processing of images of pollen grains based on the use of mechanisms for extracting texture, specific characteristics, and geometric features of micro-objects. A technique is proposed for the trend description of the contour according to the characteristics of the growth of a dynamic curve based on a wide range of functional dependencies. Extrapolation mechanisms have been developed for exponential, autoregressive, moving average, adaptive smoothing, statistical models of forecasting and linking of reference points of the image contour. The principles of masking reference points, pixel representation, adjusting their width, modifying biquadratic interpolation, as well as mechanisms for segmenting and detecting changes in the dynamics of the contour curve are proposed. The mechanisms for describing a contour curve by an interpolation spline - the Daubechies function, 4, 8 orthogonal polynomials in order to identify the characteristics of pollen grains - are investigated. A set of programs has been implemented that focuses on the application of a mechanism for reducing zero points, regulating the beginning, center, segment boundaries, mask width of the reference point, and color-brightness picture. The program modules of the complex are created in C++ and tested in the «CUDA parallel computing environment.

Methodology of Optimization of Identification of the Contour and Brightness-Color Picture of Images of Micro-Objects

A methodology for optimizing the identification, recognition and classification of micro-objects has been implemented using dynamic models for transforming the original image, synthesizing mechanisms for extracting redundant information structures and using histological, morphological, and texture characteristics of images. Implemented mechanisms for detecting, highlighting and correcting the values of distorted points by embedding new data, segmenting the contour, reducing redundant fragments, and setting variables. Identification algorithms have been developed that take into account the correlations of a set of points, the dynamics of their change, matrices of coordinates of points, deformation of segments and the selection of homogeneous areas of images. A comparative analysis of the effectiveness of the mechanisms for pre-processing images, recognition and classification based on testing pictures of medical diagnostics, and pollen grains in solving problems of selection and seed production of grain was carried out. Modified component schemes, and learning algorithms for a three-layer neural network. A generalized mechanism of identification is implemented in case of non-linearity of connections "inputs and outputs" under conditions of a priori insufficiency and uncertainty. Five types of neural network software modules were tested with supervised and unsupervised learning algorithms based on network learning with modified samples with vector quantization, clustering, and identification with "sliding windows". The efficiency of identification with filtrations in the presence of "noise", approximation of the reference points of the curve of the contour of images of micro-objects has been studied. 

Optimization of Micro-Object Identification by Correcting Distorted Image Points

Scientific and methodological foundations of identification, recognition, classification of micro-objects using redundant information structures - morphometric, histological, fractal characteristics of images have been developed. Mechanisms for extracting statistical, dynamic, specific characteristics and rarefaction of redundant information structures are proposed. Dynamic models have been developed, combined with the capabilities of neural networks. Computational schemes for preliminary processing of images, texture, segmentation, filtering, approximation, regulation of variable values and optimization have been developed. Methods of recognition and classification of micro-objects with tools for obtaining images from a photo, video camera, digital microscope, interactive measurement, counting, structure determination, statistical analysis, isolation and segmentation, and the formation of informative (reference) points of image contours have been investigated. The traditional computational schemes of a multilayer neural network, a Kohonen neural network, and a radial-basic network are investigated. Modified algorithms for training a network with mechanisms for adjusting the values of variables, monitoring errors along the boundaries of permissible values, accounting for stationary, quasi-stationary and non-stationary behavior of image points during the formation of training samples have been developed. Generalized algorithms for the identification of images of pollen grains are proposed. The efficiency of the algorithms was investigated according to the criteria of the root mean square error and the speed of information processing. A software package for visualization, recognition, classification of images of pollen grains has been developed, the implementations of which have been tested taking into account the conditions of a priori insufficiency, uncertainty and nonstationarity of processes.

Optimization of recognition of micro-objects based on reducing excessive information structures of images

A methodology has been developed for optimizing the identification of micro-objects based on the use of neural networks (NN) of various topologies, synthesis of image processing mechanisms, extracting statistical, dynamic, specific characteristics, selecting and segmenting a contour, selecting reference points and reducing redundant points, taking into account systematic error factors, choosing an adequate model, setting variables and optimization. Methods and algorithms for determined and multivariate analysis, obtaining the coefficients of influence and elasticity of factors, approximating the contours represented by time series are proposed. Modified component schemes of the NN, training algorithms, developed a software package (SP) for visualization, recognition, classification of images of pollen grains, implemented a hybrid identification model taking into account the non-linearity of the effects of factors under the condition of a priori insufficiency and uncertainty of parameters. The efficiency of the SP was studied on the basis of a three-layer NN of forward and backward propagation of errors, learning algorithms with and without a teacher, Kohonen network with procedures for vector quantization, clustering and segmentation and the formation of a "sliding windows". The results of image identification in the presence of "noise", optimization based on filtering systematic error and NN extrapolation of the trend of the contour curve of the images of pollen grains were obtained.

Optimization of Identification of Images of Micro-Objects Taking Into Account Systematic Error Based on Neural Networks

Researched and developed mechanisms for optimizing the identification of random time series based on mechanisms for searching for unknown knowledge, hidden properties, patterns, relationships, features of non-stationary objects under the condition of limited a priori information, uncertainty, non-stationary processes. A generalized model for optimizing the identification of RTS based on the use of neural networks, neuro-fuzzy networks of dynamic models, as well as fuzzy logic algorithms is implemented. Instruments for data reliability control are obtained based on statistical, dynamic, intelligent approaches suitable for the conditions of transmission of incomplete, heterogeneous, partially given information with large parametric uncertainty. The principles and methods of data reliability control in a fuzzy environment are developed based on the generalization and use of the properties of neural networks, fuzzy logic and statistical modeling methods. Algorithms for searching for correlations, trends, interrelations and regularities in data, forming training, control and test sets for solving problems of recognition, classifying images of micro-objects and predicting power supply indicators are synthesized. The results of the research are implemented in the form of a software package for identifying non-stationary objects, which ensures the adaptability of model variables, high data processing performance and accuracy of results.

/pdf/improving-the-accuracy-of-identification-of-non-stationary-3dnv8y7n.pdf

Improving the Accuracy of Identification of Non-Stationary Objects Based on the Regulation of Model Variables

The problem of visualization, recognition, classification of images of micro-objects, in particular, pollen grains, unicellular organisms, fingerprints based on the definition of their variety, belonging to a class, the use of information of geometric shapes, morphology, dynamic, specific characteristics, unique features of neural networks has been investigated, in control systems of industrial and technological complexes, environmental monitoring, ecology, and medical diagnoses. Methods, learning algorithms, component computational schemes of neural networks have been developed, which provide the best quality of image identification in conditions of a priori insufficiency, uncertainty of parameters, and low accuracy of data processing. Mathematical expressions are obtained for estimating identification errors associated with information distortions at the measurement, input, and transmission stages due to nonstationarity, the inadequacy of approximation, interpolation, and extrapolation of the image contour. A software package for the recognition and classification of pollen grains has been built and implemented, which includes algorithms for a three-layer, loosely coupled neural network, Hopfield’s network, bidirectional associative memory, Kohonen. Results are obtained for correct, incorrect recognition, and rejected pollen samples based on with-teacher and unsupervised learning algorithms, which are synthesized with cubic, biquadratic, and interpolation spline functions.

Mechanisms For Using Image Properties And Neural Networks In Identification Of Micro-Objects

Scientific and methodological foundations for the identification of images of unknown objects with the allocation of local areas of objects of interest, the use of traditional Gaussian, median filtering, fast Fourier transform, cosine transform, wavelet transforms, shift transforms, segmentation, factorization based on the use of redundancy of information structural components and their features. Mechanisms for detecting and correcting distorted points in fragment pictures have been developed based on the use of statistical, dynamic and neural network models. A wide range of mechanisms for filtering, segmentation, factorization, recognition and classification of images of non-stationary objects in conditions of a priori insufficiency, parametric uncertainty, and low reliability of information have been investigated and implemented. Operators of wavelet, shift and other transformations, color coding, visualization of objects with regulation of their formalized parameters are investigated. Tested realizations of the generalized mechanism of identification, recognition and classification of pictures of a human brain tumor have been tested, in particular, unicellular elements of the brain with and without pathology have been investigated. A software package for visualization, recognition, classification of images of unknown objects has been developed, the are carried out of which are carried out in conditions of a priori insufficiency, parametric uncertainty, and low reliability of information.

Olimjon I. Djumanov

Papers

Optimization of Identification of Images of Micro-Objects Taking Into Account Systematic Error Based on Neural Networks

Methodology of Optimization of Identification of the Contour and Brightness-Color Picture of Images of Micro-Objects

Improving the Accuracy of Identification of Non-Stationary Objects Based on the Regulation of Model Variables

Mechanisms For Using Image Properties And Neural Networks In Identification Of Micro-Objects

Detection of Distorted Points on Images of Micro-Objects Based on The Properties and Peculiarities of the Wavelet - Transformation