Development automated system of data processing (ASDP).
Mentor: DOCTOR OF TECHNOLOGICAL SCIENCES, PROF. Prystavka P. O.
ASDP are intended for automatically processing of observation results and can include:
- subsystems of data saving and data editing (data base);
- subsystem of visualization of the processing results;
- libraries of procedures and functions of mathematical support for information processing;
- hardware-integrated systems for collecting and processing information, in particular, unmanned systems with intelligent information analysis functions with client-server architecture.
Applied tasks that can be solved with ASDP:
- estimation of the period of active existence of highly reliable technical systems, in particular space and aviation engineering;
- estimation and forecasting of ecological state of the environment and technogenically loaded regions;
- Processing of the results of geological exploration of mineral deposits;
- hydrochemical monitoring in zones of human activity;
- medical diagnostics;
- digital image processing (filtering, scaling, compression, digital stabilization);
- video analytics for target loading cameras for unmanned systems;
- reconnaissance-search unmanned aerial systems with the possibility of automatic maintenance of the unmanned complex of found objects.
The mathematical support of the developed ASDP includes:
- classical methods of statistical analysis of data (primary, correlation, regression, factor, cluster, hypothesis testing, classification);
- methods of non-parametric estimation of observation functions;
- parametric spline estimation (distribution and regression functions);
- Time series processing and forecasting methods;
- methods of modeling systems based on the theory of Markov processes;
- methods of approximation using local polynomial splines based on B-splines;
- methods for processing digital images based on linear operators, providing real-time processing;
- video processing methods in real time: detection and tracking of moving objects, object recognition, orthorectification of aerial photography;
- Object recognition on the basis of firewall neural networks (CNN-networks) and based on the statistical approaches of Data Science;
- Intelligent filtering of visual information based on recognition methods, Deep Learning and Data Mining.