Hardware and software realization of stochastic geometry methods
in image recognition
R.G.Bukharaev, N.R.Bukharaev, V.M.Zakharov, N.N.Nurmeev,
F.I.Salimov, N.G.Fattakhov
The work studies applied stochastic geometry methods in plane images recognition. For the purpose of experimental research in that field, a software simulation system "RandomScan", demonstrating wide possibilities of the stochastic recognition methods founded on two decision, procedures parallel Bayes method and metric analysis, was developed. The system provides user optimization of recognition methods parameters, based on experimental data analyses. The results of numerous experiments series show high effectiveness of studied methods.
Apart of proper software simulation, possibilities of parallel processing and hardware realization of system blocks were also investigated.
Methodology, algorithms, hardware and software as well as rich experimental results of the project can be used in applied recognition systems design.
[Contents]
|
