Control of an Extended Object Recognition Process in Real Time Based on Evaluation Complexity of Each Stage of Recognition
Keywords:
Pattern Recognition, Recognition Process Control, Real Time, Extended Object, Complexity of Pattern RecognitionAbstract
Purpose: Efficiency of a real-time pattern recognition system is determined by reliability and performance of the recognition
process. The aim is to develop a model of control of a recognition mode in real time researching specific classes of objects — extended
objects. Results: There has been proposed a mathematical model of pattern recognition control in real time. It is shown that based on
assessment of recognition complexity of a current section of an extended object one can make a decision on a recognition mode for the
next section. An amount of data and a velocity of a measuring device in relation to an extended object determine a recognition mode.
Recognition mode control involves changing the speed of movement and, subsequently, the volume of information incoming to a section
of an extended object in real time. Practical relevance: The research results have been tested on data from geophysical boreholes. The
described approach to control the process of pattern recognition of an extended object allows achieving optimal performance of the real
time pattern recognition system.