MODELING, ARCHITECTURE AND CYBER PHYSICAL SYSTEMS FOR COMPUTER VISION

Abstract

Many methods have been developed to identify anomalies in the images. An important area is tasks in which anomalies need to be identified when the objects are in motion. With the development of a more sophisticated exchange of infrastructure data, such as Internet things have the opportunity to obtain such real -time images with camcorders in static and dynamic modes and their further processing. Such a system can be implemented as a cyberphysical system. It would make it possible to obtain images, use of videos to collect images of both static and dynamic, as well as their processing and further clarification as needed. Intellectual computer vision systems are one of the main components of intellectual data analysis. These systems should be able to make decisions with minimal human interaction or even without any human interactions. Vision analysis systems should understand the scene by classifying and tracking objects, as well as studying their behavior. One of the most important parts of visual analysis is the detection of anomalies.

The architecture of a cyberphysical system of computer vision was developed according to the developed method. The cyberphysical system is a centralized system and controls the overall process. The agent's autonomy is limited to safety measures to ensure a safe overall process. As a result, hybrid system architecture is proposed, which is able to scale the level of autonomy for each task on the requirement and integrate expert knowledge into the system. The developed cyberphysical system was modeling its work in order to determine the possibility of application in practice. The results obtained confirm the effectiveness of the proposed decision.

The areas of further research are the improvement of the architecture of the cyberphysical system and the method of recognition.