METHOD OF GESTURE RECOGNITION FOR INTERACTIVE COMPUTER PERFORMANCE

Abstract

Topicality. As the demand for fast and convenient interaction with computers increases across various domains (from multimedia applications to industrial systems), the need for contactless control methods also grows. Gesture recognition is becoming a key solution, as it can boost responsiveness, minimize reliance on traditional input devices, and provide a natural, intuitive means of communication.. The goal of this work is to develop a gesture recognition method that operates in real time, delivers balanced performance, remains robust under varying lighting conditions, and is capable of adapting its responses depending on the context of the active application. The object of research is the process of capturing frames from a webcam and recognizing hand gestures in real time. The subject of research involves the algorithms and technologies (in particular, video stream filtering, multi-threading mechanisms, and context adaptation) aimed at increasing the effectiveness and reliability of a gesture recognition system. Results. The work analyzes several implementations featuring automatic brightness/contrast adjustment to enhance stable hand detection in changing lighting conditions, as well as multi-threaded processing to avoid interface freezes. It is demonstrated that context adaptation of gestures makes the system flexible: one gesture can perform various actions depending on the active window. Experiments confirmed real-time operation at 25–30 FPS and high recognition accuracy given proper parameter settings. The proposed webcam-based gesture recognition method has proved its effectiveness in real-life scenarios. Combining frame filtering, multi-threading, and context adaptation enables the system to function steadily, swiftly, and be easily scalable to the specific tasks of various applications.