Рақамли технологияларнинг назарий ва амалий масалалари Volume 7 Issue 2 (2024) · pp. 98-107
Contour and Color Analysis to Improve Fire Detection Performance in Video Surveillance
Murad, Hayder
Abstract
Fire detection through CCTV systems is critical to quickly responding to fires and reducing potential hazards. In this study, we present an innovative approach to flame detection that uses shape analysis to improve the accuracy and reliability of fire detection mechanisms. Unlike traditional methods that rely solely on color characteristics to detect flames, our approach incorporates contour analysis to effectively distinguish real flames from environmental nuances. The approach includes two main algorithms: one is focused on recognizing smoke and fire based on color features, and the other specializes in recognizing smoke and fire by analyzing the shape of objects. The proposed method was experimentally validated on a diverse set of twelve videos that included different scenes such as indoor, outdoor, and day and night conditions. The experimental results show promising performance, with detection results ranging from 91.4% to 99.6%. The proposed method exhibits high detection rates while minimizing false alarms, as evidenced by low error rates for both omissions and false alarms. Thus, the results highlight the effectiveness of the introduced flame detection method in accurately detecting flames in video sequences. The integration of contour analysis enhances the system's ability to distinguish real flames from environmental factors, resulting in improved detection accuracy and reliability. This research contributes to the development of fire detection technologies in video surveillance systems, which can potentially be applied to improve safety and security in various environments.
видеонаблюдениеобнаружение пожараанализ контурараспознавание пламениvideo surveillancefire detectioncontour analysisflame detection
Metadata source: the journal's OAI-PMH archive · Sindex does not store the full text; it links to the source.