METHODS AND ALGORITHMS FOR SMOKE DETECTION IN OPEN PLACES BY VIDEO SEQUENCES
DOI:
https://doi.org/10.47390/ydif-y2026v2i12/n32Keywords:
video surveillance, dynamic texture, visual signs of smoke detection, chromaticity of smoke, color segmentation, pixel intensity, smoke transparency, optical flow, Wavelet transform, SVM classifierAbstract
This scientific thesis presents a comprehensive study of methods and intellectual algorithms for the early-stage detection of smoke in open areas and large-scale territories using modern video surveillance and Computer Vision technologies. During the research, the limitations and disadvantages of traditional fire sensors under open-air conditions, continuous wind, and air currents were thoroughly analyzed. Furthermore, practical and scientific recommendations have been developed to extract chromatic (color), dynamic (motion), and texture characteristics of smoke in video streams, thereby reducing the rate of false alarms using artificial intelligence classifiers.
References
1. Babari, R. Oddiy kameralar yordamida atmosfera ko'rinishini baholash uchun modelga asoslangan yondashuv. R. Babari, N. Hautiere, E. Dumont, R. Bremond, N. Paparoditis . Atmosfera muhiti. 2011-jild. 45. 30-son. Pp. 5316–5324.
2. Barmpoutis P., Dimitropoulos K., Grammalidis N. Fazoviy-vaqtinchalik tahlil, harakatni modellashtirish va dinamik to'qimalarni tanib olish yordamida tutunni aniqlash. Yevropa signal konferensiyasi (EUSIPCO - 2014) Lissabon, Portugaliya. 2014. 1078-1082 bet.
3. Catrakis, H. J. Turbulentlikdagi shakllarning murakkabligi. HJ Catrakis, Pedimotakis. Jismoniy ko'rib chiqish xatlari.1998. 80-jild. 968-971-betlar.
4. Bilkent universiteti ma'lumotlar bazasi [Elektron resurs]. http://signal.ee.bilkent.edu.tr/VisiFire/Demo/ (kirish sanasi: 08.09.2016)
This work is licensed under a 