Flame detection is an important capability for intelligent robot perception in unknown environments, which is essential for responding to devastating fire hazards in advance. This paper presents a vision based flame detection method using video compressed domain motion prediction and multi-feature pattern recognition. Firstly, the motion information in the macroblocks and motion vectors of H.264/AVC compressed domain data is fully utilized to extract the motion foreground. Secondly, the flame color conditions in the YCbCr color space are established to segment candidate flame regions. Then, the chromaticity, texture, and geometric features of a candidate flame region are extracted and combined into a multi-dimensional feature vector. Finally, the Support Vector Machine classifier is adopted to identify real flames or non-flames to improve sensitivity and stability. Experimental results of indoor and outdoor applications for flame and flameless scenes illustrate that the proposed method achieves superior performance with higher accuracy and lower false alarm rate.
Vision Based Flame Detection Using Compressed Domain Motion Prediction and Multi-Feature Fusion
Lect.Notes Computer
International Conference on Intelligent Robotics and Applications ; 2023 ; Hangzhou, China July 05, 2023 - July 07, 2023
2023-10-21
12 pages
Article/Chapter (Book)
Electronic Resource
English
Flame detection , Video processing , Motion prediction , Multi-feature fusion Computer Science , Artificial Intelligence , Software Engineering/Programming and Operating Systems , Computer Applications , Computer Communication Networks , Special Purpose and Application-Based Systems , User Interfaces and Human Computer Interaction
Flame Detection Algorithm Based on Video Multi-Feature Fusion
British Library Online Contents | 2006
|Traffic flow prediction method based on multi-feature fusion
European Patent Office | 2024
|Vision-Based Distracted Driver Detection Using a Fusion of SIFT and ORB Feature Extraction
Springer Verlag | 2023
|Fatigue Driving Detection Based on Multi Feature Fusion
IEEE | 2019
|