Flame Detection for Video-Based Early Fire Warning for the Protection of Cultural Heritage

Abstract

Abstract. Cultural heritage and archaeological sites are exposed to the risk of fire and early warning is the only way to avoid losses and damages. The use of terrestrial systems, typically based on video cameras, is currently the most promising solution for advanced automatic wildfire surveillance and monitoring. Video cameras are sensitive in visible spectra and can be used either for flame or smoke detection. This paper presents and compares three video-based flame detection techniques, which were developed within the FIRESENSE EU research project.

A Review on Early Forest Fire Detection Systems Using Optical Remote Sensing

“…Detection methods that use optical sensors or RGB cameras combine features that are related to the physical properties of flame and smoke, such as color, motion, spectral, spatial, temporal, and texture characteristics. The following color spaces have been used for the task of early fire detection: RGB [ 17 , 18 , 19 ], YCbCr [ 20 ], CIELAB [ 21 ], YUV [ 22 , 23 ], and HSV [ 24 ]; however, a drawback of color-based fire detection models is the high false alarm rates, since single-color information is insufficient in most cases for the early and robust fire detection. Thus, many of the developed methodologies combine color and motion information in images and videos [ 25 ].…”

Spatio-Temporal Flame Modeling and Dynamic Texture Analysis for Automatic Video-Based Fire Detection

“…To examine the proposed fire-flame detection algorithm, we estimated the number of correctly detected flame frames out of the total number of flame frames (true positive) and the number of non-flame frames that erroneously recognized as flame frames out of the total number of non-flame frames (false positive). For comparison reasons, a frame is labeled as a fire frame if it contains at least one fire block, as in [13]. The average frame rate of the proposed method was 5.2 fps for video sequences with resolution 320x240, which is considered adequate for an early fire warning system.…”

“…Both datasets have been used in the past for the comparison of flame detection algorithms. Specifically, three state of the art algorithms [13] have been tested with the first dataset, while three other [5][11] [12] with the second dataset. To examine the proposed fire-flame detection algorithm, we estimated the number of correctly detected flame frames out of the total number of flame frames (true positive) and the number of non-flame frames that erroneously recognized as flame frames out of the total number of non-flame frames (false positive).…”

“…More recently, within the FP7 EU-funded Firesense project [13] various flame detection algorithms were developed. More specifically, Habiboglu et al [14], proposed a video-based fire detection system, which uses color, spatial and temporal information.…”

“…12). The performance of the proposed method was compared with three existing flame detection algorithms that were tested with this dataset in [13]: i) Correlation descriptors [14], ii) Multiple features and rulebased classification [15], and iii) Spatio-temporal consistency energy [27]. As shown in Fig.…”