Flame detection method and application based on RGB-HSI model and initial flame growth characteristics

Abstract

Abstract The safe and stable operation of hydraulic machinery equipment cannot be separated from various auxiliary electrical equipments and lubricating oils. And these are important factors in the occurrence of fire. Fire detection is used to detect fire hazards during the operation of hydraulic machinery equipment, which is of great significance to ensuring safety and stable operation of hydraulic machinery equipment and reducing economic losses. Traditional fire detection methods require the detector to be put close to the fire source, and the detector is also prone to interference, resulting in delayed detection, false alarm and missed alarm. To solve these problems, a flame detection method based on RGB-HSI model and initial flame growth characteristics is proposed in this paper. Firstly, after image processing such as binarization, filtering and dilation, the connected region of pixels is divided and the total amount of flame pixels is calculated. Finally, the region where the growth ratio of flame pixels exceeds the growth ratio threshold for three consecutive calculations is marked and alerted. The experiment proves that the real-time and accurate identification of the initial flame can be realized by combining the above image processing technology with the initial fire growth characteristics.