Flame Detection Method Based on Feature Recognition

摘要

This paper introduced technique of current flame detecting system based on the CCD camera from which the color images are transferred into a computer, then the image processing algorithm is used to determine whether there is fire in the image sequence, the monitoring method is the most important in the whole system. The initiation of flame is a slowly process in which the image characteristics are very clearly, As the shape, area, and intensity of the flame in different time, each one varies every time. The image information of flame is analyzed in this paper, the regularity is summarized in color feature and dynamic characteristics, which is the main basis for the design of the identification algorithm. The color model is established based on the analysis of the characteristics of flame color, and the dynamic characteristics of the flame are identified according to the irregularity, the similarity and the stability of the flame, so as to provide the accurate basis for the flame detection.