Video Image Fire Recognition Based on Color Space and Moving Object Detection

Flame recognition based on video image is an important method for fire detection. In order to improve the accuracy of flame recognition and the applicability of complex scenes, the flame color model was improved on the basis of RGB and HSI color space models, and the flame color model with adaptive threshold values was proposed for different background spaces, which could be adapted to the extraction of suspected flame areas in different environments. The flame has motion characteristics during combustion. ViBe(Visual Background Extractor) algorithm can quickly identify moving objects, but it cannot detect moving objects quickly when the first frame of the image contains moving objects. In this paper, an improved ViBe algorithm is proposed. Frame difference method is used to build the background model through the difference of the first two frames. The method of combining three frame difference and VIBE algorithm can reduce the influence of noise. The hole in the target graph is solved through image morphology processing. Experiments show that the algorithm can identify the flame region accurately and quickly.