基于空间区域生长和模糊推理的视频烟雾检测

摘要

This paper proposes the video smoke detection method based on space region growing and fuzzy inference. It analyses the fire smoke color characteristic and diffusion characteristics, and uses the three frames method and three-channel region growing segmentation algorithm to segment the smoke suspected area of the first frame image and the second frame image. It analyses the image light flow main direction of the suspected regions, and obtains the two images primary orientation angle ratio. This paper also uses the fuzzy logic intelligent analyzer to test whether the fire has happened. Experimental results show that the detection speed is faster, and can resist the interference.%提出基于空间区域生长和模糊推理的视频烟雾检测方法.研究火灾烟雾颜色特性、扩散特征,采用三帧法与三通道区域生长分割算法分割出第1帧、第2帧图像的烟雾疑似区域,分析该疑似区域的图像光流主方向,求得2幅图像的主方向角比率,通过模糊逻辑智能分析器检测火灾是否发生.实验结果表明,该方法的检测速度较快,抗干扰能力较强.