一种基于颜色不变性的视频烟雾检测方法

Video-based smoke detection method based on color invariants

文中提出一种基于颜色不变性的视频烟雾检测方法,该方法将颜色特征作为烟雾的突出特征,通过分别应用照度不变的颜色表示、基于光学增益的背景减法、颜色滤波和基于颜色不变性描述子进行烟雾识别。实验结果表明,所提出的方法能够有效地检测烟雾,对光照变化和噪声具有很好的鲁棒性,尤其适用于火灾视频监控环境。

In this paper,a video smoke detection method based on color invariance is proposed. The method uses the color feature as the prominent feature of smoke. More precisely,the proposed detector employs consecutively of an illumination invariant color representation, a photometric gain based background subtraction,color filter and a smoke identification based on invariant color descriptor. The experimental results show that the proposed method can effectively detect smoke,which is very robust to light changes and noise,especially for fire video surveillance environment.