大空间图像型火焰检测方法研究

Research on large space image flame detection method

为了提高图像型火焰检测算法的准确率,满足其对实时性的要求,采用三帧差分和背景更新相结合的方法,提取运动前景区域。然后在RGB空间和Lab空间建立颜色模型,分割出火焰疑似区域。用Lucas-Kanade稀疏光流算法跟踪运动区域,获取火焰的主运动方向作为火焰识别特征,判断是否是真实的火灾发生。实验结果表明,该算法具有较好的实时性、鲁棒性,能够有效地提高火焰识别的准确率,降低误检率,在大空间公共建筑消防系统中具有重要的应用价值。

In order to improve the accuracy of the image flame detection algorithm and meet its real-time requirement,the three-frame difference and background update combining method is adopted to extract the motion foreground region.The color model is established in the RGB space and Lab space to segment the suspected flame region.The Lucas-Kanade sparse optical flow algorithm is used to track the motion region,and obtain the main motion direction of the flame which is taken as the flame recognition feature to determine whether any real fire occurs.The experimental results show that the algorithm has a good realtime performance and robustness,can effectively improve the accuracy of flame recognition,reduce false detection rate,and has an important application value in the large space public building fire-proof system.