多特征融合的视频火焰检测算法研究

基于视频的火焰检测算法为解决传统感烟感温火焰检测方法受环境制约的问题提供了一条新的路径。通常的视频火焰检测算法主要利用火焰的颜色、形状、频域特征等信息来进行检测,计算较为复杂,往往不能达到实时性。文中结合火焰的颜色、运动特性以及频闪特性,提出一种简单高效的视频火焰检测方法。首先使用ViBe算法提取出视频中的运动区域作为火焰候选区域,以降低计算量,再通过火焰的颜色模型筛选出疑似火焰区域,最后根据火焰的频闪特性建立一个简单的频闪模型,进一步滤除与火焰颜色相似的非火焰运动区域。通过实验证明,该文提出的算法能够检测出不同环境下火焰的发生,且执行效果较高。

The Research of Flame Detection in Video Based-on Multi-feature Fusion

The flame detection based-on videos provide a new way for solving the problem that the conventional flame detection methods based-on detecting smoke or heat are limited by circumstance. Common video flame detection principally employ the color, shape and frequency features to detect flame. These methods are too complex to be used in real-time environment. This pa-per combines the color, movement and flicker and proposes a simple and efficient video flame detection method. Firstly use the ViBe algorithm to extract moving regions in video as candidates for reducing calculation, then select the suspects through the color model of flame, and lastly according to the flicker property build a simple model to remove the non-flame areas having color simi-lar to flame. The experiments show that the proposed method can detect flame in various circumstances and perform efficiently.