基于视频的火焰检测技术研究.pdfVIP

第l7卷 第1期 云南民族大学学报(自然科学版) V01．17 No．1 2008年1月 Journal of Yunnan Nationalities University(Natural Sciences Edition) Jan．2OO8 基于视频的火焰检测技术研究 高爱莲 刘 辉 林 宏 王汉友 (1．昆明理工大学信息工程与自动化学院，昆明650051； 2．西南林学院计算机与信息科学系，昆明650224；3．昆明理工大学机电学院，昆明650093) 摘 要 综合分析了火焰的颜色、闪烁频率及其面积增长性等静态特征和动态特征，设计了一种基于视频的火焰检测自 动预警算法．利用混合高斯模型有效地表示了火焰颜色及颜色的多态性．实验表明该算法具有误报警低、识别率高、实时性强 等优点，是火灾自动报警的一种可行、科学的解决方案，具有广泛的应用前景，该研究在国内外还处于发展阶段． 关键词 火焰检测；混合高斯模型；静态特性；动态特性 【中图分类号|TP391．4 【文献标识码】B 【文章编号2008)01—0094—03 Flame—．Detection Technique Based on Video Processing Gao Ailian Liu Hui Lin Hong2 Wang Hanyou (1．School of Information Engineering and Automation，Kunming University of Science and Technology，Kunming 650051，China；2．Department of Computer and Information Science， Southwest Fores University，Kunming 650224，China；3．School of Mechanical and Electrical En~neering，Kunming University of Science and Technology，Kunming 650093，China) Abstract：The paper analyzes smtic characteristics and dynamic ones of flam e such as flame’s colors、flicker- ing frequency and augmentation．puts forward a new flam e—detection algorithm that could automatically give an a． 1arm on video processing．Flam e’s color and its polymorphism were described effectively by mixture Gauss mode1． The experimental results show that the flam e—-detection algorithm has many advantages such as lower false alarm rate，hisher recognition rate and it can be used in many real—time control environments．It is an efficient and sci— entitle solution，and it has widely applied prospect．The research of flame—detection is stiU in the early stage in China． Key words：flam e—detection；mixture Gaussian model；static characteristics；dynam ic characteristics