摘要
在智能防火的研究中,通过远程视频对火焰进行识别是一个新的研究方向。远程视频与高清晰的采集图像不同,由于采集的火焰特征空间发生改变,导致代表火焰可以识别的动态特征发生明显变化。传统的算法进行火焰的识别,一旦缺少火焰的可识别动态特征,无法对火焰图像进行颜色特征提取,导致远程视频火焰的识别效率差、精度低。提出采用多特征融合算法的远程视频火焰的识别方法。先依据在燃烧过程中火焰具有的物理特征来描述火焰图像识别区域的封闭轮廓线集合,并与火焰的动态特征相结合来建立火焰识别的概率模型,通过提取火焰图像的颜色特征计算火焰图像的颜色特征向量,并利用火焰图像的颜色特征向量间的距离进行远程视频火焰图像匹配,进而精确地实现对远程视频火焰的识别。仿真证明,基于多特征融合算法的远程视频火焰的识别方法识别精确度高,具有较好的适应性。
A remote video flame identification method wass proposed based on multi-feature fusion algorithm. According to the physical characteristics of the flame in the process of combustion, a closed contour line set in flame image identification area was described and combined with the dynamic characteristics of the flame to establish the probability model of flame identification. Color feature vector of flame image was calculated by extracting the color features of flame image, and the distance between the color feature vectors of flame images was used to match the remote video flame image, and then to realize the accurate identification of the remote video flame. The simulation results show that the recognition method of remote video flame based on multi feature fusion algorithm has high accuracy and good adaptability.
出处
《计算机仿真》
CSCD
北大核心
2016年第7期355-358,共4页
Computer Simulation
基金
濮阳职业技术学院自然科学研究项目(2015PYZYZR09)
关键词
颜色特征
远程视频
火焰识别
Color feature
Remote video
Flame identification
