基于图像相关性提取的火焰振荡频率  

Flame Oscillation Frequency Based on Image Correlation

作　　者：陈志斌[1] 胡隆华[1] 霍然[1] 祝实[1] 

机构地区：[1]中国科学技术大学火灾科学国家重点实验室,合肥230027

出　　处：《燃烧科学与技术》2008年第4期367-371,共5页Journal of Combustion Science and Technology

基　　金：国家自然科学基金资助项目(50676090)

摘　　要：振荡频率是燃烧火焰的一个重要基本参数,利用Matlab图像分析工具,分析了连续火焰图像序列的相关性,并对图像相关系数随时间的变化过程进行快速傅里叶变换,从而提取出火焰的振荡频率.采用该方法分析实验获得的0.7 m×0.7 m油盘火焰的视频图像,得出火焰的振荡频率为1.80 Hz,Pagni公式的理论预测值为1.71 Hz,两者比较接近.另外,研究发现同一图像序列中红、绿、蓝3个颜色分量的图像相关性系数差别较大,但它们的变化频率一致,表明采用该方法分析得到的振荡频率值不受颜色分量选择的影响.Oscillation frequency is a key parameter of flame. Based on Matlab image processing tool, this paper provides a new method of getting the flame oscillation frequency by applying fast Fourier transform to time variation of image correlation coefficients of a continuous image sequences. The gasoline flame oscillation frequency of a square burner of 0.7 m×0.7 m is obtained by the new method, which is 1.80 Hz, agreeing well with the theoretically predicted value by Pagni equation, which is 1.71 Hz. In addition, the image correlation coefficients are of great difference for different colors considered, which are R, G and B, in the same image sequence; however, their changing frequency is the same, demonstrating that colors have no effect on frequency value obtained by this method.

关 键 词：图像相关性 快速傅里叶变换 火焰 振荡频率 

分 类 号：TK411.12[动力工程及工程热物理—动力机械及工程]