亮背景下基于图像提取火源振荡频率研究

摘要

This paper presents an improved method to extract the oscillation frequency,a key flame param⁃eter,from ship fire images with bright backgrounds. The method is based on the Matlab image processing toolbox,and its primary principle is to remove the background noise from fire images through image bina⁃rization and by setting a corresponding optimum threshold. The flame oscillation frequency is then obtained by performing the Fast Fourier Transform(FFT)to the fire regions within continuous image sequences. To validate the approach,fire tests have been conducted on four oil pools with different sizes in a ship cabin model. The results show that the oscillation frequencies extracted by the proposed method agree well with those obtained through the theoretical Bejan equation.%　　振荡频率是燃烧火源的一个重要基本参数。在前人的基础上，发展一种亮背景下利用图像测量火源振荡频率的新方法。该方法基于 Matlab 平台，通过去除背景图像并选用最佳分类阈值进行图像二值化处理，可去除图像中的亮背景干扰，有效识别出图像中的火焰区域。在此基础上，对图像亮区面积的周期性变化过程进行快速傅立叶变换（FFT），提取出火源的振荡频率。在模拟船舶舱室内进行4组不同尺寸油盘的火源振荡试验并分别获取视频图像，然后采用该方法分析得到火源的振荡频率。通过将所得结果与 Bejan 理论公式进行比较发现，4组振荡频率计算结果均与理论预测值吻合较好，验证了新方法的可行性。