摘要
火焰图像识别现在已有较为成熟的方法和实现手段,然而大多数方法都是针对脉动强烈,且主色为红色的火焰的。而高炉煤气火焰的特征为焰色淡蓝,且基本无脉动。因此,传统的火焰检测方法不适用于高炉煤气火焰的检测。文中针对高炉煤气火焰的特点,提出了一套基于蓝色色彩的图像处理的检测方法。摄像头采用全光谱摄像机,保证取得完整的燃气火焰视频。在后期的处理上,首先使用背景差分法过滤掉与火焰不相关的背景信号;然后利用火焰的颜色信息对图像进行滤波增强,主要是增强火焰的蓝色分量的信息;最后用O tsu法确定阈值对图像进行二值化处理。这种方法的特点是利用了燃气火焰的颜色特征,对其RGB 3个通道图像分别进行不同的处理;同时利用其脉动小的特征对其进行背景分割。模拟实验结果表明,该方法可以准确地分割出燃气火焰的图像,从而提高了识别此类火焰的准确率。
Various practical methods have been proposed for flame image recognition.Most of the methods face the flames flickering and with red as its main color,while the high furnace gas flame is stable and light blue in color.This paper presented an image processing approach for high furnace gas flame examination based on the color characteristic of the flame.A full spectrum camera was adopted to ensure complete video clip of the flame.Post process was performed as follows.Unrelated background information was removed by background difference method.Then imagine enhancement using filtering methods was performed with color information of the flame.Image binarization was conducted using Otsu method.The proposed approach is unique in utilizing the color and relative stable character of the gas flame.Simulation results show that the approach can extract the image of the flame with high precision.
出处
《仪表技术与传感器》
CSCD
北大核心
2010年第11期85-87,90,共4页
Instrument Technique and Sensor
