摘要
提出一种以图像识别技术为基础的火焰识别方法。为将火焰与背景分离,首先通过灰度转换将彩色图像转化为灰度图像,然后用中值滤波处理法对图像进行平滑处理去除图像噪声,再用最大类间方差法进行二值化处理分离火焰与背景图像,用腐蚀与膨胀算法进行数学形态处理突出火焰图像特征,最后采用区域形状参数作为火焰图像判别依据,去除灯光等疑似干扰图像。实验表明,该方法可较好地辨别火焰和干扰源。
A flame recognition method based on digital image processing was put forward.In order to separate the flame and background,the color image was transformed into grey level image by grey level transforming,the noise was removed by smoothing based on median filter method,separate the flame and background by Otsu method to perform the binarization,highlight the flame characteristics by morphological processing such as corrupt and expand,and take regional shape parameter as reference for the flame image recognizing basis to remove the interrupt such as lamplight.The result of experiment proved that the method could recognize the flame and disturbs availably.
出处
《消防科学与技术》
CAS
北大核心
2010年第12期1051-1054,共4页
Fire Science and Technology
