摘要
针对由红外热像仪直接采集到的红外热图很难直接获得红外辐射温度场分布的问题,本文拟提出一种提取红外热图每像素点温度值的方法。以某红外热像仪采集到的JPEG格式的人体面部红外热图作为研究对象,首先通过对数据的统计分析,获得几种常用的面部测温范围;然后采用最小二乘拟合曲线的方法,获得这几种测温范围下图像灰度值与实际温度值之间的对应关系;最后根据上述对应关系提取出图像上每像素点处的温度值。实际测试表明:该方法能能有效提取人体面部温度分布,且温度估计值的偏差均值和精度均小于0.2℃。
It is difficult to know the distribution of infrared radiation temperature field from infrared thermographs captured by an infrared thermal camera, so this paper put forward a method for extracting values of temperature from each pixel of an infrared thermograph. Aiming at human face thermographs collected with an infrared thermal camera in JPEG format, it firstly performed a statistical analysis of data and confirmed several common ranges of temperature. Next, the relation- ships between gray scale and actual temperature of images were determined under such ranges of temperature by the least square fitting curves. At last, the values of temperature at each pixel were extracted from the images based on the above relationships. The actual tests suggest that this method is effective in extracting temperature distribution on human face, ensuring that both the mean deviation and accuracy of estimated temperature are below 0.2℃.
出处
《计算技术与自动化》
2016年第3期85-89,共5页
Computing Technology and Automation
