基于双边滤波与帧差方差法的微小泄漏点检测

Device of Device Micro Leakage Point Based on Infrared Image

为了对器件的微小泄漏进行检测定位,介绍了一种利用红外热像对复杂器件进行微小泄漏点的定位方法,利用红外相机采集多帧经热气加热并进行差压脉冲操作时被测器件的红外图像。对红外图像进行双边滤波去除噪声后,利用泄漏点的温度场的变化即图像灰度值细微变化,提出了一种帧差方差法,通过增强泄漏点灰度值变化,准确定位到泄漏点位置,并对得到的泄漏点进行灰度值帧差概率计算确定是否为真泄漏点。实验证明:与其他方法相比,所提方法能够有效去除周围噪声,克服了复杂器件易受红外反射率的影响、微小泄漏点变化幅度小、变化较慢的问题,并得到了较好的接受者操作特征(receiver operating characteristics,ROC)工作曲线,同时可以满足工业生产中对微小泄漏检测的要求。

In order to detect and locate the tiny leakage of the device,the infrared thermal image was used to investigate a method of locating the tiny leakage point of the complex device.The infrared camera collects infrared images of the measured device when multiple frames are heated by hot gas and are operated by differential pressure pulse.After removing the noise from the infrared image by bilateral filtering,a frame difference variance method was proposed by using the change of the temperature field of the leakage point,that is,the slight change of the gray value of the image.By enhancing the change of the gray value of the leakage point,the leakage point was accurately located.The gray value frame difference probability of the leakage point was calculated to determine whether it is a true leakage point.The results show that compared with other methods,the proposed method can effectively remove the surrounding noise and overcome the problems of complex devices vulnerable to infrared reflectivity,small change range and slow change of tiny leakage points.A better receiver operating characteristic curve(ROC)was obtained,which can meet the requirements of tiny leakage detection in industrial production.