The Design Of Image Fusion Temperature Detection System Based On Floating Population

Influenza viruses have been harassed in recent years.Because these diseases are easy to cross-infect,especially in densely populated and mobile public places,it is difficult to prevent and control them.The main characteristic of influenza virus is the rise of body temperature.Therefore,rapid and accurate measurement of body temperature in densely mobile population is an important means of prevention and control.Common infrared point thermometers can not be imaged and can only measure temperature at a single point.Handheld infrared imaging thermometers rely on manual operation.Because the thermal infrared thermometer needs to keep the measured object at a relatively close distance,it is difficult to measure the surface temperature of the object quickly and accurately in complex environment.In this paper,a non-contact thermal infrared temperature measurement system is designed to accurately detect the face and extract the temperature of the face.Firstly,the principle of infrared radiation is introduced,and the working principle of infrared temperature measurement is elaborated in detail.Secondly,the key technologies of thermal infrared temperature measurement system are elaborated in detail,including face detection of video content,face orientation recognition,visible light and thermal infrared image registration.In visible image detection,residual network is introduced based on YOLOv2 algorithm,which enhances the applicability of data set and improves the detection accuracy and speed of non-positive face and partly occluded face.The parallel concatenated LVQ network is used for face orientation recognition,and the face candidate frame is judged to further determine the face forehead region.In the part of visible and thermal infrared image registration,a correction method is proposed based on affine transformation model to achieve accurate image registration.The forehead region of visible image is mapped to thermal infrared image in real time,which can effectively solve the image registration deviation of different imaging distances.Finally,the main factors affecting the accuracy of thermal infrared temperature measurement are analyzed,and external blackbody correction,distance correction and other methods are used to improve the overall accuracy of temperature measurement.In this paper,a temperature measurement method based on visible and thermal infrared images is implemented,and an infrared temperature measurement system is built,and the actual scene is experimented.The experimental results show that the method has high accuracy and fast detection speed for non-frontal and partially occluded faces,and can accurately detect the region of forehead position and quickly extract the surface temperature of the region of interest according to the face orientation recognition algorithm.The error of this method is less than 0.2 C,which can meet the requirement of rapid screening of human surface temperature in complex scenes.