Temperature Analysis Methods For Human Thermal Infrared Images

Thermal imaging technology is widely used in military,aerospace,industrial,medical,security and other fields to record surface temperature information.In medicine,body temperature is one of the indicators of abnormal activity of human tissues,so in-depth analysis of thermal infrared images can be used to diagnose specific conditions.Limited by the thermal infrared imaging equipment,the resolution of the thermal infrared image is low.Usually,the upper and lower body images of the human body are collected by the camera’s movement in the vertical direction.At the same time,due to the lack of visible texture details in thermal infrared images,it is difficult to accurately locate human body parts.Because of the above problems,this dissertation carries out the research work as follows:(1)To complete the Mosaic of upper and lower thermal infrared images and eliminate the background,a series of image preprocessing algorithms are proposed.In the binarization stage,the sliding window method is used for local threshold segmentation.In the splicing stage,the binary image is first used for rough matching,then the grayscale template is used for fine matching,and finally,the image is fused.The algorithm is validated on 300 sets of sample images.The experimental results show that both binarization and whole body splicing are satisfactory.(2)To analyze the temperature of infrared thermal images of the human body,several image processing algorithms are proposed,including the division of human body parts,abnormal temper-ature detection,texture similarity region detection,etc.Based on the curvature analysis of contour,the key points of the human body are calculated and the regions are divided by the key points.The algorithm is validated on 300 sets of sample images.Abnormal temperature detection can be real-ized by contour analysis and morphological processing,and also by gradient analysis.Besides,the gray co-occurrence matrix is used to distinguish the texture similarity.The realization of the above algorithm brings data support and efficiency improvement for infrared thermal image diagnosis of doctors.(3)To accurately divide the parts in the infrared thermal image of the human face,a multi-mode image acquisition method of the human face is proposed.The acquisition device is a binocular camera consisting of a visible light camera and a thermal infrared camera.Firstly,image acquisition is carried out on the calibration plate and the parameters of image homography transformation are calculated.Then the visible-infrared thermal image of the human face is collected and processed.We use a Windows form application program to realize face image acquisition and processing,which integrate calibration and registration,face image acquisition,temperature statistics,database storage,and query and other functions.