Research On Human Body Tracking Temperature Measurement Technology Based On Fusion Of Infrared And Visible Light

As the epidemic situation in China gradually improves,epidemic protection has entered a critical stage,and strong epidemic protection work has been carried out in public places around the country.Taking subway stations as an example in indoor public places,responding to and preventing accidental sudden cases has become the primary task.Therefore,normal temperature monitoring is implemented in stations.The existing temperature measurement solutions have shortcomings such as slow temperature measurement speed,low temperature measurement accuracy,and small monitoring range.In recent years,thanks to the rapid development of deep learning technology,the application of target detection technology in security,transportation and other fields has gradually increased.However,the direct application of target detection algorithm to infrared or visible light images cannot simultaneously obtain and utilize the two types of images.The characteristic information will affect the target detection accuracy and temperature measurement accuracy.In view of the above problems,this paper proposes a target detection algorithm based on the fusion of infrared and visible light images based on the idea of feature-level fusion of infrared and visible light images,and designs a set of tracking temperature measurement algorithms and systems based on dual video source input.It is difficult to accurately and quickly measure temperature in complex multi-person scenarios.Aiming at the problems of complex images,many interference factors,and dense crowds blocking each other in visible light images,this paper builds a feature-level fusion detection algorithm based on YOLOv5 s,adds the Dense Fuse feature fusion structure,and takes advantage of the strong anti-interference ability of infrared images and is only sensitive to temperature,combined with the characteristics of high resolution and rich texture details of visible light images,the input features of two types of images are fused at the feature level,which solves the problem that single visible light image input is easy to cause false detection and missed detection,enhances the structure of features,and obtains more Strong semantic information.At the same time,in order to further improve the performance of the fusion detection algorithm,the Decoupled Head is used to replace the original coupled detection head,which enhances the expression ability of the output end,improves the detection accuracy,and can more accurately detect high-density and large numbers of people.Experiments show that the recall rate of the improved method in this paper reaches85.73%,and the average accuracy rate reaches 92.00%,which meets the needs of efficient and accurate real-time detection in complex multi-person scenarios.In order to solve the temperature measurement problem in practical application,this paper designs a set of tracking temperature measurement algorithm and system based on the proposed fusion detection algorithm.The algorithm realizes the temporal and spatial registration of infrared and visible light images by constructing an image preprocessing module;constructing a fusion detection module to obtain the coordinate information of pedestrian head and shoulder area;constructing a multi-temperature measurement area tracking module to continuously capture the pedestrian head and shoulder area to improve the temperature measurement accuracy;The structure temperature extraction and calculation module compare and read and correct the output pedestrian body temperature.The system is designed and constructed based on an infrared-visible binocular camera and an external blackbody device.Finally,through experimental tests,the system achieves an effective and accurate temperature measurement effect with an error of ±0.18°C within a detection distance of 2?5m,which meets the needs of accurate and rapid temperature measurement in complex multi-person scenarios.