Research On Detection And Recognition Technology Of Infrared Thermal Imaging Human Target

Visible light is sensitive to changes in illumination,visibility at night,and other environmental effects.Compared with visible light,infrared thermal imaging can effectively avoid these problems.The infrared thermal imaging detection technology has many advantages such as non-contact,intuitive,convenient operation,safety,and large single-detection area.However,infrared thermal imaging brings other challenges due to its unique imaging characteristics.The use of infrared thermal imaging for human detection analysis is not simple.The detection and recognition of human targets has always been a research hotspot in the field of image information processing.Infrared thermal imaging can work under any external conditions and has a wider range of applications than visible light,so the infrared thermal imaging target detection has a very good application prospect.The infrared thermal imaging inspection process includes multiple stages,each stage contains multiple links,and each link is closely linked.The processing effect of each link will affect the subsequent.Infrared thermal imaging has its own special imaging mechanism and characteristics.Therefore,in the design of the detection system and the processing of algorithmic programming,it is necessary to optimize and ensure the processing effect of the next step to ensure the accuracy and speed of infrared thermal imaging human detection,and the best real time processing results are obtained.In view of the above problems,based on the analysis of existing infrared and visible light human body target detection methods and the imaging characteristics of human body targets in infrared images,a static human body target on an infrared thermal imaging image is taken as the research object,the infrared thermal imaging human target detection is studied.The main contents are as follows:(1)The current methods and processes of infrared thermal imaging human target detection commonly used at home and abroad is analyzed.The experimental principle of each step is analyzed.Using the commonly image processing algorithms to process and analyze infrared thermal imaging.Applicable algorithm of detection infrared human targetand the advantages and disadvantages of various algorithms are summarized from the theoretical level.(2)An improved fast global K-means clustering algorithm is proposed for image segmentation in view of the problem that detection and identification of infrared human target in complex background.The clustering center point is selected according to the anti-sensitivity principle of median filter.The median gray level is selected as the initial cluster center,and the problem of inaccurate cluster segmentation results due to the random selection of the original center of the K-means cluster is solved.Iteration time is shortened and the efficiency of image recognition is improved by setting gray feature threshold.(3)Human body feature extraction and recognition.The combination of HOG+SVM algorithm is selected.HOG feature extraction can maintain good invariability of image collection and the invariance of features of optical deformation.Subtle human posture changes do not affect feature extraction.The HOG feature of the target to be measured is extracted,including the feature set of the gradient size and direction,and input to the next selected SVM classifier to find the optimal hyperplane of the SVM classifier.As shown in experimental results,this algorithm has good recognition effect and applicability to human body recognition.(4)The algorithm presented in this paper is verified by taking photos in a real environment using an infrared camera and compared with the original algorithm.The experimental results show that the accuracy of the algorithm used in this paper is improved by about 3% and the average running time is also reduced by about one second.The availability and practicability of the algorithm are obtained by applying it to the actual scene and data statistics.