Hidden Human Object Detection Based On Image Analysis Technology

At present,for the field of object detection,there is no hidden human object detection algorithm.The realization of human target detection in the hidden scene can help the security monitoring system to quickly track down criminals and assist the search and rescue system to carry out rescue operations.Therefore,it is of great significance to construct the detection models for hidden human objects.The research of this paper is mainly carried out from the following aspects.Specifically,three different object detection models are constructed by data acquisition with different imaging devices and image analysis techniques.Compared with the existing object detection algorithms,the algorithm proposed in this paper has different degrees of improvement.The main contents and innovations are as follows:(1)This paper proposes a hidden human target detection model based on physiological signal features.Compared with deep-learning based detection algorithm YOLO v4 and traditional object detection algorithms such as Haar,LBP and HOG operator,it achieves performance increase by 19%,28%,48% and 34% on the homemade dataset,respectively.(2)The object detection algorithm based on deep learning is improved.Euler color amplification technology is used to amplify the video signal to improve the signal to noise ratio,the inter-frame difference method is used to extract the weak human signal,and the variance threshold method is used to suppress the background target with large movement.Finally,the hidden human target detection based on deep neural network and physiological signal features is realized.Compared with deep learning algorithms such as YOLO v4,YOLO v5,Faster RCNN and SSD,the proposed algorithm improves the detection accuracy by 11%,2%,14%and 1%,respectively.(3)The hidden human target detection model based on far infrared and visible dual-mode fusion data is proposed.Compared with single-mode visible data,the detection accuracy of the model based on dual-mode data is improved by 24%,43%,28% and 76% compared with YOLO v4,YOLO v5,Faster RCNN and SSD,respectively.In this paper,three hidden object detection models are constructed.Compared with the general object detection methods and the traditional object detection methods,the model performance has been improved in different degrees.For the hidden human object detection task,the hidden human object detection model proposed in this paper can make up for the failure of the current object detection algorithm.It is of great significance to the development of intelligent rescue system and intelligent security system.