| Prison is an important part of the national judicial system,and bears an important responsibility for the reform of the criminal,whose security and reliability of the security system are significantly higher than other places'.The traditional video monitoring system cannot analyze video in real time,and it is difficult to find and solve problems in time by means of artificial monitoring,which can not adapt to the development of supervision system.Therefore,it is of great engineering value to study the intelligent monitoring system applied to the supervision.In this paper,the software of supervision system is designed and developed for the security,reliability and real time of the monitoring system.This system mainly realizes the real-time broadcasting of panoramic video,multi-channel video stitching and personnel detection.According to the need of real-time monitoring system for a panoramic view,the client in system presents panoramic video and alarm events,achieve the purpose of prison monitoring reliability and security.In the light of the distortion of prison panoramic images and blind area,using computer vision technology and multithreading technology,designed to achieve the multi-channel video stitching scheme based on SURF(Speed-up Robust Features),which gets less distortion and blind area,and presents real-time panoramic picture of the prison.Based on the requirements of the accuracy and efficiency of character tracking,a personnel detection scheme based on SSD(Single Shot MultiBox Detector)was designed,using the Caffe for learning and testing.The intelligent monitoring system developed by this thesis has been put into practice,which realizes the panoramic intelligent monitoring system for prisons,and satisfies the system requirements in terms of function and performance.According to the test,the system video splicing module requires an average time of 20.4ms for the single stitching,and the detection accuracy of the personnel detection is about 96%. |
PDF Full Text Request