Research On Personal Safety Monitoring System Of High Voltage Laboratory Based On Computer Vision

The traditional high-voltage laboratory safety protection mainly relies on manual reminders,the professional knowledge and high safety awareness of the staff.Secondly,by installing a locking device on the laboratory door and the operation room door to prevent the people who enter or go out of the lab.There are no reliable safety precautions measure for unsafe phenomena such as personnel detention and mechanical movement in the laboratory during the testThis paper first analyzes the special application environment of the 1100KV UHV power test hall of Xiamen University of Technology and the safety risks of electrical test work and the safety protection systems used by domestic and foreign laboratories.Based on the analysis results,a set of computer vision-based personal safety protection monitoring system based on computer vision is designed and implemented.The system design first analyzes the influence of the electromagnetic field generated by the high-voltage equipment on the system.The overall structure of the system is designed based on the analysis results,and the system logical structure and design standards are formulated.Secondly,in terms of software design,the system analyzes various algorithms such as general background difference method,inter-frame difference method,mixed Gaussian model background difference method,Mean-Shift tracking algorithm.It is found that the traditional mixed Gaussian model background difference method is greatly affected by shadows and the target is incomplete.The saturation parameter is introduced to reduce the effect of shadows and enhance the integrity of moving targets.The traditional Mean-Shift tracking algorithm is greatly affected by the background due to incomplete target feature extraction.The introduction of a two-dimensional histogram can reduce the impact of the background.Then using the above two improved algorithms at the same time,the system's ability in motion detection and tracking has been effectively improved.At the end of this paper,according to the system design,Hikvision's webcam capable of outputting the video data stream of h.264 through the network cable is used in the hardware to facilitate computer video data reading.In the aspect of human detection,the recognition method of human height and aspect ratio is adopted.The system software is GPU accelerated and multi-threaded accelerated to make a complete security protection system.It was then installed and tested in the 1100KV high-voltage power test hall of Xiamen University of Technology.The test result is that the system response time is 100ms,the success rate of 18m distance motion detection reaches 96.6%,and the success rate of personnel tracking reaches 89.6%.The results of the system test met the design requirements and met the basic needs for personal safety protection in high-voltage laboratories.