Design And Implementation Of Intelligent Fire Identification Video Monitoring System

At present,the mainstream real-time video surveillance system has some shortcomings,such as unclear video input,low coding efficiency which cannot meet the requirements of bandwidth,and the acquisition terminal does not support intelligent video analysis algorithm.With the continuous improvement of the performance of embedded processor,the development process of video surveillance system is greatly promoted,and the research on the active recognition algorithm of abnormal situation based on video surveillance is of great significance.This system adopts the embedded hardware platform built by HI3516DV300 based on dual-core ARM Cortex-A7 architecture and PC,designs and implements the intelligent fire identification video monitoring system,studies the fire detection algorithm suitable for HI3516DV300 platform,and actively identifies the occurrence of fire.The video image acquisition,HEVC coding and RTSP streaming media transmission of the system are all completed on ARM embedded platform.The VI module in MPP is used to collect video images,the VENC module is used to encode and output HEVC code stream,and the ORTP library is transplanted to support the RTSP streaming media transmission protocol for wireless network transmission.At the same time,the C/S client written based on Qt transplants VLC-Qt library for remote display,communicates with the server through socket,and realizes the update of important status(such as whether the fire occurs).We also study the flame detection algorithm suitable for HI3516DV300 platform.Since most objects in the fixed monitoring picture are static in the observed scene,the flame of a natural fire will be regarded as a moving target in the monitoring image,and the influence of the static background can be eliminated by motion detection.The fire object is separated by the RGB color space characteristics of fire flame.Finally,the dynamic features of adjacent flame images are detected and recognized comprehensively.In the program design,the video capture coding transmission program and the fire flame identification program run in the server at the same time.Finally,each module of the system has been tested,and the requirements of high-definition video monitoring have been met.At the same time,the fire flame recognition algorithm based on video monitoring has a high accuracy,the introduction of the judgment of fire flame disordered motion characteristics based on SURF and Bo W greatly reduces the false alarm rate and has a certain practical value.