摘要
近年来火灾相关的消防安全问题成为研究热点之一,而电气消防中的人员定位问题又成为关注的重点问题。将DeepSORT算法与空间坐标转换算法相结合,针对火灾发生时被困人员的位置进行检测和实时定位,该算法中采用改进的卡尔曼滤波器与匈牙利计算器相结合,对人员运动的轨迹进行识别匹配与更新,识别后再通过坐标转换实时显示人员定位,解决了对目标难以进行准确识别以及定位不精确的问题,并在减小了多目标识别时对人员检测所产生的误差问题,达到在复杂建筑中对多目标准确快速识别的效果。
In recent years,the fire safety problem caused by the frequent occurrence of fires has become one of the research hotspots,and the personnel positioning problem in electrical fire protection has become a key issue.This paper combines the DeepSORT algorithm with the spatial coordinate conversion algorithm to detect and locate the position of the trapped personnel in real-time when the fire occurs,the algorithm uses the improved Kalman filter combined with the Hungary calculator to identify and match and update the trajectory of personnel movement,and then display the personnel positioning in real-time through coordinate conversion after identification.It solves the problem that it is difficult to accurately identify and accurately locate the target,and reduces the error caused by the detection of personnel in the multi-target recognition,to achieve the accurate and fast recognition effect of multi-target in complex buildings.
作者
李东泽
魏立明
刘旭
LI Dongze;WEI Liming;LIU Xu(Jilin Jianzhu University,Changchun 130118,China;Guohua(Qian’an)Wind Power Co.,Ltd.,Songyuan 131400,China)
出处
《技术与市场》
2024年第11期104-108,113,共6页
Technology and Market
基金
吉林省科技厅科技发展重点研发项目(20230203129SF)。
