摘要
在对火灾中受困人员定位研究过程中,由于火灾现场的烟雾浓度、现场环境等因素是不可控的,导致无法在火灾现场准确确定用于人员图像定位的阀值特征。采用当前的算法进行火灾中受困人员定位时,人员图像识别的阀值确定过程存在明显滞后性,现场随机干扰较强,缺陷明显。提出了基于图像NMI特征算法的火灾中受困人员定位方法,先对火灾中受困人员定位图像尺度空间的极值进行检测,消除对比度较低的火灾中受困人员定位关键点,获得较为精确的火灾中受困人员定位关键点,将火灾中受困人员定位图像的灰度值与设定的门限进行比较,依据火灾中受困人员定位图像的二值特征消除背景干扰,得到完整的火灾中受困人员定位信息,同时融合于特征灰度门限化方法确定受困人员定位的空间位置,并融合于相似度度量原理对火灾中受困人员的位置进行定位。仿真结果表明,火灾中受困人员定位方法有效的提高了火灾中受困人员定位的准确性。
A locating method of the trapped persons in fire disaster is put forward based on the feature of image NML Firstly, the extremum of the scale space of the trapped persons＇ location image in fire disaster is detected, the located key points of the trapped persons in fire disaster with low contrast are eliminated, and more accurate located key points of the trapped persons in fire disaster are obtained. A comparison between the gray value and the setting threshold of the trapped persons in fire disaster is performed. Based on the two-value feature of the trapped persons＇ locating image in fire disaster, the background interference is eliminated, and the complete locating information of the trapped persons in fire disaster can be obtained. At the same time, the spatial position of the trapped persons is determined by fusing the gray level thresholding method of feature, and the position of the trapped persons in fire disaster is determined by the fusion of the similarity measure theory. The simulation results show that the proposed method can effectively improve the accuracy of the trapped persons locating in the fire disaster.
出处
《计算机仿真》
CSCD
北大核心
2016年第7期440-444,共5页
Computer Simulation
关键词
火灾
人员定位
特征提取
Fire disaster
Personnel locating
Feature extraction
