文档介绍：安徽大学
硕士学位论文
地震动传感器的目标检测和识别算法研究
姓名:徐照胜
申请学位级别:硕士
专业:计算数学
指导教师:胡茂林
2010-05
地震动传感器的目标检测和识别算法研究
摘要
本文研究基于地震动传感器的地面运动目标的检测和识别算法及其实现。
地面目标的检测算法主要是对采集的目标信号在时域内通过改进的高阶过
零分析算法进行检测。原理是设计动态区间窗口,对每个窗口的一阶过零数进行
计算,然后与设定的阈值比较,得出检测结果。
目标识别算法是对采集的信号进行小波分解降噪,然后对目标信号在时域和
频域内的过零计数识别和改进的周期图功率谱估计分析识别方法进行判别——
基于过零数识别方法区别人和车辆,通过改进的周期图功率谱估计分析方法区别
单人和多人目标,最后综合这两种方法得出识别结果。
对于硬件系统的设计,针对整个系统数据量大、逻辑控制复杂以及低功耗、
使用、修改方便等要求,本文采用具有高速信号处理功能的数字信号处理器实现
的。硬件系统主要由数据输入模块、数据处理模块、外部存储模块、输出模块、
监控模块和电源模块等六个模块组成。系统的软件设计方案包括算法的程序编写
以及将其转化为 DSP 系统的可执行代码。
本文研究的成果在中科院上海微系统与信息技术研究所某项目中得到了应
用,在外场试验中,目标的最远检测距离和识别的准确率都达到了项目的要求。
关键词:地震动传感器,小波降噪,高阶过零,功率谱,DSP
I
地震动传感器的目标检测和识别算法研究
ABSTRACT
The property is researched on detection and recognition algorithm based on
seismic sensors
Detection Algorithm of ground target is mainly to detect the acquired target
signal in time domain by the improved higher order zero-crossings algorithm. The
principle is to design dynamic windows, calculate the zero-crossings in each window,
and then obtain the test pared with the fixed the value of a threshold.
Target recognition algorithm wavelet-poses and noise-deafens the
acquired signals. Then it inspects the periodic chart in power spectrum estimation
which is identified and improved zero-crossings in time and frequency domain. This
method is based on the zero-crossings to identify different people and vehicles,
through improving the periodic chart in power spectrum estimation, with analytical
method of estimation to identify single and multiplayer objectives, and finally
combines these two methods to obtain the identifying result.
As to the design of hardware system, aim at the volume of data, complexity of
logic control and the requirements of less power, usage, ease of amendment in the
entire system, this paper adapts high-speed signal processing digital signal processor
to obtain information. Hardware system mainl