基于知识的声目标探测识别系统  

Research and Realization of Acoustic Target Detection and Recognition System Based on Knowledge

作　　者：伍宗富[1] 杨如曙[1] 李晓峰[1] 

机构地区：[1]湖南文理学院电气与信息工程学院,湖南常德415000

出　　处：《计算机测量与控制》2010年第5期1132-1135,共4页Computer Measurement &Control

基　　金：湖南省科技计划项目(2009GK3059);湖南文理学院资助科研项目(JJYB0716)

摘　　要：被动声目标探测广泛应用于战场目标识别或自动设备的故障探测;通过对声目标的短时信号处理,使用现场可编程门阵列器件的可重构技术对声目标探测识别系统设计;提出通过提取子带Mel倒谱系数(MFCC)参数特征构建声目标信息的知识库,并使用0阶Mel倒谱系数(MFCC0)进行频谱能量分析,找寻信号起止端点,将声目标Mel倒谱系数(MFCC)特征参数映射为二值图像进行模板匹配识别;将声目标识别输出的控制指令传送给工控机或直接输出控制相关的智能系统,实现战场声目标识别或自动设备的声故障探测。The passive acoustic target detection is widely used in the battlefield target identification or fault detection. Through for the acoustic target＇s short--time signal processing, the use of FPGA designed to the acoustic target detection and recognition system. By extracting features of the formation of sub-band MFCC parameters of acoustic target information knowledge base, and use the MFCC0 to carry on the energy spectrum analysis, to find the signal starting and ending endpoints, enable acoustic MFCC feature parameters maps to binary ima- ges and carry out template matching and target recognition. The output of the acoustic target identification control instruction transmitted to the output control industrial computer or directly related to intelligent systems to achieve battlefield acoustic target identification or automatic equipment, acoustic fault detection.

关 键 词：现场可编程门阵列 声目标 MEL倒谱系数 二值图像 目标识别 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]