| 李岳鹏,杨军,李晓东,田静.基于蝙蝠回波定位模型的材质识别方法研究[J].,2007,26(6):335-340 |
| 基于蝙蝠回波定位模型的材质识别方法研究 |
| Texture classification based on echolocation used by bat |
| |
| 中文摘要: |
| 在蝙蝠的回波定位和目标识别模型的基础上完成了目标材质识别的工作。利用超声换能器发射超声脉冲,接收回波信号并对其进行处理;分别提取了时域包络、功率谱和短时傅里叶(STFT)系数作为BP(Back-Propagation)网络分类器的输入特征向量;完成了对纸板、泡沫板和木板的识别工作。实验结果表明识别正确率最高达到近90%。 |
| 英文摘要: |
| Based on bat sonar technology and pattern classification,a texture classifi- cation system was set up to classify three types of texture:pasteboard,foamboard and woodboard.It utilized an ultrasonic signal transmitter and a receiver to receive the echo signal.Three types of feature vectors were to be used as the input to BP(Back- Propagation)neural network classifier:time domain envelope,frequency domain power spectrum and time-frequency domain spectrogram).The result of experiment showed that the BP network had a maximum correct classification rate of nearly 90%. |
| DOI：10.11684/j.issn.1000-310X.2007.06.003 |
| 中文关键词: 蝙蝠 声纳 回波定位 超声 材质识别 |
| 英文关键词: Bat Sonar Echolocation Ultrasound Texture classification |
| 基金项目: |
|
| 摘要点击次数: 2789 |
| 全文下载次数: 184 |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
