论文部分内容阅读
针对宽吻海豚通讯信号自动分类提出了一种基于句法模式识别的方法。该方法首先提取海豚通讯信号基频随时间变化的轨迹曲线,然后提取基频变化的基元序列。根据海豚通讯信号分类的标准,归纳出产生各类海豚通讯信号基频基元序列的文法。对未知类别的海豚通讯信号,提取其基频变化的基元序列,根据各类模式的文法对基元序列进行分类,进而实现海豚通讯信号的自动分类。实验结果显示本文方法的分类准确率达到了95%。本文方法预期为海豚生物学行为的声学研究提供一定的技术支持。
Aiming at the automatic classification of bottlenose dolphin communication signal, a method based on syntactic pattern recognition is proposed. The method first extracts the trajectory curve of the fundamental frequency of the dolphin communication signal with time, and then extracts the elementary sequence of the fundamental frequency variation. According to the standard of dolphin communication signal classification, the grammar of generating basic element sequence of various dolphin communication signals is summarized. For primitive dolphin communication signal of unknown type, the primitive sequence of fundamental frequency change is extracted, and the sequence of primitives is classified according to the grammar of various modes, so as to realize the automatic classification of dolphin communication signals. The experimental results show that the classification accuracy of this method reaches 95%. This method is expected to provide some technical support for the acoustic research of dolphin biology.