一种宽吻海豚通讯信号自动分类的方法  

An automatic classification method for whistles of bottlenose dolphin(Tursiops truncates)

作　　者：杨武夷[1] 孙馨喆 张宇[1] 魏翀[1] 杨燕明 牛富强 

机构地区：[1]厦门大学水声通信与海洋信息技术教育部重点实验室,厦门361000 [2]中国海洋局第三海洋研究所,厦门361000

出　　处：《声学学报》2016年第2期181-188,共8页Acta Acustica

摘　　要：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。To automatically classify bottlenose dolphin＇s whistles, a method, which is based on syntactic pattern recognition, is presented. Dolphin whistles have typically been characterized in terms of their instantaneous frequency as a function of time, which is also known as ＂whistle contour＂. The frequency variation feature of a dolphin whistle is extracted according to its whistle contour. Then, the frequency variation features are used for learning grammatical patterns. A dolphin whistle is classified according to grammatical pattern of its frequency variation feature. Experimental results show that the classification accuracy of the proposed method is 95%. The method can provide technical support for acoustic study of dolphins＇ biological behavior.

关 键 词：宽吻海豚 通讯信号 自动分类 句法模式识别 分类准确率 生物学行为 后提取 轨迹曲线 

分 类 号：Q62[生物学—生物物理学] TN911.7[电子电信—通信与信息系统]