波形的自动分类

给出了一种新的波形链表达模型，提出了一种新的波峰分类方法，并给出了识别算法。所提方法数学概念清楚，易于计算机实现，可用于具有结构特征的信号的识别。

Automatic Classification of Signal Waveform Useful in Diagnosis

Waveform of one kind of signals plays a very Important role in diagnosis, either ensineering or medical. Statistical recognition is not satisfactory for automatic recognition of waveform. Syntactic recognition, introduced by world-famous Fu2] to Chinese scientists in the early eighties, appears to the authors to be a good way of recogniaing waveform automatically.We are particularly interested in signals whose main band -width is relatively narrow and whose measured signal--noise ratiO is relatively high.We present a new coding method to replace the one based on difference operator and now used in China for syntactiC recognition. This is because there are two shortcomings inherent in coding method based on difference operators: (1) chain is long and consequently speed of recognition is fow; (2) features of waveform such as concavity and convexity can not be described.We present an algorithm, which takes into consideration the features of left and right slopes and top of peak. This algorithm can automatically distinguish between 18 basic ircesof peak (8 of them are shown in Fig. 1) and is easily implementable on computer.