一种基于微分阈值法的脉搏波特征点提取优化算法

基于微分阈值法测得的脉搏波信号中通常混杂了许多噪声干扰,并且常规方法提取特征点效果不理想,故对提取脉搏波的特征点算法进行改进优化。首先对滤波后的脉搏波信号进行微分,采用滑动窗口的方式在源数据中提取出最小值点,以该点为极小值基准点,在窗口内寻找邻域内的波谷。然后对波谷的左右邻域二次微分,取最大值,在该邻域内提取出类似的波峰。将检测窗口滑动到上一次窗口的边沿,重复上述步骤,提取出检测数据中的脉搏波特征点。对于已有算法因混合许多噪声干扰,经微分后噪声信号也会过零点,导致错误地提取出脉搏波特征点,将优化后的提取算法应用到嵌入式检测系统。从实验对比发现,特征点提取误差率降低到0.12%。经优化的算法能够很好地剔除噪声信号和脉搏波信号的干扰,准确地提取出脉搏波特征点。

Optimize Pulse Wave Feature Extraction Based on Differential Threshold Method

The research in this paper is based on the pulse wave always mixed much noises,and the conventional method of extracting feature points is not ideal.For improves and optimizes the differential threshold method for extract feature point of pulse wave.First,differentiate the filtered pulse wave signal,and the sliding window is used to extract the minimum point in the source data,and taken this point as the minimum reference.For existing algorithms,many noise interferences are mixed.After the differentiation,the noise signal will also cross the zero point,resulting in the wrong extraction of the pulse wave feature points.