基于局部非线性投影滤波算法的ECG信号处理

ECG Processing Using Local Nonlinear Projective Filtering Algorithm

含有一定强度干扰的混沌系统可以通过局部非线性投影的方法逼近其吸引子。单导心电信号(ECG)在时间延迟嵌入空间中表现出一定的几何结构 ,包含其它成分的ECG可以在一个较高维的空间得到没有交叠的状态轨迹。利用局部非线性投影法可以逼近ECG的一个低维吸引子 ,从而将ECG和其它成分分离。

Attractor in state space of a chaotic system contaminated with noise can be approximated by means of locally nonlinear projective filtering. Single lead ECG signals present to a certain extent geometric structures in the time delay embedding space. In this way, a trajectory at a high dimension of the signal, that contain components induced by variables which are difficult to describe, can stretch without intersection. In practice, ECG signal can be extracted from other components successfully by approximating a lower dimension attractor using local principal component analysis (PCA) technique.