一种基于压缩域的穿戴式心电信号直接特征提取与分类方法

压缩感知是实现可穿戴式健康监测系统低能耗工作方式的一种有效途径，而现有基于压缩感知的心电信号分类方法大多需要在进行分类之前，先使用重构算法恢复出原始心电信号，这可能会导致较高的计算复杂度高，不适合于具有实时性需求的可穿戴式系统。提出一种基于压缩域的穿戴式心电信号的特征提取与自动分类方法。跳过信号重构步骤，使用改进的主成分分析法在压缩域上直接对压缩后的心电信号进行特征提取，并基于最小二乘支持向量机半监督学习方法实现心电信号的自动分类。实验结果表明，相较于在非压缩域上的分类方法，该方法在保证分类性能下降非常少的前提下，心电数据量大大地减少，有效提高了心电信号自动分类的效率。

A feature extraction and automatic classification method of wearable electrocardiosignal based on compressed domain

Compressed sensing is an effective way to achieve a low-energy working mode of wearable healthy monitoring systems. However, most current electrocardiosignal classification methods based on compressed sensing need to use a reconstruction algorithm to recover the original ECG signal before classification. This may lead to high computational complexity and is not suitable for wearable systems with real-time requirements. This paper proposes a feature extraction and automatic classification method of wearable ECG signals based on compressed domain. The signal reconstruction step is skipped. The improved principal component analysis method is used to directly extract the feature of the compressed ECG signal in the compressed domain, and the ECG signal is automatically classified based on the least-squares support vector machine semi-supervised learning method. The experimental results show that this method can greatly reduce the amount of ECG data and ensure the efficiency of automatic classification of ECG signals, compared with the classification method in the uncompressed domain.