中文摘要：

为了提高人体手部动作的识别性能,针对高维特征数据给模式识别带来的问题,提出了一种基于局部线性嵌入（LLE）算法和支持向量机（SVM）的模式识别方法。该方法从肱桡肌和尺侧腕屈肌上采集两路表面肌电信号（s EMG）,通过对样本信号的时域分析和小波分析,提取原始信号的特征,构造特征矢量。再利用LLE算法对原始特征数据进行降维,挖掘出具有内在规律的低维特征。将降维后的特征数据输入SVM分类器进行4种动作的模式识别。实验表明：此方法可以有效、准确地对人体手部动作进行分类。

英文摘要：

To improve performance of human hand movement pattern recognition,a new pattern recognition method based on locally linear embedding（ LLE） algorithm and support vector machine（ SVM） is proposed aiming at problem caused by high dimensional feature data. Two path surface electromyography（ s EMG） are acquired from the brachioradialis muscle and flexor carpi ulnaris,by means of time domain analysis and wavelet analysis,feature of original signal is extracted,and feature vector is constructed. To find low dimensional feature which is more able to express intrinsic law of the characteristics,reduce the dimension by using the LLE algorithm. Feature data after reducing dimension is input into SVM classifier for pattern recognition of four kinds of hand movements. Experiments show that the method can effectively and accurately classify human hand movements.