关键词: 表面肌电信号（sEMG）, 非负矩阵分解（NMF）, 支持向量机（SVM）, 特征矩阵, 模式识别

Abstract:

A joint model based on Non-negative Matrix Factorization（NMF） and Support Vector Machine（SVM） is proposed for action recognition based on surface Electromyography（sEMG）. The main idea of this model is the different activation degree of each muscle corresponding to different exercise patterns. An eigenmatrix with the row representing the muscle type and the column representing the eigenvalue type is constructed from data extracted from the sEMG. The coefficient matrix representing the activation degree of each muscle is obtained through the dimensionality reduction of non-negative matrix decomposition. Then the obtained coefficient matrix is transferred to support vector machine for training classification. Finally, compared with traditional SVM, the computational efficiency and recognition rate of the proposed joint model is improved by half and 5.2% respectively. Sample separation experiments show that the recognition rate of the algorithm is still more than 91.7%, which verifies the effectiveness of the algorithm.

Key words: surface Electromyography（sEMG）, Non-negative Matrix Factorization（NMF）, Support Vector Machine（SVM）, characteristic matrix, pattern recognition