改进GOMP算法在心电信号去噪中的研究与应用

心电信号的采集、传输在医学研究心脏性疾病具有重要意义,利用稀疏分解方法对其进行去噪仍存在诸多问题。针对广义正交匹配追踪(Generalized Orthogonal Matching Pursuit,GOMP)算法选择字典原子时在残差取值最小的非最优性,以及在稀疏信号过程中选择过多原子导致的重构精度问题,提出一种新的优化算法。通过增加阈值的方式来根据噪声干扰程度不同实现迭代次数的变化。利用Jaccard系数相似性来替换GOMP算法中内积余弦算法准测,优化了字典矩阵与残差匹配原子的方式。实验结果表明,改进GOMP算法明显提高了信噪比,且均方根误差最低,在时效性上达到最优。

RESEARCH AND APPLICATION OF IMPROVED GOMP ALGORITHM IN ECG SIGNAL DENOISING

The acquisition and transmission of ECG signals are of great significance in the medical research of cardiac diseases.There are still many problems in denoising by sparse decomposition method.Aiming at the non-optimality of the minimum residual value when selecting dictionary atoms in the generalized orthogonal matching pursuit algorithm and the reconstruction accuracy caused by choosing too many atoms in the sparse signal process,we proposed a new optimization algorithm.The number of iterations was changed according to the degree of noise interference by increasing the threshold.The similarity of Jaccard coefficients was used to replace the inner product cosine algorithm in GOMP algorithm and the method of matching atoms with dictionary matrix and residual was optimized.The experimental results show that the improved GOMP algorithm significantly improves the signal-to-noise ratio,and the root mean square error is the lowest and the timeliness is the best.