Abstract: In order to improve the convergence rate and accuracy of Independent Component Analysis (ICA) algorithm, an independent component analysis of innovation model was proposed. The fundamental mechanism of innovation model was to reduce the redundancy among the observed samples, thus it could increase the nonGaussianity of the latent components. Approximately independent image signals were taken to do the simulation. The simulation results show that the new method has a superior performance in both converge rate and accuracy to the traditional one.

Key words: Independent Component Analysis (ICA), innovation model, time filtering, kurtosis

摘要： 为提高独立分量分析(ICA)算法的收敛速度与收敛精度，引入ICA方法的新息模型。通过新息的计算减少观测数据间的冗余进而增加了潜在分量的非高斯性。实验中利用具有弱相关性的图像信号进行仿真，通过与传统算法的比较证明了新方法能有效提高收敛速度和收敛精度。

关键词: 独立分量分析, 新息模型, 时间滤波, 峭度