Abstract: This paper introduces the basic principle of ICA，elaborates the natural gradient algorithm and the methods of ICA wavelet.With the idea of Performance Index (PI)，it sets an improved criterion to judge whether the objective function L(w) is optimal.And based on this standard，it gets a new step-size adaptive method，by using hyperbolic tangent function.Through computer simulation，it confirms the superiority of its various aspects.

Key words: step-size adaptive, Natural Gradient Algorithm（NGA）, Wavelet Independent Components Analysis（WICA）, hyperbolic tangent function, image blind separation

摘要： 介绍了独立分量分析基本原理，阐述了自然梯度算法以及小波ICA方法。采用性能指数（PI）的思想，设定一个新的评判目标函数[L(w)]达到最优的标准，并且根据此标准，提出了一种改进的基于双曲正切函数步长自适应方法。通过计算机仿真实验，证实了其各方面的优越性能。

关键词: 步长自适应, 自然梯度算法（NGA）, 小波ICA（WICA）, 双曲正切函数, 图像盲分离