一种基于独立分量分析的模糊图像盲分离算法

利用独立分量分析（ICA）的不完整自然梯度算法对因混合而引起的多幅模糊灰度图像进行盲分离，并针对算法中的非线性函数与源信号概率分布密切相关，而源信号的分布却是未知的先验信息的问题，利用算法输出信号的峰度对非线性激活函数进行自适应选择，提出了一种改进的自适应不完整自然梯度算法，并将其应用于模糊图像的盲分离，分析了不同混合矩阵对本文算法恢复原始灰度图像的影响及算法性能。仿真结果证明了本文算法与经典的FastICA算法相比，计算耗时更少、性能指标明显优越。

Blind separation algorithm of blurred image based on independent component analysis

The original images were restored from the blurred grayscale images by using the nonholonomic natural gradient (NNG) algorithm of the Independent Component Analysis(ICA) methods, and the principle of natural gradient algorithm under nonholonomic constrain was analyzed. However, the nonlinear activation function of this algorithm is closely related to the unavailable probability distribution of the sources, though it is robust to nonstationary and strongly undulate sources. To solve this problem, the nonlinear function was selected adaptively by use of the kurtosis of the output signals, an improved adaptive NNG (ANNG) blind separation algorithm of blurred image based on ICA was proposed, and the effect of the different mixture matrices on the performance of this algorithm was researched. The simulations show the validity of the proposed method. Compared with the nonholonomic natural gradient algorithm and the classical FastICA algorithm, the performance index of this proposed algorithm is better.