Abstract: An effective method for strong noise pollution control in multiple copy images is proposed. If the noise and original image are mutually independent in mixed signals, the mixed signals are decomposed to an original image component and a noise component using ICA. The original image component is selected to reconstruct the result image according to standard deviation of its time course. Experimental results show that the proposed method can separate nonzero-mean noise Gaussian noise from multiple copy images, and is effective in the case of stable and unstable noise intensity.