利用独立分量分析和小波变换的特性,提出了一种基于小波变换的混合图像盲分离新算法。该方法首先通过小波变换将原始图像稀疏分解,然后采用基于分离度变步长的EASI算法对分解后的系数矩阵进行盲分离,最后根据盲分离确定的分离矩阵将原始图像恢复出来。仿真结果表明,该方法的分离精度高,分离速度快,稳定性强,在混合图像较多的情况下仍具有较好的分离效果。
Using the characteristics of independent component analysis (ICA)and the wavelet transform (WT), a new method based on WT to separate the mixed images blindly is proposed. Firstly,this method make the original images sparse decomposition by WT. And then use the variable step -size based on separation degree (SD)EASI algorithm to separate the coefficient matrix of sparse decomposition blindly. Finally, recover the original images ac- cording to the separation matrix got above. The simulation results show that the algorithm not only has high separation precision,fast separation speed and high stability, but also can achieve good separation effect even under more mixed images' case.