基于小波变换和KICA算法的图像盲分离

Blind Separation of Image Based on Wavelet Transform and Kernel Independent Component Algorithm

盲源分离技术在污染图像恢复与重构中起着重要的作用。近年来出现了多种盲分离算法,在无噪声的情况下,KICA(核独立分量分析)的分离方法最好。但在有噪声的情况下,传统的方法对于有噪混合图像的分离效果不佳。针对这一问题,提出了小波去噪与KICA相结合的算法对有噪混合图像进行去噪分离。仿真实验结果表明这种方法能有效地降低噪声的影响,能较好地实现了图像的分离。

Blind sources separation technology plays a significant role in recover and reconstruction of the pollution image. In recent years,several algorithms of the blind source separation have been studied in which the kernel independent component algorithm（KICA） is the optimal one in case of noiseless. On the other hand,the conventional methods have poor performance for the de-noising separation of the mixed noised image. In order to resolve this problem,a algorithm combined the wavelet de-noising approach and the KICA technique is proposed to de-noising separate the mixed noised image. Finally,some simulation results are given to illustrate that the method can reduce the influence of the noise effectively,and achieve the better de-noising separation of the image.