摘要
提出了基于UWT(非抽样小波变换)去噪与Fast ICA(快速独立分量分析)算法相结合的含噪盲源分离方法,采用先去噪后分离的方式实现了在加性高斯噪声环境下混合图像的盲分离。仿真结果表明,该方法能很好地从加性高斯噪声中分离出源图像,与曲波阈值去噪后的Fast ICA方法相比较,该方法能获得更好的峰值信噪比。
This paper proposes a method to realize noisy blind source separation based on UWT(Undecimated WaveletTransform)denoising and FastICA(Independent Component Analysis). The method employs a model of ICA after denoisingto implement noisy image separation under the environment of additive Gaussian noise. The simulation results show thatthe proposed method can separate noisy mixed images efficiently. Compared with the method based on curvelet denosingbefore ICA, the proposed method can obtain better performance of Peak Signal-to-Noise Ratio(PSNR).
作者
蔡伟华
何选森
CAI Weihua;HE Xuansen(College of Information Science and Engineering, Hunan University, Changsha 410082, China)
出处
《计算机工程与应用》
CSCD
北大核心
2016年第16期180-185,共6页
Computer Engineering and Applications
关键词
盲源分离
非抽样小波变换(UWT)
快速独立分量分析
曲波变换
峰值信噪比
blind source separation
Undecimated Wavelet Transform(UWT)
Fast Independent Component Analysis (FastICA)
curvelet transform
Peak Signal-to-Noise Ratio(PSNR)
