一种剪切波域的稀疏分量分析方法

发布时间：2018-08-21 13:20

【摘要】：在图像盲源分离应用中,传统独立分量分析(ICA)方法直接使用图像混合源作为输入求得解混矩阵,没有利用图像在变换域的稀疏特性,无法获得较好的解混效果.针对这一问题,笔者在分析剪切波具有良好稀疏表示能力的基础上,提出了一种剪切波域的稀疏分量分析方法.该方法首先将图像混合源变换到剪切波域,得到剪切波系数,接着使用峭度选择最稀疏系数,最后将稀疏系数作为ICA方法的输入实现图像分离.由于选择较少的稀疏系数,问题的求解复杂度有了显著的降低.实验结果表明,与传统ICA方法相比,该方法获得了更好的分离效果,且缩短了算法的运行时间.
[Abstract]:In the application of blind image source separation, the traditional independent component analysis (ICA) method directly uses the image mixing source as input to obtain the unmixing matrix, and can not obtain a better unmixing effect without taking advantage of the sparse property of the image in the transform domain. In order to solve this problem, a sparse component analysis method in shear wave domain is proposed based on the analysis of the good sparse representation of shear wave. In this method, the image mixing source is first transformed into the shear wave domain to obtain the shear wave coefficient, and then the kurtosis is used to select the sparse coefficient. Finally, the sparse coefficient is used as the input of the ICA method to separate the image. Because less sparse coefficients are chosen, the complexity of solving the problem is significantly reduced. The experimental results show that compared with the traditional ICA method, this method has better separation effect and shortens the running time of the algorithm.
【作者单位】：西安电子科技大学计算机学院;
【基金】：国家自然科学基金资助项目(61101248,61070143) 陕西省科技计划资助项目(2011K06-39)
【分类号】：TN911.7

【参考文献】