一种基于后非线性盲分离的图像解混技术研究-《计算机技术与发展》

文章信息/Info

Title:: Research on Image Unmixing Technology Based on Post-nonlinear Blind Separation

Author(s):: LIU Ji-cheng; BAI Hui; YAO Jian-hong; Electrical Information Engineedng Institute, Northeast Petroleum University

摘要:: 文中将一种后非线性盲分离算法应用于图像解混，该算法不需要额外的附加源信号信息，实现了非线性混合图像的全盲分离。首先，对后非线性混合模型进行微分变换，形成如同线性瞬时混合模型的形式，经论证源信号的微分形式仍保留了源信号的统计特性，达到简化的目的；其次，依据信号的相关特性来建立相应的目标函数及其递推方式，实现盲信号分离目的；最后，通过仿真试验来验证文中算法的有效性、可行性。实验证明，所采用的算法计算量小、收敛速度快、分离指标高，实现了混合图像的全盲分离，扩大了盲分离算法在图像解混技术中的应用范围及影响

Abstract:: In this paper,a nonlinear blind source separation algorithm is applied to image unmixing,the algorithm does not require additional information about the source signal,realizes the purpose of the nonlinear mixed image blind separation. Firstly, the model of post-nonlinear mixture is transformed into the model like the instantaneous linear mixture by the differential transformation,which makes the nonlinear problem simplified largely, then it is proofed that the differential of source signals conserves the same statistical characters as the source signals. Secondly, the correlation characters of signals are utilized to construct separation criterion function and iterative equation for blind source separation. Lastly, the numerical computer simulations arc performed to illustrate the validity and feasibility of the algorithm. Experiment tests show that the algorithm possesses some good excellences like low computational complexity, rapid convergence, high separation indexes, realizes whole-blind source separation and expands application and effect of the blind source separation algorithm in image unmixing technology