Research of Blind Images Separation Algorithm Based on Kernel Space

Principle of blind source separation (BSS) and kernel function method is introduced. Kernel method is a kind of new learning algorithm concerned by many scholars. More excellent new algorithm can be got by kernelizing the original algorithm using kernel trick. Kernelized blind source separation algorithm based on second-order statistics are expatiated and a new blind images separation algorithm using the kernel trick originally applied in support vector machine (SVM) is proposed. The result of experiment on realistic natural images shows that the blind images separation algorithm based on kernel space can separate mixed natural images successfully.