摘要
在研究核函数以及PCA(主成分分析)理论的基础上,提出了一种将多核函数的PCA应用于板形图像降维的方法,并使用遗传算法来确定多核模型中的关键参数,最后将多核PCA方法与PCA方法进行比对,证明了其降维性能优于普通PCA方法.
On the basis of extensive study on kernel and its relevant theories,a multi-kernel PCA method was designed in this thesis to integrate application areas and characteristics of a number of kernels,and use genetic algorithm to definite the key parameters of the model.At last,compare the multi-kernel PCA to PCA,and illustrate the effect ion of multi-kernel PCA is better than PCA.
出处
《陕西科技大学学报(自然科学版)》
2010年第6期69-71,共3页
Journal of Shaanxi University of Science & Technology
关键词
矫直机
核函数
KPCA
降维
tension leveler
kernel
KPCA
dimension reduction