摘要
基于小波变换提出了一种特征提取及特征选择的方法.通过对小波系数绝对值化,减小了特征的分布范围;对小波进行自适应的过滤提取了目标的主要特征,增加了特征的聚类程度.本文通过SVM分类器对该方法进行验证并与其他方法比较.实验证明该方法有效的提高了目标的识别率,降低了误识别率.
In this paper, a method of feature selecting and feature extractihg based on wavelet was proposed. Using this method the range of wavelet coefficients can be reduced by getting the absolute value of the coefficient; clustering the coefficient by filtering them with adaptive threshold. This method is tested by SVM classifier. Experimental results show that this method is effective for object recognition.
出处
《沈阳理工大学学报》
CAS
2006年第5期35-38,共4页
Journal of Shenyang Ligong University
关键词
小波变换
SVM
特征提取
wavelet transform
SVM（ support vector machine）
feature extracting
