摘要
以支持向量机关键部分核数的有效选定作为探究目标,从核函数选取规则着手,将遥感数据领域知识嵌入到核函数构造问题中,结合遥感数据本身特征提出一种能同时兼顾到样本亮度差异性和角度差异性的混合核函数选取方法,通过支持向量机传统核函数分类效果进行对比分析,表明混合核方法的有效性.
The technology of remote sensing image classification , selection of classification rules and the selection of kernel function affect the classification accuracy among the samples . The key part of the support vector machine（SVM ） kernel function is effectively selection as exploration target rules from the selection of kernel function . The remote sensing data knowledge embedded in the kernel function structure problems ,combined with the feature of remote sensing data itself can put forward a kind of both samples to the brightness difference and angle difference of the mixed kernel function selection method . The kernel function of SVM and traditional SVM classification effect are analyzed to show the effectiveness of mixed kernel methods .
出处
《西北师范大学学报（自然科学版）》
CAS
北大核心
2016年第3期49-56,共8页
Journal of Northwest Normal University(Natural Science)
基金
国家自然科学基金资助项目(61363066)
新疆高校科研计划重点研究资助项目(XJEDU2014I043)
伊犁师范学院院级重点基金资助项目(2015YSZD04)
吉林省科技发展计划资助项目(20120302)
关键词
支持向量机
遥感数据
核函数
光谱
support vector machine
remote sensing data
kernel function
spectrum
