基于混合多分类器结合算法的遥感分类  

Remote Sensing Classification Based on Hybrid Multi-classifier Combination Algorithm

作　　者：杨海波[1,2] 王宗敏[1,2] 张涛[1] 

机构地区：[1]郑州大学河南省信息网络重点开放实验室,郑州450001 [2]郑州大学水利与环境学院,郑州450001

出　　处：《计算机工程》2010年第11期173-175,共3页Computer Engineering

基　　金：河南省自然科学基金资助项目(0611051900)

摘　　要：为改善遥感影像分类精度,提出混合多分类器结合算法。考虑抽象级和测量级2个层次的特点,综合最优子分类器、Bagging算法和最大置信度区间法。应用到不同分辨率的遥感影像分类进行验证,结果表明,与选用的子分类器相比,该算法的总体精度和单个类别分类精度有明显提高,是有效的高中精度遥感影像分类算法。To improve the precision of remote sensing image classification, this paper proposes hybrid multi-classifier combination method. Taking the characteristic of abstract level and measurement level into consideration, the optimal sub-classifier, bagging algorithm and the most large confidence algorithm are combined. This method used in diffierent image classification shows a better enhancement, and results indicate that the hybrid multi-classifier combination algorithm is an effective algorithm for medium-high precision remote sensing image classification.

关 键 词：遥感分类 混合多分类器结合 抽象级 测量级 

分 类 号：TP75[自动化与计算机技术—检测技术与自动化装置]