并联结构组合分类器的误差分析  

The Errors Analysis of Combined Classifier based on Parallel Structure

作　　者：郑忠[1] 曾永年[1] 刘慧敏[1] 徐艳艳[1] 于菲菲[1] 

机构地区：[1]中南大学地球科学与信息物理学院,中南大学空间信息技术与可持续发展研究中心,湖南长沙410083

出　　处：《遥感技术与应用》2011年第3期340-347,共8页Remote Sensing Technology and Application

基　　金：国家自然科学基金项目(40771198);湖南省自然科学基金项目(08JJ6023)

摘　　要：虽然遥感图像分类器发展迅速,但是单分类器分类精度仍然不能满足实际应用的需要,因此组合分类器便成为遥感分类技术研究的一个重要方面。与串联结构相比,并联结构是实际应用过程中研究较早、发展较充分和应用最广泛的,因而主要就并联结构组合分类器的误差进行分析。通过理论分析得出:并联结构组合分类器的精度变化与单分类器的误差集合分布模式有关。相离时,其精度提升幅度最大;相交时,组合分类器精度得到提升,其精度提升幅度大小与组合后的误差像元集合的大小成反比;相包含时,其精度位于两单分类器之间,且更靠近精度较高的单分类器的精度。同时以长沙市局部区域为实验区进行了验证性的实验,实验结果有效地验证了并联结构组合分类器的误差分析所得到的推论。最后从理论上讨论了组合分类器提高遥感图像分类精度的可行性,并指出单个类别的组合分类结果也与单分类器对这个类别的误差集合分布模式有关,为组合分类器发展提供了一个较好的突破方向。The classification algorithm of remote sensing image rapidly develops,as the classification accuracy of single classifier still can＇t meet the needs of practical application,the combined classifier becames an important aspect of the remote sensing classification.There are various types of combined classifiers.Compared with serial structure,Parallel structure is early studied,fully developed and widely used.This paper analyzed classification accuracy of the combined classifiers based on parallel structure.The results indicated that the precision of combined classifier is related to the position of incorrectly classified pixels in each single.While the incorrectly classified pixels by single classifiers are separated,the accuracy of combined classifier is the highest;while the incorrectly classified pixels by single classifiers are intersected in classified results,the precision of combined classifier is higher than that of each single classifier and the improvement is inversely proportional to the size of error set in combined classified results;while the incorrectly classified pixels by one classifier are included in another classifier,the accuracy of combined classifier is located in between the high and low accuracy,which is near to the higher one.This paper also experiments in Changsha local area,and the result effectively tested the deduction which we get in the analysis.Finally,this paper discussed theoretically the feasibility of improvement which combined classifier can get in the remote sensing image classification and provides an effective way to improve the performance of combined classifier.

关 键 词：组合分类器 分类精度 土地利用 LandsatTM 

分 类 号：TP75[自动化与计算机技术—检测技术与自动化装置]