摘要
基于像斑的变化向量分析法CVA(Change Vector Analysis)过分依赖像斑的灰度均值信息,而未能有效利用其灰度分布信息,这在高分辨率遥感影像变化检测中存在不足。本文提出了一种基于像斑直方图相似性测度的变化检测方法。利用G统计量构建不同时期像斑之间的相似性测度。假设所有像斑的相似性测度值符合混合高斯分布模型,通过期望最大化算法EM(Expectation Maximization)求解相关参数,最后采用基于最小错误率的贝叶斯判别规则获取最终的变化结果。实验表明,本文提出的上述方法能够有效提高变化检测的精度。
The object-oriented change vector analysis method, which is excessively dependent on the mean value of each object but failed to use gray distribution information, is deficient in change detection using high-resolution remote sensing images. A new method introducing similarity measurement of object histogram is proposed in this study. First, the similarity measurement of objects between different periods is built up by G statistic. Second, the Expectation Maximization (EM) algorithm is used to calcu- late the related parameters according to the assumption that all similarity measurement values of objects fit a Gaussian Mixture Distribution model. Finally, the Bayesian rule with minimum error rate is applied to get the change/no change results. Experimental results show that the method can get results with higher accuracy in change detection, especially for high-resolution remote sensing
出处
《遥感学报》
EI
CSCD
北大核心
2014年第1期139-153,共15页
NATIONAL REMOTE SENSING BULLETIN
基金
国家自然科学基金(编号:41101412)
中央高校基本科研业务费专项基金资助(编号:3101009
20102130201000139)~~
关键词
像斑
变化向量分析
变化检测
G统计量
EM算法
object-oriented, change vector analysis, change detection, G statistic, expectation maximization