摘要
传统基于统计的高空间分辨率遥感影像变化检测需假设数据服从正态分布,如基于卡方检验的变化检测,但受困于样本数据不严格服从正态分布,检测效果并不理想。针对此问题提出一种基于箱线图的变化检测方法,不需任何假设,根据数据结构本身固有模型进行变化检测。首先,经由矢量引导的增长式分割获得对象;然后,结合纹理与光谱特征计算其余弦值;最后,采用箱线图算法获得变化对象。以高分一号融合影像为实验数据进行验证,结果表明,结合矢量夹角余弦与箱线图进行变化检测的总体精度可达88.21%,相比马氏距离与卡方检测的方法和结合马氏距离与箱线图的方法具有更好的准确性与稳定性。
The traditional statistics-based change detection method requires the prerequisite that the dataset should obey the Gaussian distribution,such as the iterative chi-square test based change detection method.However,the dataset does not strictly obey the Gaussian distribution,and hence the result is not ideal.A novel change detection method is proposed in this paper,which does not need any assumptions and can take change detection by its own structure.First,an incremental segmentation method is adapted to get objects.After that,spectral and contextual features are combined to calculate its cosine value.Finally,changed objects are found by the box-plot.High-resolution remote sensing images of GF-1 are used as the experimental data.The results are much better than the results of the traditional statistical object-based change detection.
作者
张春森
吴蓉蓉
李国君
崔卫红
冯晨轶
ZHANG Chunsen;WU Rongrong;LI Guojun;CUI Weihong;FENG Chenyi(College of Geomatics, Xi’an University of Science and Technology, Xi’an 710054, China;School of Remote Sensing and Information Engineering, Wuhan University, Wuhan 430079, China)
出处
《国土资源遥感》
CSCD
北大核心
2020年第2期19-25,共7页
Remote Sensing for Land & Resources
基金
国家自然科学基金项目“反射率与叶绿素荧光协同的小麦白粉病早期探测机理与方法研究”(编号:41601467)
陕西省自然基金项目“矿井巷道围岩变形监测摄影测量计算机视觉方法关键技术研究”(编号:2018JM5103)共同资助。
关键词
变化检测
高空间分辨率遥感影像
增长式分割
箱线图
change detection
high resolution remote sensing image
incremental segmentation
box-plot
作者简介
第一作者:张春森(1963-),男,博士,教授,主要从事摄影测量与遥感应用的研究,Email:zhchunsen@aliyun.com。
