针对地区发展不平衡现象导致DMSP/OLS数据提取城市建成区时全局最佳阈值分割法精度不高的问题，基于DMSP/OLS非辐射定标夜间灯光平均强度数据提出一种分区最佳阈值分割方法进行城市建成区面积的估算。利用陕西省1992、1997、2002、2007及2012共5个年份的DMSP/OLS数据进行验证，结果表明：与统计数据相比，该方法对各年份建成区面积估算误差均小于1.27％；与全局最佳阈值法相比，其平均误差降低0.49％。而且，该方法估算的建成区面积变化速率与统计数据基本一致。%For a low accuracy problem from those methods based on an globally optimal threshold using DMSP /OLS data to extract urban time-spatial change information due to unbalanced development of different city groups for a given study region ,a novel thresholding segmentation method by minimizing errors from each sub -region of study region was proposed to extract urban built-up area and change information of urban using a long time series DMSP /OLS averaged nighttime light intensity data without radiometric calibration .In the experiment , five time node DMSP/OLS data in Shaanxi province in 1992、1997、2002、2007 and 2012 was used to verify the proposed method .The conclusions can be drawn that the proposed method can achieve less error than 1.27%in estimating urban built-up area for each year com-pared to statistical data ,the proposed method has an error rate averagely 0.49% less than the global method .Also the extracted change information using the proposed method is in line with the actual situation .
展开▼
