为了提高植被变化检测精度和获取植被变化类型信息,在综合利用像素级变化检测、特征级变化检测和分类后变化检测的基础上,提出了一种由粗到精并准确获取变化信息的植被变化检测方法。文中选取2002 年和2007 年的QuickBird多光谱图像作为试验对象,使用像素级变化检测处理多时相遥感数据,确定出包含植被变化的候选区域;然后,使用特征级变化检测确定出变化区域;最后,结合分类后变化检测获取变化信息的位置和类型。通过对两时相遥感数据进行变化检测试验,并与各单独检测方法对比,验证新方法的有效性。结果表明:该方法在提高了变化检测精度的同时能够提供变化地物信息,可以有效地应用于植被变化信息的检测。

In order to improve the accuracy of change detection and obtain the information of vegetation changes,we proposed a method of obtain accurate from coarse to fine on the basis of utilization of pixel level change detection, feature-level change detection and after classification change detection. 2002 and 2007 QuickBird multispectral image is selected as a test object, using pixel-level change detection process the multi-temporal remote sensing data to determine the candidate region including vegetation changes, and then using feature-level change detection to determined the changes in the region. Finally the location and type of changes information are obtained by using after classifition detection change. By the test of two remote sensing data change detection, and compare with the individual detection method, the validity of the new method is verified. The results showed that: new method could improve the detection accuracy changes while providing feature information changes, and could be effectively applied to vegetation change detection information.

TP751

国家重大科学仪器设备开发专项项目(2012YQ24012705);中煤科工集团科技创新基金面上项目(2014M5030)