一种有效的减弱配准误差影响的LUCCD方法

利用多时相遥感图像进行土地利用便盖变化检测（LUCCD），其结果的精度受到很多因素的制约，而对图像进行空间位置配准时的残余误差就是其中最重要的因素之一。本文从统计学的观点出发。在对配准误差的分布特点进行统计分析的基础上，提出一种传统的变化矢量分析（CVA）与配准误差的分布特征相结合进行土地利用便盖变化检测的方法，并用某地区的多时相TM数据对该方法的有效性进行了验证。结果表明，该方法可以有效地降低漏检率和错检率，大大提高了变化检测的精度。

An effective LUCCD method to decrease the effect of registration error

The accuracy of Land Use/Cover Change Detection (LCCCD) using multitemporal remote sensing images is affected by a few of factors, of which one is registration error. On the basis of analysing the distribution of registration error by statistic theory, this paper proposes a method which combine the traditional Change Vector Analysis (CVA) and the distribution of registration error, and testify its effectiveness with local area multitemporal TM image. The result indicates that the method can decrease the ratio of miss and false detection, so the accuracy of LUCCD is improved greatly.