摘要
本文从遥感动态监测实际出发,提出利用图像差值与质量控制相结合的方法选取不变特征点(PIF),较好地克服了样本选择的主观性,得到样本的相关系数达到98%以上,采用最小二乘法求解线性关系式,对待校正图像的灰度值进行线性变换,使两幅图像上同名地物具有相同辐射值,实现遥感动态监测中多时相遥感图像的辐射归一化。
Based on the current situation of dynamic monitoring of remote sensing, we select the PIF by the way of combining the image differencing and quality control. In this way, influence of the subiectivity of sample selection can be well handled, and the correlation coefficient of samples is up to 98%. Method of least squares is adopted to solve the linear equation, we make linear transformation to the Gray-value of the pre-revised images in order that the same object in two different images has the same radiation value, then accomplish the relative radiometric normalization of multitemporal images in RS dynamic monitoring.
出处
《遥感信息》
CSCD
2008年第4期71-75,共5页
Remote Sensing Information
基金
863课题“面向地块的地物类型精细识别技术及其应用(2007AA12Z181)”
国家科技基础条件平台工作重点项目“森林灾害的监测、预警与管理系统平台(2003DIA6N007)
关键词
动态监测
光谱归一化
伪不变特征点
线性变换
均方根误差
dynamic monitoring
radiometric normalization
pseudo invariant features
linear transformation
root mean square error
