伪不变特征法在遥感影像归一化处理中的应用

提出了一套伪不变特征法（PIF）归一化处理多时相遥感影像的技术流程和方法: 即参考影像选择,参考影像大气辐射校正,伪不变特征点选取,回归模型建立及结果评价,并对处理流程中的各个环节加以详细说明。其中在样本点选取过程中尝试把人工直接选取伪不变特征点应用于实践,并用主成分分析法对处理结果进行评价,评价结果表明人工直接选取伪不变特征点,对多时相遥感影像进行归一化处理是可行的,达到了消除季节性地物辐射变化、太阳光照条件差异、气象条件（大气散射、吸收和云量变化等）波动大等因素的影响,处理后的影像能满足建立模型的要求。



Application of the Pseudo-invariant Feature in Normalization Process of the Remote Sensing Images

In this paper,a method to normalize the multi-temporal remote sensing images using pseudo-invariant feature was developed. The main processes include the selecting of reference image, correction of atmospheric radiation, the selecting of Pseudo-invariant feature points, establishing the regression model and estimating the result. The operating procedure was introduced in detail. The pseudo-invariant feature points chosen with artificial method were used in practice. The results were evaluated with principal component analysis. The result of evaluation indicated that the artificial method for choosing Pseudo-invariant feature points is feasible in the normalization process of the Multi-Temporal Remote Sensing Images; The method reduced the influences of some factors such as the seasonal variety of surface radiation, the difference of lighting condition and the great fluctuation of meteorology condition (atmospheric scattering and assimilating, change of cloud cover). The processed images can meet the demands of the model.