全球地表覆盖遥感制图与关键技术研究项目要求对两个基准年度（2000年、2010年）全球30 m分辨率的多光谱遥感数据进行辐射处理和几何精纠正处理，为地表覆盖制图完成数据准备。数据以Landsat TM/ETM+为主，HJ-1A/B CCD数据为补充，共计2万多景影像需要进行辐射处理，有1000多景HJ-1A/B CCD影像需要几何精纠正。如此大规模的数据处理，自动化处理是必然的选择。本文介绍了HJ-1A/B CCD图像几何精纠正自动化实现中关键问题的解决方法和精度评价结果，Landsat TM/ETM+和HJ-1A/B CCD图像自动化辐射校正中关键问题的解决方法和精度评价结果，以及大规模的数据处理活动引发的一些思考。

"Higher Resolution Global Land Cover Mapping" program requires global data from two base years (2000,2010) with 30 m spatial resolution multispectral images for global land cover mapping. These data should undergo radiometric and geometric precision correction. The data mainly comprise Landsat TM/ETM+ images supplemented by HJ-1A/B CCD. In total, more than 20000 images require radiometric correction and more than 1000 HJ-1A/B CCD images require precise geometric correction. With such a large data set, automated processing is necessary. This paper presents a solution to key issues on the automatic geometric correction of HJ-1A/B CCD images and precision evaluation results. This paper also presents solutions to key issues on the automatic radiometric correction of Landsat TM/ETM+ and HJ-1A/B CCD images and accuracy assessment results. Finally, this paper presents some suggestions for the processing of large data sets.