SRTM约束的无地面控制立体影像区域网平差

SRTM-aided Stereo Image Block Adjustment without Ground Control Points

针对SRTM(shuttle radar topography mission)数据在平坦地形或局部区域的高程精度远远高于其标称精度的特点,研究设计了一种无地面控制条件下利用SRTM作为高程约束的立体区域网平差方法。通过构建一个较大范围区域网并匹配密集连接点,将SRTM作为连接点物方高程初值,并在平差解算过程中确保分布于地形平坦区域(根据经验,在该类区域SRTM精度较高)的连接点的物方高程严格趋近SRTM高程,最终实现大范围区域内影像高程精度的整体提升。通过以覆盖湖北省全境的资源三号卫星三线阵立体影像作为试验影像的试验验证表明,采用该平差方案,在无地面控制点条件下资源三号立体影像的高程中误差从7.2m提升到2.0m,其中地形平坦区域高程中误差1.44m,山地区域高程中误差3.0m,达到了我国1∶25 000比例尺测图应用的高程精度要求。

With regard to the feature that the vertical precision of SRTM（shuttle radar topography mission） data in the flat terrain area is much higher than its nominol precision, it is proposed that o stereo block adjustment algorithm without ground control by adopting SRTM data as elevation constraint. In order to improve overall vertical accuracy of the image, by constructing a wide local area network and setting intensive connecting points, SRTM is set as the initial vertical volue of connecting points. In the process of adjustment calculation, ground height of connecting points distributing in the flat terrain area is made sure to rigorously approach to vertical occurocy of SRTM. Through the experiment employing ZY-3 triple linear orray stereo satellite images cover across all of Hubei province, the verification results prove that the method is effective. The height root mean square errors（RMSEs）of ZY-3 stereo images is improved from 7.2 m to 2.0 m without ground control, and the RMSEs of flat areas and mountain areas improved to 1.44 m ond 3.0 m respectively, which fulfills the elevation accuracy requirement of notional 1 - 25 000 scale mopping appl ications.