“天绘一号”卫星三线阵CCD影像自检校区域网平差

Self-calibration block adjustment for Mapping Satellite-1 three linear CCD image

遥感卫星在空间环境运转过程中,星敏感器的姿态角、成像相机的镜头和CCD等几何参数会发生不可预估的变化,从而对卫星影像的定位精度产生影响。因此,对卫星影像进行自检校区域网平差处理是实现卫星影像精确定位的一项关键技术。本文以"天绘一号"卫星为研究对象,首先分析了"天绘一号"卫星三线阵立体测绘相机的镜头和CCD几何形变,并提出了适用于"天绘一号"卫星三线阵CCD影像的自检校模型;然后建立"天绘一号"卫星三线阵CCD影像的自检校区域网平差模型,对外方位元素和自检校参数进行整体平差,消除存在于外方位元素观测值和自检校标定值中的系统误差,以提高卫星图像定位精度。最后利用嵩山实验场对"天绘一号"卫星三线阵CCD影像进行了自检校区域网平差处理,验证了自检校模型和自检校区域网平差模型的正确性和有效性,并分析研究了不同数目的控制点条件对自检校区域网平差的影响。实验结果证实采用自检校区域网平差技术可以有效地消除系统定位误差,显著提高定位精度。

Launched in 2010, ＂Mapping Satellite-1＂ （TH-1） is the first stereo surveying and mapping satellite from China. During thereal operation of TH-1 in the space environment, the attitude angles of the star sensor, lens geometric parameters, and CCD undergo unpredictable changes that affect the positioning accuracy of satellite images. This study conducts serf-calibration block adjustment research on TH-1 satellite images to achieve high image positioning accuracy. The following investigation method is used in this study： First, the lens and CCD geome^c distortions of the three linear stereo surveying and mapping cameras of TH-1 are analyzed, and a serf-calibration model suitable for the satellite＇s three linear CCD images is proposed. The self-calibration block adjustment model is built for the satellite＇s three linear CCD images. Overall adjustment is then applied on exterior orientation elements and serf-calibration parameters to eliminate systematic errors present in the above observations and to improve the position- ing accuracy. Finally, the Songshan testfield is utilized to support the serf-calihration block adjustment processing for the three linear images of TH-1 and to verify the correctness and validity of the serf-calibration and serf-calibration block adjustment models. The influence of the number of control points on serf-calibration is also analyzed. Experimental results demonstrate that using the serf-calibration block adjustment technique can effectively eliminate systematic positioning errors and significantly improve positio- ning accuracy. The spatial resolution of the three linear array images of TH-1 is 5 m. After normal block adjustment, the optimal plane accuracy along the X- and Y-directions are approximately 20 m, with a height accuracyofapproximately 20 m and plane accu- racy of 4 pixels. After serf-calibration block adjustment, the optimal plane accuracy increased from 5 m to 6.5 m （ i. e. , less than 1.3 pixels）, and the height accuracy increased to 5 m. Through a series of experimental research and analysis, the following con- clusions can be drawn ： （ 1 ） The auxiliary orbit and attitude determination files of the three linear CCD images of TH-1 contain clear systematic errors. A certain degree of geometric deformation also exists in the camera lens and CCD component. For direct space intersection using orbit and attitude determination files, a prominent systematic positioning error must be eliminated through the block adjustment approach. （2） If the serf-calibration parameter X, and its observation equation are not considered, Eq. （10） becomes the normal block adjustment model. Normal block adjustment considers only the control point errorXs and exterior orienta- tion element error X; thus, it corrects only the attitude errors. Although normal block adjustment can effectively eliminate the posi- tioning error caused by the attitude angle error, the impact of the lens and CCD component geometric deformation parameters still exists. The variation of the latter parameters is minor in a medium period, and the residual errors in the plane and elevation direc- tion are still systemic. （3） Applying serf-calibration block adjustment to the three-line array images of TH-1 with the serf-calibra- tion model presented in this paper can effectively eliminate the attitude errors and the combined effects of the focal length change, lens, and CCD geometric deformation and significantly improve positioning accuracy. The residual systematic errors of the check- points are effectively eliminated after serf-calibration adjustment, which confirms the correctness and effectiveness of the serf-cali- bration parameter and serf-calibration block adjustment models presented in this paper. （4） Increasing the control points can effi- ciently improve the accuracy of serf-calibration block adjustment in an appropriate range, but when the number of control points reaches a certain degree, the adjustment accuracy stabilizes.