Principle and method of GF-7 satellite integrated processing of stereo image and laser data
The GF-7 satellite is the world's first Earth observation satellite synchronously equipped with an optical stereo camera and an operational laser altimeter,which can simultaneously obtain submeter resolution dual linear-array stereo images and sparse ground Laser Altimetry Points(LAPs).LAPs can be used for improving the elevation accuracy of stereo images to ensure that the GF-7 satellite can be applied to the global 1∶10,000-scale stereo mapping.On the basis of the error propagation principle of the working processes of satellite cameras and laser altimeter,the basic principles of improving the elevation accuracy of stereo images by utilizing LAPs was first analyzed in this study.Then,an integrated processing method of GF-7 satellite stereo images and LAPs,including the construction of precise measurement method of LAPs on stereo images and the combined block adjustment model of LAPs and stereo images,was designed.Finally,the elevation accuracy of the stereo images was effectively improved.A total of 70 stereo images and 463 LAPs of the GF-7 satellite in the northern region of Hebei Province,China,were selected to conduct integrated processing experiments.Results show that the vertical Root Mean Square Error(RMSE)of stereo images in the flat,hilly,mountainous,high-mountainous,and entire regions was reduced from the original 3.0,4.68,2.86,2.48,and 3.19 m to 0.35,0.66,0.74,0.91,and 0.68 m,respectively,and the horizontal RMSE of stereo images was 4.99,3.52,4.42,5.99,and 4.82 m,respectively.These results reveal that the horizontal and vertical accuracy of the GF-7 stereo images satisfied the accuracy requirements of China's 1∶10,000-scale mapping.At present,the integrated processing software system of GF-7 satellite stereo images and LAPs constructed by the method described in this study has achieved business application,which is vital for the future construction of global 1∶10000-scale geographical information resources.
NETL
NSTL
万方数据
