首页 > , Vol. , Issue () : -
Objective: Using domestic satellite for global geographic resources construction, one of the most important promise is to improve the geometry accuracy of satellite imagery without ground control points (GCPs). Ziyuan3-03 (ZY3-03) satellite, launched on July 25th, 2020, was the third high-resolution stereo mapping satellite of Ziyuan3 series. And it mainly used for the global application of 1:50000-scale mapping. Triple linear-array push-broom panchromatic cameras (resolution<=2.5 m) and one multi-spectral camera (resolution =5.8 m) were loaded on this satellite. Moreover, in order to improve the vertical accuracy, an additional single-beam laser altimeter was loaded on the ZY3-03 satellite. This addition enables the satellite to simultaneously obtain high vertical accuracy laser altimetry points (LAPs). Integrating stereo images and LAPs to improve the vertical accuracy of the stereo images is considered the key to realize the application of 1:50,000-scale stereo mapping of ZY3-03 images with sparse or no GCPs. Method: This study proposed a combined block adjustment method using ZY3-03 triple linear-array stereo images and synchronous orbit LAPs to improve the elevation accuracy of the images. First, a method able to accurately obtain the image coordinates of LAPs on the stereo images was designed. By fully using this advantage that the relative planar error between LAPs and the synchronous nadir image was minimal, the image coordinate of LAPs on synchronous orbit nadir images could be correctly acquired through the image rational function model (RFM) and ground geodetic coordinates of the LAPs. Then the accurate pixel coordinates of the LAPs on triple linear-array stereo image were obtained by high-precision image matching between the forward/backward image and nadir image. Subsequently, using the whole orbit stereo image as the operation unit, a combined block adjustment model based on RFM and combined block adjustment strategy were constructed. Result: Twelve ZY3-03 satellite triple linear-array stereo image pairs and 81 synchronous orbit LAPs of plain and hilly terrains located in Heilongjiang province (China) were selected as experimental data. Simultaneously, 270 Global Positioning System (GPS) points were collected as checkpoints. Through combined adjustment with stereo images and LAPs, the vertical root mean square error (RMSE) of the images was reduced from 5.27 m to 2.58 m. Seven ZY3-03 satellite triple linear-array stereo images pairs and 6 synchronous orbit LAPs of mountainous terrains in Hebei province were selected as experimental data, with 115 GPS points as checkpoints. After the combined block adjustment, the vertical RMSE of images was reduced from 11.25 m to 4.45 m. The experimental results showed that the vertical accuracy of ZY3-03 images was enhanced significantly, satisfying the precision requirements of 1:50,000-scale stereo mapping in China. Conclusion: The experiments discussed above demonstrated that the combined block adjustment method proposed in this study was effective regardless of whether the terrain is flat, hilly, or mountainous. And the vertical accuracy of stereo images can be greatly improved so that it satisfies the accuracy requirements of 1:50000-scale stereo mapping in China.