The spatial and radiation resolutions of domestic satellites have been remarkably improved in recent years. The sensor design and geometric calibration technologies have improved with the development and application of stereo mapping satellites (ZY-3). However, the imaging systems typically use the linear CCD imaging technology. The long focal length and narrow field angle causes the geometrical model to have 3D parallel projection characteristics, and the satellite images will still have residual system errors caused by the drift error of satellite-borne GPS/IMU and asynchrony between the pose and trace of the satellite, especially after the rigorous on-orbit geometry calibration. A block adjustment technique is still required to meet the requirements of remote sensing monitoring and mapping application. Therefore, satellite images-based accurate positioning without ground control point information is the precondition to obtain the global geographic and resource environmental information and monitoring changes in the global resource environment. The accuracy of “ZY-3” satellite image is improved to 15 m after calibration and its internal precision is better than 1 pixel. After the overall block adjustment without GCPs, the plane and elevation accuracy of the image can be improved to 5 m (medium error). In this study, on the basis of the widely used optimization method of solving the constraint problem in photogrammetry–alternating direction method (ADM) and RFM least-squares block adjustment, we propose a GCP-independent block adjustment method for large-scale domestic high-resolution optical satellite images–GCP-independent satellite imagery block adjustment (GISIBA) based on the geometric features of ZY-3 satellite images. The proposed method is highly efficient and easy to parallelize. GISIBA of satellite images can be considered as the overall block adjustment of the multi-source optical satellite imagery with specific constraints (distance, angle, etc.) when no ground control points are available. Most GCP-independent adjustments use a form of virtual control points. However, the precision of these virtual control points is low (varied system error), and the precision is inconsistent in the measurement area. Thus, the GCP-independent adjustment is a type of block adjustment under different precision controls. The law of error propagation of this approach is complex, and gross error detection and positioning are difficult to perform using this approach. This study presents an “average” virtual control point-based stereoscopic GCP-independent block adjustment method for large-scale satellite image GISIBA. On the basis of the automatic and reliable acquisition of uniformly distributed image tie points, the method comprehensively uses the “ADM” introduced from aerial photogrammetry and RFM-based least squares adjustment algorithm to realize the combined block adjustment of satellite images. First, the “ADM” is used to solve the initial values of the unknowns and to perform automatic detection and elimination of the above medium-scale gross error based on the parallel processing platform. All unknowns are assigned priori weights based on the results. Next, the RFM-based least square method is used to solve the large-scale reformation normal equation to obtain the orientation parameters with high-precision, which meets the production requirement of high-precision image products. Block adjustment by constructing virtual “average” control points addresses the “rank” problems in the GCP-independent adjustment and improves the state of normal equation of the block adjustment system, which benefits the stability and fast convergence of the block. Moreover, the method makes it convenient to analyze the relationship among the data coverage, imaging time interval, and satellite image GCP-independent block adjustment. In addition, parallel processing based on the OMP parallel method is used to realize the parallel processing of the “ADM” and multi-thread parallel computing based on least-squares adjustment to ensure the efficiency of block adjustment. We used multiple sets of the ZY-3 satellite image data in typical regions to verify our method. The following experiment results are summarized as follows: 1) On the basis of the widely used optimization method of constraint problem called the “ADM” and RFM least-squares block adjustment, the proposed GISIBA method is easy to parallelize and is highly efficient in terms of reliability, accuracy, and performance of the developed procedure. 2) In this method, virtual “average” control points are built to solve the rank defect problem and qualitative and quantitative analyses in block adjustment without control. Assuming the positioning accuracy is located on the same number order (such as 50 m), the final positioning accuracy of satellite image must be improved after GCP-independent block adjustment by using the virtual “average” control points. The final positional accuracy is stronger than the worst initial positioning accuracy of the original image. Furthermore, the increase on the coverage of satellite images does not consistently improve the overall positioning accuracy. However, the use of considerable high-resolution satellite images to cover the same area improves the positioning accuracy after the final block adjustment in the statistical sense. The horizontal and vertical accuracies of multi-covered and multi-temporal satellite images are greater than 6 m and 5 m, respectively. 3) The mosaic problem of adjacent areas in large area DOM production can be solved when third-party geographic information data are introduced as horizontal and vertical constraints. This approach is considered as weak-sense auxiliary control in the block adjustment process.

GCP-independent orientation;Rational Function Model (RFM);virtual control points;block adjustment;alternating direction method;ZY-3 satellite image