| Popis: |
Using stereo synthetic aperture radar (SAR) to extract 3-D information has become an important means of continuously improving the spatial resolution and geometric positioning accuracy of spaceborne SAR. However, owing to the large differences in a stereo configuration, such as intersection angle (ITA), it is difficult to guarantee localisation accuracy. This study proposed a method for analysing the stereoscopic intersection capability and verifying the accuracy of spaceborne SAR images. The method is improved from three aspects. First, it considers the complexity of the OPSS and SASS modes and discusses the two cases separately in the theoretical modeling and experimental analysis. Secondly, in the theoretical modeling stage, the error is no longer simplified as originating from only one image, but all images have errors. Finally, the validity of the modeling conclusions is verified using multiple sets of simulated experiments and real SAR experiments. Experiments show that stereo SAR has more potential than single SAR. In the opposite-side stereo (OPSS) mode, the planimetric positioning accuracy of 35.028 m (single SAR) can be improved to 3.132 m (stereo SAR, optimal 1.641 m). By designing a reasonable stereo configuration, the geometric positioning accuracy of the original data can be significantly reduced by 31.896 m. In addition, the results show that, (1) for the symmetric OPSS mode, the planimetric accuracy can be improved, and an increase in the ITA will decrease the vertical accuracy. (2) For the same-side stereo (SASS) mode, the smaller the ITA, the higher the planimetric accuracy (except for a weak ITA). In addition, if the incidence angle (ICA) is fixed, an increase in the ITA increases the vertical accuracy. (3) The base-to-height ratio is not the only prerequisite for the stereo configuration. The recommended ITA value for symmetric OPSS mode is 52–80°, and the suggested ITA threshold for SASS mode is 10–21°. The theoretical results are generally compatible with the previous achievements, which can guide applications such as ground control point production, DSM generation, and 3-D information extraction. |
Full Text from ScienceDirect