개별 및 번들 순차 고해상도 위성 데이터 RPCs 보정

High-resolution satellite data often provide RPCs (Rational Polynomial Coefficients) that inherently contain errors, making error correction using GCPs (Ground Control Points) essential. Typically, position accuracy is improved through individual RPCs correction for each image; however, such an approach may not ensure high consistency for overlapping or time-series imagery. Therefore, this study proposes and evaluates a sequential method combining individual RPCs correction and bundle adjustment-based RPCs error compensation to enhance positional accuracy and mutual consistency among overlapping images. For effective bundle adjustment, it is crucial to generate reliable tie points. To achieve this, individual RPCs are first corrected by matching each image with GCPs. Based on the corrected RPCs of individual images, tie points between target images for bundle adjustment are generated through image projection. Finally, bundle adjustment is performed using both the GCPs matching results and tie points, and the improvement in consistency is evaluated. Experimental results show that the proposed method improved the alignment accuracy of data, which initially exhibited inconsistencies greater than 1.4 pixels, to within 1 pixel.