기구축 고해상도 위성 데이터를 이용한 정밀 시계열정사영상 자동 생성

A variety of high-resolution satellites including large, mid, and small size fleet satellites are expected to operate in near future. The high frequency data acquisition based on the multiple satellites is expected to enable the rapid change detection and map updates but this is impossible without the rapid data preprocessing including precise orthorectification. Traditional manual orthorectification is accurate but not efficient such that an automated approach using accurately created GCP (Ground Control Point) chips is favored. The accurate GCP chips based on aerial photogrammetry have problems that they are not easy to acquire and update. Therefore this study proposes an method of using preprocessed satellite data to georeference time-series images that are acquired afterward. In the approach, an orthorectified reference image is used to generate reference image chips that show better similarity to targeted time-series images. The chips are image-matched to time-series images for RPCs bias-compensation. Experimental results presented that the proposed approach showed better precision of RPCs bias-compensation than that of conventional GCP chip-based approach. In addition, an edge matching was efficient to overcome image intensity differences due to the seasonal difference such as snow cover.