AUTOMATED SENSOR ORIENTATION ESTABLISHMENT FOR HIGH RESOLUTION SATELLITE IMAGES BY IMPROVED GCP CHIP MATCHING

초록

Correcting geometric distortions in satellite images is crucial for their effective utilization. Geometric distortion correction is performed by establishing a precise sensor orientation. Automated precision sensor orientation is essential for the rapid delivery of satellite images. For this, one of the key processes is GCP (Ground Control Point) chip matching. Our previous research developed a method of GCP chip matching at sub-pixel scale of satellite images and achieved a sub-pixel accuracy for mid-resolution images. In this paper, we enhance our previous method further with high-resolution satellite images and to achieve precise geometric accuracy automatically. Our method begins by establishingan initial sensormodel usingrationalpolynomialcoefficients(RPC) providedwith the original satellite image. The initial model is then used to search for all GCP chips within the boundaries of the satellite image. Next, we define areas surrounding each GCP chip and resampled GCP chips and satellite images according to desired match scale. Then, we apply GCPchip matchingfromcoarse scale to finer scaleup to desiredsub-pixellevel.Finally,we establish a precision sensor orientation based on the matchingresults and verify the accuracy of the sensor model. We tested this method using RapidEye images with a resolution of 5m and Kompsat3A with a resolution of 0.5m. We used GCP chips with 0.25m ground sampling distances. We applied GCP chip matching, up to a desired sub-pixel level. As a result of the study, it was possible to reduce the position error of more than 0.5 pixel in mid-resolution satellite images, and about 0.5 pixel in high-resolution satellite images. This study showed that performing GCP chip matching up to enhanced resolution by upsampling both chips and satellite images could establish a better performance than performing GCP chip matching up to the original resolution. The results of this study are expected to contribute to the improv ement of precision sensor orientation performance of satellite images with various GSDs. © 2023 ACRS. All Rights Reserved.

키워드