Automated quality monitoring of high resolution satellite images using natural target-based analysis techniques

초록

Periodic quality evaluation of images acquired from the satellite is very important work to guarantee the quality of image for end-users and to inspect the performance of the satellite. As one parameter of the quality, the NIIRS is widely used to describe interpretability of high resolution satellite imagery. Commonly, this parameter is measured using images of specially made tarps. Due to this, measuring NIIRS is costly and not carried out often. This problem may be eased if natural targets, such as edge profiles from object boundaries, building shadows, etc., can be used for NIIRS estimation. Our previous research attempted to use natural targets for accurate and robust NIIRS estimation. The purpose of this study is to automate the quality monitoring of satellite imagery by applying the natural target-based method. An algorithm assessment was performed through direct comparison with manual method using KOMPSAT-1 and -2 images acquired on difference dates. The results showed that the NIIRS measured by automated method was almost identical with the manual one. This indicates that the proposed method is possible to monitor the image quality effectively.