Assessment of Different Image Transformation Methods on Diwata-1 Smi Images Using Structural Similarity Measure

Abstract

Abstract. This paper aims to provide a qualitative assessment of different image transformation parameters as applied on images taken by the spaceborne multispectral imager (SMI) sensor installed in Diwata-1, the Philippines’ first Earth observation microsatellite, with the aim of determining the order of transformation that is sufficient for operationalization purposes. Images of the Palawan area were subjected to different image transformations by manual georeferencing using QGIS 3, and cloud masks generated and applied to remove the effects of clouds. The resulting images were then subjected to structural similarity (SSIM) tests using resampled and cloud masked Landsat 8 images of the same area to generate SSIM indices, which are then used as a quantitative means to assess the best performing transformation. The results of this study point to all transformed images having good SSIM ratings with their Landsat 8 counterparts, indicating that features shown in a Diwata-1 SMI image are structurally similar to the same features in a resampled Landsat 8 data. This implies that for Diwata-1 data processing operationalization purposes, higher order transformations, with the necessary effort to implement them, offer little advantage to lower order counterparts.