SATELLITE IMAGE DATASET TAXONOMY: A MULTIDIMENSIONAL FRAMEWORK FOR EFFECTIVE DATASET UTILIZATION

Abstract

Satellite imagery datasets are employed in research and industry. A classification system must apply the dataset according to its purpose and intention to increasing the utilization value of satellite imagery datasets. Therefore, we propose a method to organize and classify satellite imagery datasets, departing from existing research trends. Thus, this study developed a classification system using currently available satellite image datasets called a taxonomy, consisting of eight dimensions to describe satellite imagery (image quality, satellite type, label information, dataset acquisition channel, coverage area, time, applicable models, and available fields) and a table containing characteristics comprising each dimension. This paper details the iteration process of the development stage of the taxonomy and the theoretical background and evidence used to create the content for each dimension. Examples of how the taxonomy classifies actual datasets and how it can be applied in practice are included to provide practical guidelines for using satellite image datasets.