This paper describes the development of a shoreline extraction and change monitoring system aimed at providing coastal-environment information using high-resolution KOMPSAT series satellite images.
For the satellite-image-based shoreline automatic extraction, the machine-learning-based object extraction algorithm was developed, and to utilize the developed algorithm for services, the OpenCV-based monitoring system was developed. In addition, to verify the accuracy of the extracted shoreline information, the reliability of the developed algorithm was verified by comparing the proposed system with the existing diverse image object extraction methods and manually digitized results. First, to develop the high-resolution-image-based shoreline automatic extraction algorithm, the artificial-neural-network-(ANN)-based machine learning technique was used. For the application of this technique, training sample data extracted in advance from KOMPSAT images were created, and the clustering technology was applied to the data. The water and land were divided into binary categories to extract vector-format shorelines. Thus, data with more precise accuracy compared to the existing NDVI-based shoreline data extraction technique can be extracted, and the final vector-format data were calculated, making it possible to maximize their use as quantitative data. That is, the final output was calculated in terms of the type of standardized data in the geographical information category, thus securing the diverse uses of the analysis results. In addition, to develop a monitoring system for its effective utilization, instead of using the existing commercial software, an OpenCV-based system was implemented for extracting, comparing, and analyzing shoreline data. As a result, the system can be used in diverse platform environments, and in particular, the multiple-time image-based data comparison and analysis function makes it possible to conduct quantitative analysis and to monitor shoreline change trends. Thus, the system is believed to be usable as an effective tool for analyzing coastal- environment changes. Coastal-environment changes occur more slowly and are wider in scope compared to land environment and weather changes, making it difficult to define their occurrence time as well as to quantify the coastal damage, if any. The main purpose of analyzing the satellite-image-based global observation information is to monitor the change trends from the macro perspective. Given this purpose, the proposed shoreline data extraction algorithm and the monitoring system using such algorithm are deemed to be suitable as tools for analyzing the coastal-environment change data.