METHOD OF AUTOMATICALLY EXTRACTING WATER-LAND BOUNDARY IN FRESHWATER REGIONS BY USING HYPERSPECTRAL REMOTE SENSING IMAGE

摘要

The present invention relates to a method of automatically extracting a water-land boundary in freshwater regions by using a hyperspectral remote sensing image, including: (a) receiving hyperspectral image data from a hyperspectral camera by an input device, and performing an image preprocessing process (radiation correction, atmospheric correction, and geometric correction) by a preprocessing module; (b) preliminarily classifying, by a decision tree pre-classification module, a water region and a land region by using the hyperspectral image that has been subjected to the preprocessing process while misclassification is included; (c) re-classifying, by a decision tree reclassification module, the preliminarily-classified water region including the misclassification by subdividing the preliminarily-classified water region into at least one of water, an artificial structure, and a land shadow; (d) re-classifying, by the decision tree reclassification module, the preliminarily-classified land region including the misclassification by subdividing the preliminarily-classified land region into at least one of green algae, soil, and vegetation; and (e) finally combining, by the decision tree reclassification module, the water and the green algae to determine a combined result as the water region, and integrating remains into the land region. Accordingly, a decision tree for classifying water and land from the hyperspectral image is expanded and subdivided to compensate for the misclassification of the water and the land.;COPYRIGHT KIPO 2020