|
ИСТИНА |
Войти в систему Регистрация |
Интеллектуальная Система Тематического Исследования НАукометрических данных |
||
FOREST INVENTORY: REMOTE SENSING IMAGERY PROCESSING VERSUS GROUND-BASED OBSERVATIONSстатья
- Авторы: Kozoderov V.V., Dmitriev E.V., Kuleshov A.A.
- Журнал: The Complex Systems
- Том: 29
- Номер: 4
- Год издания: 2018
- Первая страница: 13
- Последняя страница: 29
- Аннотация: Enhancing our supervised classification procedures of remote sensing imagery processing, we compare the relevant results of airborne hyperspectral and satellite multispectral images given by WorldView-2 for the same test area. The string and rod problems are considered to solve the regularization procedure of the ill-posed inverse problem. The technique of error correcting output codes (ECOC) is used to test different designs of the related classifier by employing the listed airborne and spaceborne data. As a result, confusion matrices are shown of the main objects classification using these hyperspectral and multispectral remote sensing systems. Comparisons are made with typical ground-based forest inventory map of the test area. We can conclude similar errors of the forest canopy recognition while separating the pixels in accordance with their three categories of the Sun illumination (completely shaded; half shaded; sunlit tops) for each class of the forest objects on the images under processing. This is a characteristic feature of processing images of high spectral and spatial resolution.
- Добавил в систему: Козодеров Владимир Васильевич