Search academic texts

Information × Registration Number 0217U004218, 0116U001557 , R & D reports Title Methodology for the classification of agricultural crops using satellite data - the sekond year of the project popup.stage_title Head Kravchenko Oleksiy, Registration Date 24-01-2017 Organization Space Research Institute popup.description2 Study object - information technology for data acquisition for remote and ground measurements for crop classification. Purpose - to develop crop classification and area estimation methods based on satellite images. Research method - intelligent computing, support vector machine, statistical methods, decision trees. We analyzed methods of machine learning and remote sensing data that can be used for land cover classification. We developed crop classification methods based on time series of satellite data. Thus, to achieve the goal we analyzed the classification algorithms on the basis of incomplete data and ensured their reliable operation in the presence of information gaps. In this work we proposed approaches to solve the problem of non-representative training set. They are based on spectral channels of satellite imagery analysis and ground measurements during sampling. For crop classification we proposed use optical data. Ground measurements were obtained by ground surveys and Ukrainian JECAM test site (http://jecam.org.ua) in 2013. Within the first year of the project, we provided digitization the ground observation data in the JECAM test site. Using the developed methods and algorithms we provided crop classification maps for the Kyiv region. The methods and algorithms satisfy the principles of deep learning: for solving applied problem it is necessary to use geospatial data and large volume of satellite data. Product Description popup.authors Колотій А.В. Кравченко О.М. Лавренюк М.С. Яйлимов Б.Я. popup.nrat_date 2020-04-02 Close