Search academic texts

Registration Number

0216U003050, 0115U006556 , R & D reports

Title

Development of crop classification methods on the base of optical and radar satellite data.

Head

Kravchenko Oleksiy,

Registration Date

12-01-2016

Organization

Space Research Institute

Description

Purpose - new methods development for crop classification based on optical and radar satellite images. Services of crop monitoring is based on the obtained classification maps. We proposed universal approach for improving crop classification technique and disseverance crop types. It is based on hierarchical deep learning, analysis of spectral channels of optical and radar satellite images and establishing hidden dependencies. For crop classification we proposed use not only optical but also radar data that have become free since 2015. Training and test data were received from ground measurements and from international Ukraine JECAM polygon. Within the project in 2015 we improved optical and radar crop classification method that based on neural networks by taking into account vector information about fields boundaries. In the result, adding radar data allows improve overall classification map accuracy by 7.3% compare to results based on only optical data. The proposed method of filtration allows to consider annual boundaries changes by dividing parcels into a few fields that allows to increase accuracy of the final product. We obtained crop classification map based on proposed approach and upload it on the geoportal.

Product Description

popup.authors

Басараб Руслан Михайлович
Колотій Андрій Всеволодович
Компанієнко Ліля
Кравченко Олексій Миколайович
Лавренюк Микола Сергійович
Олійник Тамара Віталіївна
Яйлимов Богдан Ялкапович
Ящук Дар'я Юріївна