Object-Oriented Crop Classification Using Time Series Sentinel Images from Google Earth Engine

The resulting maps of land use classification obtained by pixel-based methods often have salt-and-pepper noise, which usually shows a certain degree cluttered distribution image elements within the region. This paper carries out study on crop and identification based time series Sentinel images object-oriented takes recognition National Modern Agricultural Industrial Park in Jalaid Banner, Inner Mongolia, as research object. It uses Google Earth Engine (GEE) cloud platform to extract satellite radar optical remote sensing combined with simple noniterative clustering (SNIC) multiscale segmentation random forest (RF) support vector machine (SVM) algorithms classify identify major regional crops spectral features. Compared method, combination SNIC can effectively reduce phenomenon improve accuracy highest 98.66 kappa coefficient 0.9823. provides reference for large-scale work.