Based on Google earth engine platform to extract cropland using OBIA methods

Using remote sensing technology to monitor and protect cultivated land is of great significance for ensuring food safety, maintaining ecological balance, promoting economic development, maintaining social stability and cultural inheritance. Large-scale cultivated land extraction using pixel-based methods will cause serious "salt-and-pepper" phenomenon. In this paper, the method using object-based image analysis (OBIA) was used to extract cultivated land. Firstly, image objects are obtained based on simple non-iterative clustering (SNIC) segmentation, and phenological features of cultivated land are extracted. Finally, random forest (RF) and support vector machine (SVM) classifiers are used to extract cultivated land. The results indicated that SVM classifier achieved the better results than RF classifier.