Random Forest Winter Wheat Extraction Algorithm Based on Spatial Features of Neighborhood Samples

In order to effectively obtain the winter wheat growing area in a large part of Guanzhong plain, this paper proposes random forest plain extraction algorithm based on spatial features neighborhood samples using 250 m resolution spectral imager (MERSI) FY-3 satellite as data source. paper, first, training and validation were obtained by constructing sample space sampling model, then study was classified an integrated learning Classifier, finally classification from different time phases fused voting game theory final result map. The land use change distribution 2011 2014 also analyzed. experimental results showed that overall accuracy after fusion processing highest compared with traditional algorithm, reaching 98.63%. At same time, LANDSAT 8 images used wheat, areas MERSI generally consistent terms shown at county scale.