Investigation on crop planting structure based on synergy of high spatial-temporal resolution remote sensing data

ObjectiveThe present study explored information including space, time and spectrum in remote sensing image, explored automatic identification of multi-source remote sensing crop information supported by plot. The research techniques could provide reference for rapid and refined survey of crop planting structure. Method Fusui county was taken as research region. Firstly, boundary of fine farmland block information were extracted by multi-scale segmentation and object contour editing of high spatial resolution images. Then, time series spectrum features of the covering crop during growth period were obtained by blot. Finally, multidimensional feature space combining static spectrum and dynamic pro-cess feature was formed based on temporal spectrum and its changing definition as well as description parameters of crop growth conditions. Crop planting information extraction model was established based on crop phenology rhythm characteris-tics, which realized extraction of main crop planting structure information. ResultBased on this method, fine plots of sugarcane, rice and other crops in Fusui, as well as grassland, woodland, water and urban construction land were drawn. The extracted area of sugarcane in Fusui was 82420.01 ha and that of rice was 6806.67 ha. The overall classification accuracy was 86.8% and Kappa coefficient 0.84.ConclusionThe extracted crop planting structure in Fusui is in accor-dance with usage precision demand, which can serve as basis for agricultural subsidy granting and loss assessment of agri-cultural disaster. The technique used in the research can provide reference for rapid and refined survey of crop planting in-formation.