Multi-Temporal Remote Sensing Image-Based ...

The crop planting information extraction is crucial to the estimation of crop output, the key of which is to speedily and accurately extract planting information through the remote sensing image. The multi-temporal remote sensing data, together with the supervised classification and decision tree classification method, are used in this study to speedily and accurately extract crop planting information from TM/ETM+ remote sensing images and sixteen MODIS time series remote sensing images, to interpret major crops in the Heilonggang area. Overall, classification accuracy is up to 91.3%, compared with one simple supervised classification of TM images. The relative errors of cotton, maize, wheat and vegetables are reduced by 1.3%, 20.5%, 2.0% and 13.8%, respectively. It has been proved that this method has high accuracy and it can serve as an effective index for reflecting the crop planting distribution. The data can provide important scientific basis for the adjustment of the major crop planting structure in the Heilonggang area, and could also provide reference for information extraction in other areas.