Abstract:

In this paper, the spectrum characteristics of different objects in Southern Shanxi was obtained through the combination of PPI (pixel purity index) extraction and practical verification that based on FY-3A MERSI 250 m data from visible to near-infrared. The identification of different objects was realized by the methods of supervised classification such as maximum likelihood and neural network, meanwhile, the winter wheat area was extracted by the mixed-pixel decomposition technique. The results indicated that the extraction of winter wheat area was feasible based on mono temporal of FY-3A MERSI data. The accuracy of winter wheat area extracted by neural network method was better than that of maximum likelihood, while the result of mixed-pixel was the closest to actual planting area.