Mapping of Agricultural Area Using Multitemporal Remote Sensing Images.

Abstract

A new method is proposed for agricultural area mapping using multitemporal satellite Landsat TM images. This statistical method is designed for the discrimination of agricultural area and non-agricultural area and crop identification, based on the fusion of multitemporal multispectral data. In the proposed algorithm, an important role is played by the transition probabilities, which take into account the temporal dependence of images. The transition probabilities are estimated directly from the pattern of Landsat TM NDVI. The feasibility of the new method is verified by experimental results.