Abstract: This paper discusses a specific method to be used to extract arable areas efficiently and effectively. As a key step, the classification method of remote sensing image processing is often regarded as one of the most important technological systems in crop estimation. Considering the features of natural environment, economic conditions as well as growing procedure of corns, based on the characteristics of the TM data, author puts forward an ecoland classification method (also called geoprocessing method) by which precision classification would be conducted in the project. The ecoland classifacation method is basically divided into three steps: the first one is called subarea in which similar spectrum is put into one group, that is, there is almost the same ecologic environment in the area, combined with GIS data such as social and economic statistics data, in this way, better recognized land patterns; the second is an extraction stage, based on the sublayers, different patterns can be easily extracted by means of geodata stored in the database; the last one is considered as a merge and analysis part, different classified subareas would finally be merged together in the basis of the same coordinate, subsequentlly, a synthetic analysis on the merged result should be made.As a result, coordinating with field surveying data in the process of automatic classification, the classified image with geodata is produced by means of the TM data source and can be easily extracted out related information for the further applications.