Semi-Operational Identification of Agricultural Crops from Airborne SLAR Data

摘要

Preprocessing, segmentation, and pseudo-hierarchical classification of a multitemporal data set of three test sites for identification of potatoes and other agricultural crops is described. Good discrimination capabilities between crop types, especially potatoes, is achieved. Unique identification of the three major crop types (sugar beets, potatoes and winter wheat) with an accuracy greater than 90% is possible. In one area, this accuracy is obtained when using only the July data set. It appears possible to identify more than one species of winter wheat and potatoes. Oats and barley are difficult to distinguish from other crop types. It is possible to obtain the required results with a high and a low track in July and one other run in May.