The studying of expert system for soil remote sensing classification and recognition

An expert system for soil surveying is presented. In this system the processing is divided into two steps. First, the original data is classified by statistical unsupervised classification method. And then, the result of classification is labeled by expert knowledge and experience. In this system, non-spectral information as well as spectral information are employed to improve the accuracy of classification. The knowledge and experience of an expert are introduced and the processing of soil interpretation of expert is simulated to get the relationship of the result of classification to the kind of soil.<
>