Classification Research Of Digital Remote Sensing Image In Interpretation

Remote sensing technology plays an important role in resource assessment including mineral resources, water resources and environmental resources. The traditional classification on the remote sensing image is based on the statistical relationship between the statistical characteristics of remote sensing image and the training sample data, such as maximum likelihood classification and minimum distance classification. Which result in the effect is not good. At the same time, the other reason is the complexity of distributed object type. On the basis of clustering algorithm, a new method of classification of digital remote sensing image based on clustering algorithm is proposed. At first, in order to improve the classification effect of different shape clusters, this paper proposes Clustering Algorithm Based on Density and Density-reachable (CADD).The algorithm effectively improves the efficiency of clustering.At the same time, in order to achieve classification of large remote sensing image, this paper analyzes Clustering Algorithm Based on Density and Density-reachable (CADD) at first, and then makes some improvements, furthermore proposes Incremental Clustering Algorithm Based on Subcluster Feature (ICSCF). Theoretic analysis and experimental results demonstrate that ICSCF algorithm has higher clustering efficiency, because of using batch mode. At the same time, it can handle large databases through partition and has good scalability. It plays an important role in spatial clustering, such as remote sensing image processing.