Study on Rule Set Construction and Application to Land Use Classification

The high spatial resolution images are used and object-oriented classification method is adopted to determine the appropriate scale of land use types by various methods and the hierarchical network system of image object with multi-scale are obtained.The features and spatial relationship of image objects are derived,the auxiliary characteristic and the expert knowledge summarized effectively and build the classification rules set of image objects.The results of the classification indicate that the accuracy of classification can be improved effectively and get better semantic distinguish and more accurate results of the classification by using features and spatial relationship rules.This program can be used to classification of high resolution image with similar natural environment conditions and with a little modification to the proposed program.