The Discussion of Knowledge-Based Classification of Remote Sensing Data Using Spatial Data Mining Approach

摘要

Incorporating of ancillary data into image classification can increase classification accuracy. Knowledge-based expert classification system is a useful approach of incorporating ancillary data, but difficult in obtaining knowledge used in classification. This paper has developed a method for creating the knowledge based on the concepts and technologies of data mining, particularly for data focusing and classification rule decision in the remote sensing classification process. This method integrates the character of spectrum, texture and space into geosciences information which combines the characters of images with the information from other ancillary datum. A practical demonstration has been performed on the land cover classification in Jiu-Zhai-Gou county, Sichuan province. The result shows the approach is effective, and it can improve the level of automation in classification.