Land-Use and Land-Cover Change Detection Based on Object-Oriented Theory

Traditional change detection methods base on pixel with a low accuracy and a poor automation. It could not meet the needs of actual production. To improve the accuracy, the object-oriented theory will be introduced to land-used and land-cover change detection. This paper will use decision tree classification method to reduce the classification error propagation for improving the accuracy of test result. First, to do multi-scale segmentation, group and analysis of features and establish a class hierarchy structure (Decision Tree), with spectral information of remote sensing data, texture features, topological and thematic information. And then do the vector analysis for the result of the classifications to get the change information. The test result convinces the effectiveness of the method with comparing different detection methods. And get guiding significance to update data on land use, agricultural land survey applications and so on.