Research On Updating Method Of Land Cover Data Based On CHANGE-ONLY Information

Global land cover change has become a hot issue. Land cover change information is important for exploring the change driving factors, explaining change rules, and evaluating the regional ecological environment. Furthermore it is the foundation of land cover database updating. Remote sensing imagery has become the popular data source. The author have discussed the updating methods based on remote sensing imagery in this paper.The remote sensing imagery of two periods in 2000 and 2009 of Shanxi province is used as experiment data in this research. Using some training samples, the land classification data is updated based on change-only information. The main research work is as follows:Firstly, the author preprocessed two images. Different change extraction methods are compared comprehensively. Some combined methods are adopted for the gain of change-only information and part of the pseudo change information is removed and its accuracy has been enhanced at some extent.Secondly, the existing data and auxiliary means are combined to extract the training and test samples. Different classification methods are analyzed. Specifically, the author chooses maximum likelihood method of supervised classification to classify the change-only information.Finally, the non-change area is extracted according to the classification data in 2009, and the change part of the area is replaced by change-only information obtained by former steps. However, the result is not suitable for visibility. Thus the head file is modified in a more intuitive way and then the whole renewal process is completed. Experiment results show that the method can reduce labor consumptions, improve work efficiency and feasible.