Change Detection Research For Remote Sensing Image Land Use Based On The Change Vector Analysis Method

Since the last century, due to the growing world population, the available landresources gradually decrease, so, some researchers pay more and more attention to landuse problems both at home and abroad. Nowadays, with the rapid development of socialand economic, land development and utilization activities are also increasingly active,land development intensity is gradually increased, terrain category change becomesincreasingly frequent. Land use change detection refers to the analysis of landutilization through the same area in different times, the traditional detection methods aremainly through artificial statistics outdoor, Remote sensing technology analyses theland’s information changes in the region through the test method in over time, and usesthe same area, but not the same time remote sensing images. The detection methods arealways the focus of research.This paper is mainly based on the change vector analysis method, and uses the TM,QuickBird, RapidEye and so many kinds of remote sensing data to study the land usechange detection.The main work of this paper includes prophase survey, collectingremote sensing images and computer processing, extracting change area and unchangedregion, statistic the change type information, etc.Revolve around the center of change vector analysis method, the paper has obtainedthe following results:1. The thesis puts forward a method to determine the optimal threshold based on theOTSU method, this method not only realizes the automation of the optimum thresholdselection, and avoids the choice of error with human-computer interaction.2. And the thesis also improves the OTSU method, histogram curvature is used toobtain upper and lower bounds of the range as the traverse range for the OTSU method.Thereby the computational complexity of the computer is decreased, and the efficiencyis greatly improved.3. At the same time, IDL is used as the paper core computing programminglanguages. Through the program, land use change area will be automatically extracted,and there is zero artificial participation in the extraction process, it avoids artificial errorcaused by objective factors. The experiments show that the algorithm is feasible andpractical, it is very good. 4. In this paper, different sensors and different resolution images of the land are alsoexplored in the land use change detection, it is proved that the change vector analysismethod is applicable for multi-source data, but for the sake of change detectionaccuracy, it is best to use the same data.5. Postclassification Comparison Method is also used for contrast experiment, theexperimental results prove that the change vector analysis method is superior to theformer. The land use change detection based on the change vector analysis is worth ofapplication and popularization.