The Research On The Detection Method Of Built-up Land And Its Change Information By Remote Sensing Data

The urbanization nearly reflects the development and progress of an area, marking culture and civilization for one era. With the change of the urban system, the urban problem is becoming more complicated and is urgent to be solved. Studying the change of space expanding and ecological environment of the city is an import way to find and resolve the urban problem. Built-up land expansion is uppermost symbol of the urban space expansion.The paper analyses imagery feature of built-up land based on 3S technologies and Nanjing city imagery in 2002 and then classifies the built-up land information by using data-mining technology to classification of land using. As the same time data-mining technology was used to detection of built-up land change information. And superposition analysis was applied in result of change information detected to check result and refine the classification. And the adaptability of the detection method was proved .The core of the study and conclusions are as follows:1. Research on the automatic extraction of building information. At first, the characteristics of built-up land and their feature in year imagery were analyzed deeply. And the built-up land of feature sets was made. At the same time, the region, in which objects were dispersed and complex, was segmented into three sub-regions in which objects had representation by using region-dependent segmentation technology. Then the data-mining technology was used to extract optimum feature subset of built-up land and corresponding knowledge rules. After that, the decision tree algorithm was used to classify the study area and construction land using information was extracted. Accuracy of classification of decision tree was 84.7% and the first classification accuracy is 91.6%. It was showed that it could work well in the automatic extraction of built-up land and had the value of using. The technique was easy to be employed and the generated knowledge rules are objective.2. Research on the change detection of built-up land using. The characteristics of change of built-up land and their feature in year imagery were analyzed deeply. And the construction of feature sets was made. The data-mining technology in machine study was used to detect the change of built-up land, and the result of stair original classification was gained. Then the result of classification was fold analyzed by the built-up land extraction of former age (for example three types construction land using of year 2002 ), and was proofed by which the stair and second classification were obtained. Finally the same change detection technology flow was used to get change information between 1988 and 1994. The result of fold analysis and accuracy of the result show that data-mining technology used to detect the change was feasible. At the same time, on the condition of lack ground data, using the result of former age classification could proof the change detection result could examine the detection result as well as refine the result.