Fusion of difference images for change detection in urban areas

Land cover in urban areas in China is changing rapidly during the past years as a result of urbanization. Changes detected from multi-temporal remote sensing images may help significantly in understanding urban development and supporting urban planning. Indeed, differences in reflectance spectra, easily obtained by satellite sensors, are important indicators for characterizing these changes. Although many algorithms were proposed to generate difference images, the results are usually greatly inconsistent. In this work, a complete procedure for land cover change detection by fusing change information obtained from multiple difference images is designed and implemented. Measurement and decision level fusion techniques are used to combine multiple difference images, and support vector machine (SVM) is selected to detect the changes. Multi-temporal CBERS images acquired in 2002 and 2008 are used to detect land cover changes and urban expansion in Shanghai, and experimental results confirm the effectiveness of the proposed approach. Using more change information, both the omission error and commission error could be reduced.