Road density provides meaningful structural information for urban development analysis. This paper investigates the integration of road density structural measure with Landsat TM spectral information for urban area change detection. Beijing, the Chinese capital, serves as the study area, where there have been great changes in the last two decades. Two Landsat TM images used in this study were acquired in the same season from 1984 and 1997. To reduce the spectral confusion between urban 'built-up' and rural 'non built-up' land cover categories, we propose a new structural method using road density combined with spectral bands for post-classification comparison change detection. Road density maps for both dates were produced using a Gradient Profile Direction Analysis (GDPA) algorithm and then integrated with spectral bands. The results from the combined spectral-structural datasets were evaluated and compared with the results from datasets of spectral bands alone. Our study shows that the addition of road density information greatly reduced spectral confusion and increased the accuracy of land cover classification, which in turn improved the change detection results.
展开▼