Impervious surface extraction from high-resolution satellite image using pixel- and object-based hybrid analysis |
| |
Authors: | Xueliang Zhang Xuezhi Feng |
| |
Affiliation: | 1. Department of Geographic Information Science , Nanjing University , Nanjing , 210093 , PR China;2. Jiangsu Provincial Key Laboratory of Geographic Information Science and Technology, Nanjing University , Nanjing , 210093 , PR China |
| |
Abstract: | High-resolution satellite imaging provides a wealth of details about the Earth's surface, but it is still a challenge to determine the complex, impervious surface from high-resolution satellite images. A pixel- and object-based hybrid analysis (POHA) method is proposed for the extraction task. Pixel-based analysis is first applied to provide prior knowledge; then, based on prior knowledge, the subsequent object-based analysis is simply to find similar rather than new impervious objects using a weighted minimum distance strategy. In order to combine different image analysis methods, the segmentation masking strategy was introduced to transform the image analysis from pixel level to object level. A QuickBird image of Hangzhou City in China was used to test POHA. Furthermore, POHA was compared with both the pixel-based analysis and object-based image analysis (OBIA) methods, showing that POHA runs with limited human–computer interactions, and can provide accurate impervious surface mapping. |
| |
Keywords: | |
|
|