2015 International Symposium of Computer Application and Information Technology (IsCait 2015)

324 / 1971-01-01 00:00:00

Extracting Buildings From Polarization Sar Image Using Object-oriented Classification Method

To explore the method of extracting buildings from polarization SAR image, this paper deeply analyzes the thought of H/alpha/A-Wishart classification which is based on pixels. Considering the spots phenomenon of the classification results, the object-oriented technique was introduced in the process of classification. We used object-oriented H/alpha/A-Wishart classification method to extract buildings from the matrixes which were generated by multi-scale segmentation method, and then compared the results with that of the unsupervised classification method based on pixels (H/alpha/A-Wishart classification). The experimental results show that the accuracy of extracting buildings reached 98% from 83.5%. It can be well distinguished the buildings with water, roads, etc.