1 / 18

imaginlabs

COSI- Corr Automatic Imperviousness Classification Study Cases. Sebastien Leprince Francois Ayoub Jiao Lin Jean-Philippe Avouac leprincs @ caltech.edu Office: 626-395-2912 Cell: 626-240-9041 California Institute of Technology. Patent U.S. 8,121,433 B2

ilana
Download Presentation

imaginlabs

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. COSI-Corr Automatic Imperviousness Classification Study Cases SebastienLeprince Francois Ayoub Jiao Lin Jean-Philippe Avouac leprincs@caltech.edu Office: 626-395-2912 Cell: 626-240-9041 California Institute of Technology Patent U.S. 8,121,433 B2 California Institute of Technology imaginlabs.com

  2. Case Study: Automatic classification of impervious surfaces Data: GeoEye image, 4-band multispectral, 2m GSD, above Indianapolis, with impervious surface classification benchmark (courtesy of MWH). Worldview 8-band multispectral images, 2m GSD (courtesy of DigitalGlobe): - Image of San Clemente, CA - Image of Sydney, Australia Goal: Testing automatic methods to extract the percentage of impervious surfaces using satellite images. Applications: Better management of storm water run-offs, tax identification.

  3. Indianapolis Test Image - GeoEye GeoEye Image

  4. Indianapolis Test Image Imperviousness Benchmark Provided Warmer color represents more % impervious

  5. Indianapolis Test Image COSI-Corr automatic imperviousness analysis Black is 0% impervious, White is 100% impervious Some inconsistencies exist but land boundaries are well defined. In particular, bare soils are harder to classify. More robustness can be achieved using Worldview-2 8-band multispectral images

  6. San Clemente CA, Test Image #1 – Worldview 2

  7. San Clemente CA, Test Image #1 – Worldview 2 COSI-Corr Imperviousness result

  8. San Clemente CA, Test Image #2 – Worldview 2

  9. San Clemente CA, Test Image #2 – Worldview 2 COSI-Corr Imperviousness result

  10. Sydney, Test Image #1 – Worldview 2

  11. Sydney, Test Image #1 – Worldview 2 COSI-Corr Imperviousness result

  12. Sydney, Test Image #2 – Worldview 2

  13. Sydney, Test Image #2 – Worldview 2 COSI-Corr Imperviousness result

  14. Sydney, Test Image #3 – Worldview 2

  15. Sydney, Test Image #3 – Worldview 2 COSI-Corr Imperviousness result

  16. Sydney, Test Image #4 – Worldview 2

  17. Sydney, Test Image #4 – Worldview 2 COSI-Corr Imperviousness result

  18. Conclusions • COSI-Corrcan provide automatic classification of impervious surfaces. It was found that classification accuracy is improved when using Worldview-2 8-band multispectral images instead of GeoEye 4-band images. • The most difficult parts to map are bare soils. Combining images at different seasons should alleviate most problems. More discussion is needed to decided how water bodies should be classified – should we differentiate between swimming-pools and natural water bodies? • We could introduce a “no data” class when the classification is not accurate, in particular in shadow areas. COSI-Corr can implement an automatic shadow detection if using Worldview-2 images. • More characteristics could be added if coupled with high resolution terrain model, which can also be extracted using COSI-Corr and Worldview stereo imagery (more competitive than LiDAR). • The results of this study are preliminary and can be improved. Please contact the authors for more information.

More Related