| 1. |
|
|
| 2. |
|
|
| 3. |
|
|
| 4. |
|
|
| 5. |
- Ban, Yifang, et al.
(författare)
-
Fusion of RADARSAT fine-beam SAR and QuickBird data for land-cover mapping and change detection
- 2007
-
Ingår i: Geoinformatics 2007Proceedings of SPIE - The International Society for Optical Engineering. - : SPIE. - 9780819469120 ; , s. H7522-H7522
-
Konferensbidrag (refereegranskat)abstract
- The objective of this research is to evaluate multitemporal RADARSAT Fine-Beam C-HH SAR data, QuickBird MS data, and fusion of SAR and MS for urban land-cover mapping and change detection One scene of QuickBird imagery was acquired on July 18, 2002 and five-date RADARSAT fine-beam SAR images were acquired during May to August in 2002. Landsat TM imagery from 1988 was used for change detection. QucikBird images were classified using an object-based and rule-based approach. RADARSAR SAR texture images were classified using a hybrid approach. The results demonstrated that, for identifying 19 land-cover classes, object-based and rule-based classification of Quickbird data yielded an overall classification accuracy of 86.7% (kappa 0.857). For identifying I I land-cover classes, ANN classification of the combined Mean, Standard Deviation and Correlation texture images yielded an overall accuracy: 71.4%, (Kappa: 0.69). The hybrid classification of RADARSAT fine-beam SAR data improved the ANN classification accuracy to 83.56% (kappa: 0.803). Decision level fusion of RADARSAT SAR and QuickBird data improved the classification accuracy of several land cover classes. The post-classification change detection was able to identify the areas of significant change, for example, major new roads, new low-density and high-density, builtup areas and golf courses, even though the change detection results contained large amount of noise due to classification errors of individual images. QuickBrid classification result was able add detailed change information to the major changes identified.
|
|
| 6. |
- Ban, Yifang, et al.
(författare)
-
RADARSAT Fine-Beam SAR Data for Land-Cover Mapping and Change Detection in the Rural-Urban Fringe of the Greater Toronto Area
- 2007
-
Ingår i: Proceedings, Urban Remote Sensing Joint Event, 2007.
-
Konferensbidrag (övrigt vetenskapligt/konstnärligt)abstract
- This research investigates the capability of the multitemporal RADARSAT Fine-Beam C-HH SAR imagery for landuse/land-cover mapping and change detection in therural-urban fringe of the Greater Toronto Area (GTA). Five-date RADARSAT fine-beamSAR images were acquired during May to August in 2002. One scene of Landsat TM imagery was acquired in 1988 for change detection. The major landuse/land-coverclasses were high-density built-up areas, low-density built-up areas, roads, forests, parks, golf courses, water and three types of agricultural lands. These ten classes were chosen to characterize the complex landuse/land-cover types in the rural-urban fringe of the GTA. The results demonstrated that, for identifying landuse/land-cover classes, five-date raw SAR imagery yielded very poor result due to speckles. Much better results were achieved with combined Mean, Standard Deviation and Correlation texture images using artificial neural networks (ANN) and with raw images using object-based classification. The change detection procedure was able to identify the areas of significant changes, for example, major new roads, new low-density and high-density built up areas and golf courses, even though the overall accuracy of the change detection was rather low.
|
|
| 7. |
- Ban, Yifang, et al.
(författare)
-
RADARSAT SAR data for landuse/land-cover classification in the rural-urban fringe of the greater Toronto area
- 2005
-
Ingår i: Proceedings 2005.
-
Konferensbidrag (refereegranskat)abstract
- This research investigates the capability of the multitemporal RADARSAT Fine-Beam C-HH SAR imagery for extracting landuse/land-cover information in the rural-urban fringe of the Greater Toronto Area (GTA) using various image processing techniques and classification algorithms. Five-date RADARSAT fine-beam SAR images were acquired during May to August in 2002. The major landuse/land-cover classes were high-density built-up areas, low-density built-up areas, roads, forests, parks, golf courses, water and three types of agricultural lands. These ten classes were chosen to characterize the complex landuse/land-cover types in the rural-urban fringe of the GTA. The results demonstrated that, for identifying landuse/land-cover classes, five-date raw SAR imagery yielded very poor result due to speckles. The best result was achieved for combined Mean, Standard Deviation and Correlation texture images using artificial neural networks (ANN) (overall accuracy: 89.7% and Kappa: 0.886). These high accuracies indicated that RADARSAT fine-beam SAR has the potential for operational landuse/land-cover mapping in urban environments.
|
|
| 8. |
|
|
| 9. |
|
|
| 10. |
|
|
