| Title | Integration of RADARSAT-2 multi-polarization data for stereo image matching |
| |
| Author | Zakharov, I; Toutin, T |
| Source | Proceedings of the 30th Canadian symposium on remote sensing 'Bridging Excellence'; 2009, 9 pages |
| Year | 2009 |
| Alt Series | Earth Sciences Sector, Contribution Series 20090178 |
| Publisher | Canadian Remote Sensing Society |
| Meeting | 30th Canadian Symposium on Remote Sensing; Lethbridge, Alberta; CA; June 22-25, 2009 |
| Document | book |
| Lang. | English |
| Media | paper |
| Subjects | geophysics; remote sensing; radar imagery; satellite imagery; radar methods; RADARSAT-2 |
| Illustrations | satellite images; plots |
| Program | Topographic Mapping |
| Released | 2009 01 01 |
| Abstract | The problem of processing RADARSAT-2 multi-polarization data was considered for stereo-radargrammetric applications. A generation of digital terrain model requires a stereo pair of intensity images
acquired by satellite from different incident angles. Several techniques (span, restoration and application polarimetric filters) for noncoherent integration of an intensity (or magnitude without taking into account phase information) image from
multi-polarization data were investigated for improving stereo image matching. The method of image restoration makes it possible to recover a single image from several images degradated during the image formation process. The image integration
techniques were compared and evaluated with different quality measures such as Equivalent Number of Looks (ENL), mutual information, correlation coefficient and textural measures calculated from the gray level co-occurrence matrix. The comparison
shows that an image processed by the restoration method performs better than the other image integration techniques. A combination of some second order statistic measures with ENL makes it possible to estimate stereo image matching accuracy. In
addition the image processing results were analyzed at the signal level. An estimation of influence on the point target and the edge features makes it possible to conclude that there is a good preservation of small details and edge with the use of
the image restoration technique. |
| GEOSCAN ID | 247826 |
|
|