MULTI-SATELLITE IMAGE ALIGNMENT OVER LARGE AREAS WITH FEATURELESS REGIONS
There is a great deal of interest in fusing together the information provided by different satellites for real-time change detection on the surface of the earth. For detecting important types of changes on the ground, it is necessary to inject geometry into the data provided by low-res satellites us...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Copernicus Publications
2023-09-01
|
| Series: | The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
| Online Access: | https://isprs-archives.copernicus.org/articles/XLVIII-M-3-2023/211/2023/isprs-archives-XLVIII-M-3-2023-211-2023.pdf |
| _version_ | 1827824370522783744 |
|---|---|
| author | C. J. Roros R. Deshmukh A. C. Kak |
| author_facet | C. J. Roros R. Deshmukh A. C. Kak |
| author_sort | C. J. Roros |
| collection | DOAJ |
| description | There is a great deal of interest in fusing together the information provided by different satellites for real-time change detection on the surface of the earth. For detecting important types of changes on the ground, it is necessary to inject geometry into the data provided by low-res satellites using multi-view imaging satellites that record each point on the ground from multiple perspectives. Combining these perspectives and generating a DSM (Digital Surface Model) gives us the geometry needed for a more meaningful analysis of satellite data. Before such an analysis can be carried out, it is necessary to align the images from all available satellites. Automatic image alignment, however, requires features on the ground that can be identified and correctly matched across different images using computer vision algorithms. While such features are common in urban areas, that is not always the case in predominantly rural areas that present a more-or-less uniform texture to the sensors. In this paper we present methods for automatic identification and alignment of featureless regions. Featureless regions are identified using point spread maps, which are a byproduct of DSM generation. The subsequent strategy for aligning featureless regions depends on the proportion of featureless regions to feature-rich regions. If most of the AOI (Area of Interest) is feature-rich, we ignore featureless regions when estimating inter-satellite image alignment parameters and apply those parameters to the entire AOI. Finally, we present a technique to propagate and fuse parameters from feature-rich regions to featureless regions. |
| first_indexed | 2024-03-12T02:24:33Z |
| format | Article |
| id | doaj.art-e5c10291efcd4c52a1170b200a5f4dc2 |
| institution | Directory Open Access Journal |
| issn | 1682-1750 2194-9034 |
| language | English |
| last_indexed | 2024-03-12T02:24:33Z |
| publishDate | 2023-09-01 |
| publisher | Copernicus Publications |
| record_format | Article |
| series | The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
| spelling | doaj.art-e5c10291efcd4c52a1170b200a5f4dc22023-09-05T22:26:10ZengCopernicus PublicationsThe International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences1682-17502194-90342023-09-01XLVIII-M-3-202321121810.5194/isprs-archives-XLVIII-M-3-2023-211-2023MULTI-SATELLITE IMAGE ALIGNMENT OVER LARGE AREAS WITH FEATURELESS REGIONSC. J. Roros0R. Deshmukh1A. C. Kak2Robot Vision Lab, Purdue University, School of Electrical and Computer Engineering, West Lafayette, IN, USARobot Vision Lab, Purdue University, School of Electrical and Computer Engineering, West Lafayette, IN, USARobot Vision Lab, Purdue University, School of Electrical and Computer Engineering, West Lafayette, IN, USAThere is a great deal of interest in fusing together the information provided by different satellites for real-time change detection on the surface of the earth. For detecting important types of changes on the ground, it is necessary to inject geometry into the data provided by low-res satellites using multi-view imaging satellites that record each point on the ground from multiple perspectives. Combining these perspectives and generating a DSM (Digital Surface Model) gives us the geometry needed for a more meaningful analysis of satellite data. Before such an analysis can be carried out, it is necessary to align the images from all available satellites. Automatic image alignment, however, requires features on the ground that can be identified and correctly matched across different images using computer vision algorithms. While such features are common in urban areas, that is not always the case in predominantly rural areas that present a more-or-less uniform texture to the sensors. In this paper we present methods for automatic identification and alignment of featureless regions. Featureless regions are identified using point spread maps, which are a byproduct of DSM generation. The subsequent strategy for aligning featureless regions depends on the proportion of featureless regions to feature-rich regions. If most of the AOI (Area of Interest) is feature-rich, we ignore featureless regions when estimating inter-satellite image alignment parameters and apply those parameters to the entire AOI. Finally, we present a technique to propagate and fuse parameters from feature-rich regions to featureless regions.https://isprs-archives.copernicus.org/articles/XLVIII-M-3-2023/211/2023/isprs-archives-XLVIII-M-3-2023-211-2023.pdf |
| spellingShingle | C. J. Roros R. Deshmukh A. C. Kak MULTI-SATELLITE IMAGE ALIGNMENT OVER LARGE AREAS WITH FEATURELESS REGIONS The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences |
| title | MULTI-SATELLITE IMAGE ALIGNMENT OVER LARGE AREAS WITH FEATURELESS REGIONS |
| title_full | MULTI-SATELLITE IMAGE ALIGNMENT OVER LARGE AREAS WITH FEATURELESS REGIONS |
| title_fullStr | MULTI-SATELLITE IMAGE ALIGNMENT OVER LARGE AREAS WITH FEATURELESS REGIONS |
| title_full_unstemmed | MULTI-SATELLITE IMAGE ALIGNMENT OVER LARGE AREAS WITH FEATURELESS REGIONS |
| title_short | MULTI-SATELLITE IMAGE ALIGNMENT OVER LARGE AREAS WITH FEATURELESS REGIONS |
| title_sort | multi satellite image alignment over large areas with featureless regions |
| url | https://isprs-archives.copernicus.org/articles/XLVIII-M-3-2023/211/2023/isprs-archives-XLVIII-M-3-2023-211-2023.pdf |
| work_keys_str_mv | AT cjroros multisatelliteimagealignmentoverlargeareaswithfeaturelessregions AT rdeshmukh multisatelliteimagealignmentoverlargeareaswithfeaturelessregions AT ackak multisatelliteimagealignmentoverlargeareaswithfeaturelessregions |