Image Acquisition and Overlap
Depending on the type of object/scene and the type of camera used, the following acquisition patterns are used:
Type of Scene
Circular acquisition around the object
Regular grid acquisition
Panoramic acquisition along corridor
3D information is extracted from overlapping images based on the stereoscopic effect. For each stereo pair, the corresponding pixels between the images are estimated using Dense Image Matching. Within SURE, generation of the 3D point cloud is based on multiple stereo models (Multi-Stereo Triangulation). Thus, the final surface quality is strongly influenced by the number of overlapping images and their similarity.
We strongly recommend that enough image overlap is established to ensure every point on the surface of interest is observed by 5 images (minimum 3 images). Furthermore, images should be taken at a similar image scale and thus images should be captured at similar distance to the object.
For details on image overlap of Aerial images, please see the Nadir Image Overlap page (also useful for Oblique imagery).
Image Capture Configuration
Irregular and unknown image distributions
Regular airborne grid acquisitions with “Maltese cross” configuration systems
Regular airborne grid acquisitions in Nadir configuration
For the latter two scenarios, the regular structure is exploited in order to identify a minimum set of stereo models.