This dataset contains eight triples of historical images for four different sights. Images were chosen with respect to their possible matching quality. The images show combined differences in illumination, field of view, viewpoints, blurring and slight rotation. Some of the images show building reflections in water or extreme shadowing. The images are saved after digitization in full quality as *.tif files with a maximum sidelength of 3543 Pixels. Since no inner orientation could be determined for all image triples the Trifocal Tensor is provided - calculated using Ressl's method (Ressl, 2003). Additional metadata information, copyright disclaimer and permalinks are provided in License.txt. The purpose of the dataset is the evaluation of different feature detection and matching methods using the given orientation with the Trifocal Tensor. Point transfer calculation is possible using the equation on p. 382 in Multiple View Geometry in Computer Vision (Hartley and Zisserman, 2003). Another method uses the corrected Fundamental Matrices calculated in eq. 15.8 from the Trifocal Tensor on p. 374 in Multiple View Geometry in Computer Vision (Hartley and Zisserman, 2003). Ressl, C., 2003. Geometry, constraints and computation of the trifocal tensor. TU Wien. Hartley, R. and Zisserman, A., 2003. Multiple view geometry in computer vision. Cambridge university press. https://opara.zih.tu-dresden.de/xmlui/handle/123456789/1371 2026-03-16T19:48:55Z 2026-03-16T19:48:55Z Maiwald, Ferdinand https://opara.zih.tu-dresden.de/xmlui/handle/123456789/1383 2019-03-26T11:30:34Z 2019-01-01T00:00:00Z

Benchmark dataset using historical images for an automated evaluation of feature matching methods Maiwald, Ferdinand Image dataset to the submitted ISPRS GSW2019 publication "Generation of a benchmark dataset using historical photographs for an automated evaluation of different feature matching methods". This dataset contains eight triples of historical images for four different sights. Images were chosen with respect to their possible matching quality. The images show combined differences in illumination, field of view, viewpoints, blurring and slight rotation. Some of the images show building reflections in water or extreme shadowing. The images are saved after digitization in full quality as *.tif files with a maximum sidelength of 3543 Pixels. Since no inner orientation could be determined for all image triples the Trifocal Tensor is provided - calculated using Ressl's method (Ressl, 2003). Additional metadata information, copyright disclaimer and permalinks are provided in License.txt. The purpose of the dataset is the evaluation of different feature detection and matching methods using the given orientation with the Trifocal Tensor. Point transfer calculation is possible using the equation on p. 382 in Multiple View Geometry in Computer Vision (Hartley and Zisserman, 2003). Another method uses the corrected Fundamental Matrices calculated in eq. 15.8 from the Trifocal Tensor on p. 374 in Multiple View Geometry in Computer Vision (Hartley and Zisserman, 2003). Ressl, C., 2003. Geometry, constraints and computation of the trifocal tensor. TU Wien. Hartley, R. and Zisserman, A., 2003. Multiple view geometry in computer vision. Cambridge university press. The images were taken from around 1880-1992 and digitized for the purpose of this publication from 2018-2019.

2019-01-01T00:00:00Z