@article{fdi:010088779,
  title = {{S}olving photogrammetric cold cases using {AI}-based image matching : new potential for monitoring the past with historical aerial images},
  author = {{M}aiwald, {F}. and {F}eurer, {D}enis and {E}ltner, {A}.},
  editor = {},
  language = {{ENG}},
  abstract = {{W}ith the ongoing digitization in archives, an increasing number of historical data becomes available for research. {T}his includes historical aerial images which provide detailed information about the depicted area. {A}mong the applications enabled by these images are change detection of land use, land cover, glaciers, and coastal environments as well as the observation of land degradation, and natural hazards. {S}tudying the depicted areas and occurring 3{D} deformations requires the generation of a digital surface model ({DSM}) which is usually obtained via photogrammetric {S}tructure-from-{M}otion ({S}f{M}). {H}owever, conventional {S}f{M} workflows often fail in registering historical aerial images due to their radiometric characteristics introduced by digitization, original image quality, or vast temporal changes between epochs. {W}e demonstrate that the feature matching step in the {S}tructure from {M}otion ({S}f{M}) pipeline is particularly crucial. {T}o address this issue, we apply the two synergetic neural network methods {S}uper{G}lue and {DISK}, improving feature matching for historical aerial images. {T}his requires several modifications to enable rotational invariance and leveraging the high resolution of aerial images. {I}n contrast to other studies our workflow does not require any prior information such as {DSM}s, flight height, focal lengths, or scan resolution which are often no more extent in archives. {I}t is shown that our methods using adapted parameter settings are even able to deal with quasi texture-less images. {T}his enables the simultaneous processing of various kind of mono-temporal and multi-temporal data handled in a single workflow from data preparation over feature matching through to camera parameter estimation and the generation of a sparse point cloud. {I}t outperforms conventional strategies in the number of correct feature matches, number of registered images and calculated 3{D} points and allows the generation of multi-temporal {DSM}s with high quality.{W}ith the flexibility of the method, it enables the automatic processing of formerly unusable or only to be interactively processed data, e.g. aerial images where the flight route is unknown, or with difficult radiometric properties. {T}his makes it possible to go back even further in time, where the data quality usually decreases, and enables a holistic monitoring and comparison of environments of high interest. {T}he code is made publicly available at https://github.com/tudipffmgt/{HAI}-{SFM}.},
  keywords = {{H}istorical aerial images ; {F}eature matching ; {N}eural networks ; {S}tructure-from-motion ; {D}igital surface model ; {M}ulti-temporal ; {CONGO} ; {FRANCE}},
  booktitle = {},
  journal = {{ISPRS} {J}ournal of {P}hotogrammetry and {R}emote {S}ensing},
  volume = {206},
  numero = {},
  pages = {184--200},
  ISSN = {0924-2716},
  year = {2023},
  DOI = {10.1016/j.isprsjprs.2023.11.008},
  URL = {https://www.documentation.ird.fr/hor/fdi:010088779},
}