Imaging models for the evaluation of fiber-optic image processing algorithms

Flexible, fiber-optic image guides can be considered as the key technology in order to get access to complex hollows through entry points with limited extent. Unfortunately, their physical properties introduce artifacts to the images that hinder further image processing tasks. Fortunately, literature provides several approaches to remove this structure from the images. However, practically none of them provides an objective evaluation. This is due to an inherent problem being associated with fiber-optic imaging. Since there is only access to the subsampled image at the proximal end of the image conductor, there is no possibility to relate it to the unaffected image in front of the distal end. Hence, it is hardly possible to objectively assess the performance due to a lack of ground truth information. In this paper, we proposed a method for the synthesization of fiber-optic images with varying apertures and number of fibers and eliminate the shortcoming of state-of-the-a rt approaches that do not take these aspects into account. Consequently, this contribution enables an appropriate synthesization of artifical fiber-optic views, providing access to the ground truth information as it is required for a comprehensive evaluation of fiber-optic image processing techniques such as comb-structure removal algorithms.