Computational vision techniques to support active capsule endoscopy

Abstract

The present paper considers endoscopic procedures based on actively-locomoting capsules, whose movement inside the gastrointestinal tract may be controlled independently of peristalsis. Computational vision and image processing techniques are exploited, to assist both diagnostic and capsule navigation tasks, by enhancing, interpreting and displaying information acquired by the on-board micro-camera. These techniques are applied (i) towards generating a “texture map”, namely a mosaic produced by stitching together the acquired images, and (ii) towards extracting visual navigation cues, in particular specifying the location of the lumen in these images.

5 Figures and Tables

Cite this paper

@inproceedings{Tsakiris2008ComputationalVT, title={Computational vision techniques to support active capsule endoscopy}, author={Dimitris P. Tsakiris}, year={2008} }