Laparoscopy is a reliable imaging method to examine the liver. However, due to the limited field of view,
a lot of experience is required from the surgeon to interpret the observed anatomy. Reconstruction of organ
surfaces provide valuable additional information to the surgeon for a reliable diagnosis. Without an additional
external tracking system the structure can be recovered from feature correspondences between different frames.
In laparoscopic images blurred frames, specular reflections and inhomogeneous illumination make feature tracking
a challenging task. We propose an ego-motion estimation system for minimal invasive laparoscopy that can cope
with specular reflection, inhomogeneous illumination and blurred frames.
To obtain robust feature correspondence, the approach combines SIFT and specular reflection segmentation with
a multi-frame tracking scheme. The calibrated five-point algorithm is used with the MSAC robust estimator to
compute the motion of the endoscope from multi-frame correspondence.
The algorithm is evaluated using endoscopic videos of a phantom. The small incisions and the rigid endoscope
limit the motion in minimal invasive laparoscopy. These limitations are considered in our evaluation and are
used to analyze the accuracy of pose estimation that can be achieved by our approach. The endoscope is moved
by a robotic system and the ground truth motion is recorded.
The evaluation on typical endoscopic motion gives precise results and demonstrates the practicability of the
proposed pose estimation system.
Motion estimation based on point correspondences in two views is a classic problem in computer vision. In the
field of laparoscopic video sequences - even with state of the art algorithms - a stable motion estimation can not
be guaranteed generally. Typically, a video from a laparoscopic surgery contains sequences where the surgeon
barely moves the endoscope. Such restricted movement causes a small ratio between baseline and distance
leading to unstable estimation results. Exploiting the fact that the entire sequence is known a priori, we propose
an algorithm for keyframe selection in a sequence of images. The key idea can be expressed as follows: if all
combination of frames in a sequence are scored, the optimal solution can be described as a weighted directed
graph problem. We adapt the widely known Dijkstras Algorithm to find the best selection of frames.<sup>1</sup> The
framework for keyframe selection can be used universally to find the best combination of frames for any reliable
scoring function. For instance, forward motion ensures the most accurate camera position estimation, whereas
sideward motion is preferred in the sense of reconstruction. Based on the distribution and the disparity of
point correspondences, we propose a scoring function which is capable of detecting poorly conditioned pairs of
frames. We demonstrate the potential of the algorithm focusing on accurate camera positions. A robot system
provides ground truth data. The environment in laparoscopic videos is reflected by an industrial endoscope and