Monocular Template-Based 3D Reconstruction of Extensible Surfaces for Laparoscopy
vendredi 29 mars 2013, 11h00 - 12h00
Over the past few years efforts have been made to
develop systems for computer aided laparosurgery. It consists in helping
the surgeons during an intervention to improve their perception
of the intra-operative environment: new view point synthesis, augmented reality with 3D pre-operative data (MRI, CT, etc). 3D sensing offers a virtual controllable view-point and is one of the major possible improvements to the current technology. However, due to the unpredictable, complex and elastic behaviour of the peritoneal tissues, 3D shape recovery from laparoscopic images is a difficult and open problem.
In this talk, I will first present our 3D reconstruction method which combines both motion and shading cues. The formal uses a conformal deformation prior and the latter a Cook-Torrance reflectance model.
Our method runs in two phases:
First, a 3D geometric and photometric template of the organ at rest is reconstructed in-vivo. Second, the surgeon manipulates and deforms the environment. Here, the 3D template is conformally deformed to globally match a few set of correspondences between the 2D image data provided by the monocular laparoscope and the 3D template. Then the coarse 3D shape is refined using shading cues to obtain a final 3D deformed shape. This second phase only relies on a single image. Therefore it copes with both sequential processing and self-recovery from tracking failures. I will then present our most recent result on replacing the conformal prior with a more realistic linear elastic prior.
In the final part of the talk, I will discuss the major limitations and open problem that raise from these formulations.
My papers can be downloaded from my webpage: