Final Report Summary - FLEXABLE (Deformable Multiple-View Geometry and 3D Reconstruction, with Application to Minimally Invasive Surgery)
Project FLEXABLE is about 3D computer vision, the field of computer science which studies how to make a computer understand its 3D environment from digital images and videos. By drawing on the very successful geometric approach described over the past decades for the rigid objects, project FLEXABLE has proposed a fundamentally new general approach applicable to rigid and deformable objects. With this approach, a computer is now able to reason geometrically about the images of common deformable objects and infer their time-varying 3D structure. Project FLEXABLE studied the use of the inter-image displacement as the main visual cue and complemented it with other visual cues such as shading. These theoretical and computational results were applied to computer-aided surgery in laparoscopy, for which an augmented reality based gesture guidance system was developed within the project. With this system, the surgeon can see the internal anatomy through the organs and optimize their gesture to improve tumour resection and enhance safety with respect to sensitive structures such as vessels.