The News at the Chair for Computer Aided Procedures & Augmented Reality

• Dipl. Inf. Andreas Hofhauser: Deformable object detection @ Chair for Computer Aided Medical Procedures and Augmented Reality, TU Muenchen.
• Ph.D. Giorgio Panin: Multi-Modal Face and Object Tracking Strategies @ Chair of Robotics and Embedded Systems, TU Muenchen.
• Dipl. Inf. Wolfgang Sepp: Efficient Shape-Texture based Tracking with the Image-Constancy Assumption in 3-d @ Institute of Robotics and Mechatronics, German Aerospace Center (DLR).
• Dr.-Ing. Didier Stricker: Presentation of the activities of the department of "Virtual and Augmented Reality" @ Fraunhofer Institute for Computer Graphics (IGD) in Darmstadt.

• Irfan Essa: Presentation of the Interactive and Intelligent Computing Division (College of Computing - Georgia Institute of Technology, Atlanta)
• Vincent Lepetit: Feature Harvesting for Tracking-by-Detection (Ecole Polytechnique Federale de Lausanne)
• Matthias Grundmann: A real-time learning-free texture-based tracking algorithm (TU München)
• Andreas Hofhauser: Deformable object detection (TU München).

Abstract: Medical Augmented Reality: state of art, recent developments and remaining challenges

In this talk I first present the definition, requirements and state of art in medical augmented reality. I then describe the two main approaches of augmenting the medical imaging device and augmentation of physicians' view of procedures. In the latter case, I discuss different solutions for such view augmentation. These include the use of external monitors, video and optical see-through displays and surgical microscopes. I will finally present several applications in medical training as well as minimally invasive surgery. I will focus on presentation of three main systems and their applications:

I) Camera Augmented Mobile C-arm (CAMC): developed and tested by the Chair for Computer Aided Medical Procedures and Augmented Reality (CAMP-AR) at Technical University of Munich and Siemens Medical in collaboration with Trauma Surgery Department of Innenstadt Hospital in Munich.

II) Reality Augmented Medical Procedures (RAMP): originally developed at Siemens Corporate Research by Sauer, Khamene et al. Currently the system is extensively used and integrated in different medical applications for medical training and minimally invasive surgery at CAMP-AR. I will show results obtained in collaboration with Orthopedics department of Hospital rechts der Isar, and within the NARVIS project in partnership with ART, SCR and Trauma Surgery department of Hospital Innenstadt.

III) Endoscopic Augmentation for Minimally invasive surgery: This system is developed at CAMP-AR and has been integrated into scenarios defined in collaboration with Munich Heart Center and Internal Surgery Department of Hospital Rechts der Isar.

The above systems are used to illustrate the technical issues, existing capabilities and future challenges in the field of Medical Augmented Reality.

Abstract: Image Registration by Combining Thin-Plate Spline Mappings With a 3D Morphable Model

Registering images of a deforming surface is a well-studied problem. It is common practice to describe the image deformation fields with Thin-Plate Splines. This has the advantage to involve small numbers of parameters, but has the drawback that the 3D surface is not explicitly reconstructed. In this talk, I shall describe an image deformation model combining Thin-Plate Splines with 3D entities - a 3D control mesh and a camera - overcoming the above mentioned drawback. An original solution to the non-rigid image registration problem using this model will be presented and demonstrated on simulated and real data.

Topic: 3D Object Tracking and Detection for Augmented Reality

Abstract: Estimation of Non-Rigid Registrations

In this talk, I will present our methods for registering images of non-rigid surfaces, which is a well-studied problem. Solutions include computing optical flow or estimating a parameterized transformation. In the case of optical flow, regularization is necessary. Our method is based on representing the inter-image transformations by Radial Basis Mappings (RBM). The method can be viewed as a direct method, i.e. intensity-based, or as a way of using RBMs as nonlinear regularizers on the optic flow field. I will show the results we obtained on challenging image sequences. Finally, I will shortly present the recent non-rigid factorization algorithms for point features, and the point and curve based algorithm we proposed.
This is joint work with Andrew Zisserman from Oxford University, UK.

The talk takes place at 10:30 in room 03.013.010.

Abstract: ARCave – or how to realize a CAVE(TM) without Screens?

Immersive or semi-immersive, stereoscopic displays, such as CAVES, walls, or workbenches are frequently used in industrial applications for scientific visualizations, simulations or industrial presentations. Their projection screens are built into dedicated rooms and are consequently not portable. What if we could transform a real room into an immersive virtual environment or into a semi-immersive mixture of real and virtual spaces? What if we could use everyday surfaces as displays, and make them temporarily appear or disappear – fully or partially? And what, if the technology that makes this possible was portable to allow this in any room? Which would be the applications for it? Maybe such a technology can support architects by visualizing their construction data and lighting simulations – but inside the real building lot rather than on a desktop monitor. Maybe it would allow visitors of historical sites to dive into a multimedia presentation of ancient times – without destroying the ambience of a castle’s vaults by a permanent installation of projection screens. Or maybe it would simply transform our living rooms into the arena of a 3D game. This talk will outline our early efforts and current results that lead us a small step forward towards this vision. It describes new real-time and pixel-precise rendering techniques for radiometric compensation, geometric image warping, and multi-focal projection. All these components are realized with state-of-the-art pixel-shading technology. They enable a view-dependent stereoscopic projection in real environments.

Oliver Bimber
Bauhaus-University Weimar
http://www.uni-weimar.de/medien/AR

The talk takes place at 13 o'clock in room 03.013.010.