| Image matching | My
first task at the ifp, under the supervision of Michael Hahn. I
programmed multi-channel least squares image matching. Investigations
were e.g. published in "Hahn, M. & Brenner, C. (1995), Area
Based Matching of Colour Images". |
| Close range 3D measurement machine | The
project I was hired for at the ifp in 1994. The idea behind the
measurement machine was to replace specialized gauges
by a multi-sensor optical measurement device which first recognises the
part, automatically selects the measurement task from a database, and
finally carries out the measurement using the appropriate sensors. My
contribution was object recognition, object-model alignment, image
matching for surface reconstrutction, and also an innovative "barcode"
based calibration/ camera location approach. Published e.g. in "Brenner, C. & Fritsch, D. (1997), Object
Recognition for a Multi-Sensor Measuring System". Later the research group was extended to
the "Special Research Area" (Sonderforschungsbereich) SFB514. |
| Hybrid measurement system | This
was a research project done together with Prof. Schlemmer, Darmstadt.
The hybrid system was a combination of a theodolite and a CCD line
sensor, I contributed some of the image processing. See "Hovenbitzer,
M., Schmitt, J., Schlemmer, H., Brenner, C., Fritsch, D., Kraus, D.
(1998), Ein
hybrides Messsystem zur visuellen, dreidimensionalen Koordinaten- und
Formbestimmung in der Ingenieurvermessung". |
| Texture mapping | While
being at the NRC in Ottawa, Canada, I designed a texture mapping using
TINs and camera images. It corrects for the usual brightness
differences at triangle borders using a triangle swap, as well as a
local and global brightness adjustment. See "El-Hakim, S. F., Brenner,
C. &
Roth, G. (1998), A multi-sensor approach to creating accurate
virtual environments". The software survived at the NRC for quite
a while, see "El-Hakim, S., Gonzo, L., Picard, M., Girardi, S., Simoni,
A., Paquet, E., Viktor, H., Brenner, C. (2003), Visualisation
of Highly Textured Surfaces". |
| 3D building reconstruction from aerial laser scanning | I
indeed worked for this for a while (since around 1997). Note that I
actually tried out three different approaches. The first one is based on
the straight skeleton (contained in the 1997 Ascona proceedings "Haala,
N. & Brenner, C. (1997), Interpretation
of Urban Surface Models using 2D Building Information"). The second one
is based on a rectangular ground plan subdivision (contained in the
PhoWo '99 proceedings "Brenner, C. (1999), Interactive
modelling tools for 3D building reconstruction"). This principle
is nowadays still contained in the ATOP software.
The third one is based on a simple rule-based acceptance scheme for
planar patches, patch intersection, and a discrete relaxation/ search
tree based intersection procedure (see e.g. at the ISPRS congress
Amsterdam, "Brenner, C. (2000), Towards Fully Automatic
Generation of City Models"). All approaches are also described in my Ph.D thesis (which, however, is in German). |
| CAD based object recognition | This
work was mainly done by Jens Gühring and Jan Böhm, myself being their
supervisor at that time. We segmented planar patches in the CAD model
and the scanned object and recognized and aligned a part by using a
search tree. See "Böhm, J., Brenner, C., Gühring, J. & Fritsch, D.
(1999), CAD-basierte
Objekterkennung für ein multisensorielles Messsystem". |
| Machine safety | In
an industry project for the company Pilz GmbH & Co KG, I worked on
a robust vision system for machine safety. The result was a prototype
system, delivered in 1999. See Pat. WO0200113029A1 and others. |
| Robot calibration | This
was a joint project between the ISW (mechanical engineering) of
the University of Stuttgart and us (providing the photogrammetry part).
Mainly carried out by Jürgen Hefele in my group, we investigated the
calibration of a KUKA industry robot using a camera at the end
effector. See Hefele, J. & Brenner, C. (2000), Robot
pose correction using photogrammetric tracking". |
| Augmented reality car navigation | While being at the Robert Bosch Research facility
in Hildesheim, Germany, I designed, implemented and evaluated a
prototype for an augmented reality car navigation system which projects
virtual route guidance information (pictograms) into the driver's view
at the correct location. This required a real-time map module, GPS/Gyro
filtering (worked on by my student Christian Löffelholz), and overlay
of camera image and 3D information. |
| Weak primitives and constraints | The
idea of bundling BRep representation and constraint equations to obtain
"weak" CSG models. See "Brenner, C. (2004), Modelling
3D Objects using Weak CSG Primitives". There is also a demonstrator
on this. I also investigated the implementation of constraints beyond
"linearize & solve iteratively", see "Brenner, C. (2005),
Constraints
for Modelling Complex Objects". |
| Continuous/ streaming generalization | Joint
work together with Monika Sester. The basic idea is to store topologic and
geometric operations instead of geometry to represent geometric
entities (in this case, 2D polygons of buildings or other objects).
This representation allows for a seemingly "continuous" generalization
and also for an easy way of incremental (streaming) transmission.See "Brenner, C. & Sester, M. (2005), Continuous
Generalization for Small Mobile Displays". There is a demo software on this. |
| Extraction of landmarks for navigation systems | Being
the main topic of Birgit Elias, I provided the method to determine the
visibility of objects using existing ground plans (from the cadastre)
and aerial laser scans. See "Brenner, C., Elias, B. (2003), Extracting
Landmarks for Car Navigation Systems Using Existing GIS Databases and
Laser Scanning". There is a demo software on the visibility analysis. |
| Road extraction from aerial laser scanning | Together
with Carsten Hatger, I investigated two methods for the extraction of
roads from aerial laser scanning data, one based on the scanline
grouping approach, the other based on RANSAC. See "Hatger, C., Brenner,
C. (2003): Extraction
of Road Geometry Parameters form Laser Scanning and Existing Databases"
and also the 4/2004 PFG article. |
| Driving simulator scene generation | In
a research project carried out for the Robert Bosch GmbH, Jan-Henrik
Haunert, Hauke Neidhart and myself investigated the derivation of
driving simulator scenes from GIS data. See "Haunert, J.-H., Brenner, C., Neidhart, H. (2005), Using
a geographic information system for the generation of driving simulator
scenes". |
| Terrestrial scan alignment using the NDT | Mainly
carried out by Nora Ripperda, the idea is to align terrestrial scans
using the "Normal Distributions Transform", a fast method proposed in
robotics. See our ISPRSJ paper "Brenner, C., Dold, C., Ripperda, N. (2007), Coarse
Orientation of Terrestrial Laser Scans in Urban Environments". |
| Terrestrial scan alignment using planar patches | Being
the main research topic of Christoph Dold, terrestrial scans are
aligned without any initial information using an assignment of planar
patches. Also covered in the 2007 ISPRSJ paper. I recently added a
hierarchical selection scheme based on angle constraints, see "Brenner, C., Dold, C. (2007), Automatic
relative orientation of terrestrial laser scans using planar structures
and angle constraints". |
| Interpretation of building facades | This
is the main research topic of Nora Ripperda in my group. The basic idea
is to represent knowledge about facades in a formal grammar
(structural) and prior densities (stochastic). Then, the search space
is explored using a Markov Chain Monte Carlo approach. See "Ripperda, N., Brenner, C. (2007), Data
driven rule proposal for grammar based façade reconstruction". |
| Augmented Reality: GeoScope | I
have built the "GeoScope", a telescope-like device which allows the
on-line augmentation of a video stream with artificially generated 3D
information. It consists of a camera, a touch-screen display and
mechanical trackers for the horizontal and vertical angles. In a
project seminar, students have explored possible usage scenarios and
also built a prototype software which includes augmentation, panoramic
images, and map display using a GIS. See "Brenner, C., Paelke, V., Haunert, J. & Ripperda, N. (2006), The
GeoScope - A Mixed-Reality System for Planning and Public
Participation". |
| Driver Assistance: EnviMap | A
very exciting project. The goal is to find out which features can be
used for high precision positioning of cars using future driver
assistance sensors such as in-car LiDAR. There are well-known
algorithms for relative positioning using LiDAR, especially from
robotics, but the goal here is to produce a map beforehand and to
predict which accuracies will be feasible for car positioning when a
certain in-car sensor is given. In contrast to contemporary maps
for car navigation systems, all map features are required to be
extracted fully automatically from dense mobile mapping laser scans.The project is funded by Volkswagen Commercial Vehicles. |
