Research and Innovations

About 300 publications and 15 patents reflect the research work and innovations.  This falls into about 16 specific topics. They are presented roughly chronologically by starting with the most recent and going back in time. 

 

1. Object Reconstruction for the Automated Modeling of 3-Dimensional Objects

At one end of the spectrum this addresses 3D endosocpy and 3D-microscopy. The later led to the commercial spin-out Alicona GmbH in Graz in 2001. This general topic resurfaces below with larger objects of interest and even entire planets. 

 

2. Three-Dimensional Urban Modeling

Much larger objects are being studied in 3D mapping of the human habitat in cities.  First publications appeared in 1994. By 1998, the application to signal propagation in wireless telecommunications systems was developed. A team spun out into a separate research company VRVis GmbH in 1999; and subsequently rolled into a Microsoft-facility in Graz in 2006.

 

3. Medical Computer Vision and Augmented Reality

Liver surgery planning was first a concept and then developed into a prototype at the intersection of computer vision, computer graphics, photogrammetry and of course medicine. This represented an Augmented Reality application for the planning of liver tumor operations. In 2003, this was awarded the Eurographics Medical Prize.

 

4. Defining and Promoting Photogrammetric Computer Vision

Image pattern recognition morphed into a segment of computer science under the new name “computer vision”. Work to persuade practitioners of photogrammetry to involve themselves in computer vision culminated in the acceptance by the International Society for Photogrammetry and Remote Sensing in 2004 to rename its Technical Commission for "Theory and Algorithms" with the new name “Photogrammetric Computer Vision and Image Analysis”.

 

5. High Redundancy Photogrammetry

The fully digital end-to-end photogrammetry workflow became feasible with the introduction of digital large format aerial cameras. This produced input image blocks with previously unthinkable overlaps and thus redundancy. This achieves full automation of major photogrammetric image analysis processes. Tangible results are a fully automated workflow for image triangulation and dense matching for image-based 3D point cloud generation to compete with point clouds from LiDAR. Internal work at the Graz Institute led to a key doctoral thesis was by Andi Klaus in 2007.

 

6. Invention of a High Accuracy Large Format Digital Aerial Camera UltraCam with Software Leverage

The basic innovation consisted of a method to fuse multiple separate image tiles in a sub-micrometer manner, and to produce these separate tiles in an approach that theoretically produces all tiles in one and the same geometry. The  patent application is from 2001, the first such camera was introduced in 2003.

 

7. Invention of High Precision Photogrammetric Film Scanners with Software Leverage

1989 saw the first VX3000 scanner for large format aerial film, based on a stare-step technology and invisible réseau. The patents were applied in 1987. An entirely different technology was the basis for the initial 1999 deliveries of the UltraScan 5000, based on a push-broom digitzation approach and on a precision calibration, at a much reduced cost compared to the previous technology, and much higher radiometric capability.

 

8. Planetary Radargrammetry

At the far end ot the 3D object reconstruction scale are entire planets. This started with the 1974/5-analysis of radar imagery from the Apollo-17 Kunar Sounder Experiment [ALSE], obtained in 1972. The appointment was in 1979 as a Co/Investigator on NASA's science team for the Venus Orbital Imaging Radar (VOIR) mission, later renamed the "Magellan" mission to map planet Venus by an orbiting radar sensor. This supported the participation in numerous innovations over the project’s 15-year life time until the mission's termination in 1994.

 

9. Satellite Radargrammetry

Sensor models needed to be developed for the emerging terrestrial satellite radar systems for global mapping and for ocean observations. This concerned the early sensors such as SEASAT (1978), Shuttle imaging radars SIR-A and SIR-B, where the interest with the Shuttle sensors was on the stereo-based generation of 3D point clouds (1984). The attention for this work culminated in a cover of the journal SCIENCE in 1986.

 

10. Airborne Radargrammetry

Geometric sensor models and procedures were developed to create ortho-maps from airborne side-looking radar images (1972). These were extended to the stereo analysis of overlapping radar images (1974). Later, radar interferometric mapping was introduced into the photogrammetric application (1985). The commercial exploitation of ths tecdhnology was achieved via the then Canadian company Intera [today Intermap].

 

11. Sea Ice Radargrammetry

Initial experimental radar sea ice mapping was in 1975 from airborne SAR in the Beaufort sea from a NASA aerial platform and sensor. The subsequent extension to satellite SAR images from SEASAT served to develop systematic ice maps of the Arctic, resulting in the Alaska SAR facility using imagery from subsequent space SAR systems, and today existing as a routine data offering of value to meterology.

 

12. Automated Map Generalization

A 1984/5-need for innovative commercial services in map digitization led to the development of an automated workflow for the so-called generalization of digitized maps. While the paper map data were originally at one scale, for example at 1:25,000, they needed to get scanned, vectorized and then represented in a different scale, for example  for a visualization at 1:50,000 or1:100,000.

 

13. Pattern Recognition

1974 saw the first publications in photogrammetric media of early computer processing of images, and was denoted as Image Processing. The mathematical tools were called “Pattern Recognition” and by 1978, the first publications employed Pattern Recognition methods for remote sensing and photogrammetry problems, defining edges, lines, regions, neighborhoods etc., and applying these to map-guided image analysis.

 

14. Tools for Software-Based Classical Photogrammetric Tasks

Réseau photogrammetry, relative and absolute stereo orientation, image block triangulation and adjustment, image rectification and mapping, point positioning in 2D and in 3D were a pervasive theme of work for a long time [1969-1974]. 

 

15. Photogrammetric Aspects of Remote Sensing

First analyses of geometric sensor models for remote sensing images began in 1969 and dealt with multi-spectral scanners, thermal imaging, radar sensors and multi-line sensors. Initial publications began in 1970.

 

16. Least Squares Interpolation for Digital Elevation Models DEMs

Least squares interpolation was first introduced into photogrammetric digital terrain models DTMs in 1970. At issue was the transition of irregualry spaced elevation samples from manual observations into a dense regular grid DTM. This "densification/interpolation" was soon extended to fuse measurements from different campaigns and with different error characteristics (1972).