Institut fr Photogrammetrie ifp Research and Development at

  • Slides: 36
Download presentation
Institut für Photogrammetrie ifp Research and Development at ifp Keynote dedicated to the 80

Institut für Photogrammetrie ifp Research and Development at ifp Keynote dedicated to the 80 th Birthday of Professor Dr. mult. Fritz Ackermann Universität Stuttgart Dieter Fritsch, Institute for Photogrammetry (ifp), Universität Stuttgart Copyright: Institut fuer Photogrammetrie, Univ Stuttgart 1

70 th Birthday Fritz Ackermann ifp § Prof. Ackermann created the world-renowned Institute for

70 th Birthday Fritz Ackermann ifp § Prof. Ackermann created the world-renowned Institute for Photogrammetry at Univ. Stuttgart Universität Stuttgart § Supervised many national and international Ph. D students § Gave outstanding contributions to Analytical and Digital Photogrammetry § Was a visionary and pragmatic engineer who always combined basic research with technology transfer and applications! His heritage is our commitment!!! 2

Contents ifp Introduction From Oberschwaben to Vaihingen/Enz Universität Stuttgart Mapping by High Definition Airborne

Contents ifp Introduction From Oberschwaben to Vaihingen/Enz Universität Stuttgart Mapping by High Definition Airborne Imaging Come-back of Image Matching – Multiray Photogrammetry Automated Processing of Virtual City Models Conclusions Copyright: Institut fuer Photogrammetrie, Univ Stuttgart 3

1. Introduction – ifp Activities ifp Universität Stuttgart Institute for Photogrammetry (ifp) Universitaet Stuttgart

1. Introduction – ifp Activities ifp Universität Stuttgart Institute for Photogrammetry (ifp) Universitaet Stuttgart Teaching, Further Education L 3 § § § Geodesy & Geoinformatics GEOENGINE Infrastructure Planning WAREM Aerospace Engineering Environmental Engineering Research & Development (R&D) § § § Photogrammetry, RS Geoinformatics Terrestrial Positioning, Close Range Photogr. Signal Processing, Statistical Inference § 4 Technology Transfer § § The Photogrammetric Week Series (bi-annual) SW Development Consultancy Workshops (eg Qua. Con, bi-annual)

 Fritz Ackermann at Photogrammetric Week ‘ 09 ifp § Prof. Ackermann created the

Fritz Ackermann at Photogrammetric Week ‘ 09 ifp § Prof. Ackermann created the world-renowned Institute for Photogrammetry at Univ. Stuttgart Universität Stuttgart § Supervised many national and international Ph. D students § Gave outstanding contributions to Analytical and Digital Photogrammetry § Was a visionary and pragmatic engineer who always combined basic research with technology transfer and applications! His heritage is our commitment!!! 5

1. Introduction - Research through Ph. D Theses at ifp Universität Stuttgart § §

1. Introduction - Research through Ph. D Theses at ifp Universität Stuttgart § § At present, 13 doctoral theses under preparation National and international Ph. D students (50/50) Duration of a Ph. D Thesis: 3 -4 Y Individual regular structure and best-fit supervision

2. From Oberschwaben to Vaihingen/Enz Universität Stuttgart ifp Testsite Oberschwaben 7

2. From Oberschwaben to Vaihingen/Enz Universität Stuttgart ifp Testsite Oberschwaben 7

2. From Oberschwaben to Vaihingen/Enz ifp The Airborne Pushbroom Camera DPA (1995) § Universität

2. From Oberschwaben to Vaihingen/Enz ifp The Airborne Pushbroom Camera DPA (1995) § Universität Stuttgart § 8 stereo module focal length line array pix/line 10 mm resolution convergence angle spectral range spectral module focal length line array pixelsize data resolution spectral range 520 - 600 nm 610 - 685 nm 770 - 890 nm 80 mm 2 x 6000 pixelsize data 8 bit ± 25° 515 -780 nm 40 mm 6000 pix/line 10 mm 8 bit 440 - 525 nm

2. From Oberschwaben to Vaihingen/Enz German Camera Evaluation Test ifp GSD 20 cm Universität

2. From Oberschwaben to Vaihingen/Enz German Camera Evaluation Test ifp GSD 20 cm Universität Stuttgart GSD 8 cm Tutorial Phowo 2009 © Institute for Photogrammetry, Univ. Stuttgart

2. From Oberschwaben to Vaihingen/Enz Test flights, evaluated by ifp Universität Stuttgart Testf flight

2. From Oberschwaben to Vaihingen/Enz Test flights, evaluated by ifp Universität Stuttgart Testf flight Sensor Date month/year GPS/inertial components Digital line scanner DPA 07/95, 08/96 10/96, 11/98 Digital line scanner WAAC 11/97 WAAC – system specific Digital line scanner HRSC-A 02/98 POS/AV-510 DG, LR 86 Analog camera RMK-Top 15 12/98 POS/AV-510 DG, LR 86 Analog camera RMK-Top 15 06/00, 09/02 AEROcontrol-IId, IMU-IId DMC 04/03 POS/AV-510 DG, AIMU ADS 40 06/04 POS/AV-510 AIMU / LN 200, AEROcontrol-IId d. IGIcam-K 14 06/04 AEROcontrol-IId large / medium format d. IGIcam -H 39 09/07 AEROcontrol-IId 4 Head Frame Rollei AIC 4 12/07 Applanix POS/AV 510 several 06/08 ff Applanix, AEROcontrol Digital frame camera large format Digital line scanner Digital frame camera medium format Digital frame camera DGPF DAC Test DMC, DPA – system specific

ifp 3. Mapping by High Definition (HD) Digital Airborne Cameras Today’s situation Leica Geosystems

ifp 3. Mapping by High Definition (HD) Digital Airborne Cameras Today’s situation Leica Geosystems ADS 40/ADS 80 Intergraph/ZI DMC Di. MAC Systems Di. MAC Universität Stuttgart Vexcel Imaging Ultracam series D, X, Xp Rolleimetric AIC series IGI Digi. CAM series Applanix DSS series 11

3. Mapping by HD Digital Airborne Cameras Empirical results – dual Digi. CAM H/39,

3. Mapping by HD Digital Airborne Cameras Empirical results – dual Digi. CAM H/39, 17. 12. 2007 ifp Dual Digi. CAM principle Universität Stuttgart System installation Block layout – GSD 7 cm, focal length 82 mm, Dpix 6. 8 mm Tutorial Phowo 2009 © Institute for Photogrammetry, Univ. Stuttgart 12

3. Mapping by HD Airborne Digital Cameras Empirical results – dual Digi. CAM H/39

3. Mapping by HD Airborne Digital Cameras Empirical results – dual Digi. CAM H/39 ifp GCP # Ch. P # s 0 DG 0 65 AT no 32 AT 44 AT io Vers. Universität Stuttgart RMS [m] DEast DNorth DUp 4. 08 0. 045 0. 075 0. 130 33 1. 50 0. 033 0. 070 0. 134 32 33 1. 41 0. 022 0. 037 0. 088 32 33 1. 47 0. 022 0. 039 0. 096 Absolute accuracy from check point analysis, GSD 7 cm © Kremer & Cramer, 2008 http: //www. ifp. uni-stuttgart. de/publications/2008/kremer_cramer_beijing 2008. pdf Tutorial Phowo 2009 © Institute for Photogrammetry, Univ. Stuttgart 13

3. Mapping by HD Digital Airborne Imaging ifp Innovative Sensor Geometries Universität Stuttgart Vision.

3. Mapping by HD Digital Airborne Imaging ifp Innovative Sensor Geometries Universität Stuttgart Vision. Map A³ Digital sweep-framing double lens metric camera © Vision. Map, 2008

3. Mapping by HD Digital Airborne Cameras Universität Stuttgart ifp Vision. Map A³ Camera

3. Mapping by HD Digital Airborne Cameras Universität Stuttgart ifp Vision. Map A³ Camera Design Camera parameters § Digital sweep-framing double lens metric camera; § Cross-track sweep motion; § Focal lens – 300 mm; § Folded optics; § Maximal sweep FOV – 104 degree; § FMC, SMC, Vibration – mirror based optical compensation and stabilization; § Weight – 15 kg; § Size – 50*50*40 cm. Tutorial Phowo 2009 © Institute for Photogrammetry, Univ. Stuttgart 27 double frames per sweep; Along strip overlap between frames 2% Cross strip overlap between frames 15%; Overlap between two sweeps – 56%. © Vision. Map, 2008

3. Mapping by HD Digital Airborne Cameras ifp Innovative sensor geometries, Vision. Map A³

3. Mapping by HD Digital Airborne Cameras ifp Innovative sensor geometries, Vision. Map A³ Universität Stuttgart Test flight Vaihingen/Enz, June 13, 2009

3. Mapping by HD Digital Airborne Cameras ifp Innovative sensor geometries, Vision. Map A³

3. Mapping by HD Digital Airborne Cameras ifp Innovative sensor geometries, Vision. Map A³ Universität Stuttgart Test flight Vaihingen/Enz, June 13, 2009

ifp 3. Mapping by HD Digital Airborne Cameras What comes next? § Towards “all-by-one”

ifp 3. Mapping by HD Digital Airborne Cameras What comes next? § Towards “all-by-one” mapping by photogrammetric image flights: Universität Stuttgart § Automatic camera alibration, Direct Georeferencing, True Orthophoto and Digital Surface Model in one flight � Determine a very dense DSM by multi-ray image matching and compute the true ortho with the DSM § Unwrap hidden buildings in the DSM using automated reconstruction algorithms, compute DTM � Reconstruct the 3 D building vectors, texture the reconstructed building § The “all-by-one” approach will improve accuracy and efficiency, cost-saving �No need for Airborne Li. DAR in map revision processes anymore. Flight-on-demand! Copyright: Institut fuer

Universität Stuttgart ifp 4. Come-back of Image Matching – Multi-Ray Photogrammetry standard design film-based

Universität Stuttgart ifp 4. Come-back of Image Matching – Multi-Ray Photogrammetry standard design film-based ACs 60% forward and 20% across overlap Test flight digital AC 80% forward and 60% across overlap § large overlap of dense HD photo blocks § comprehensive support of automatic image matching 12. 03. 2021 19

4. Come-back of Image Matching – Multi-Ray Photogrammetry Universität Stuttgart ifp Usage of multiple

4. Come-back of Image Matching – Multi-Ray Photogrammetry Universität Stuttgart ifp Usage of multiple overlaps, 3 D point measurements § Look at all possible stereo combinations § No possible permutations § Dense sampling of 3 D mass points using § Epipolar geometry § Test of consistency by forward – backward matching 20 20

ifp 4. Come-back of Image Matching DTM/DSM 3 D Point Clouds from Image Matching

ifp 4. Come-back of Image Matching DTM/DSM 3 D Point Clouds from Image Matching (N. Haala, ifp DMC 2009) Ultracam-X Digi. CAM Universität Stuttgart 8 cm GSD 20 cm GSD DGPF-DSM-Jena 12. 03. 2021 21

4. Come-back of Image Matching Sportfield Rosswag RMS of 3 D Point Clouds (N.

4. Come-back of Image Matching Sportfield Rosswag RMS of 3 D Point Clouds (N. Haala, ifp 2009) ifp RMS [cm] all Pts 1. 92 Density [Pts/m 2] 8. 25 [%] Elim. Pts 0. 54 ZMean [m] Ref. Li. DAR GSD 8 cm DMC 5. 16 9. 74 19. 67 1. 30 253. 065 Ultracam-X 6. 84 8. 04 19. 04 0. 49 253. 067 Digi. CAM 10. 18 11. 28 20. 83 0. 68 253. 023 RMK 17. 23 27. 27 0. 77 3. 18 253. 002 DMC 17. 23 25. 36 2. 71 1. 10 252. 896 Ultracam-X 22. 57 34. 21 1. 62 0. 43 252. 980 Digi. CAM 34. 13 48. 24 2. 64 2. 45 253. 030 RMK 60. 56 66. 21 0. 31 0. 73 253. 187 GSD 20 cm Universität Stuttgart RMS [cm] Filt. Pts 1. 83 § 253. 036 Image matching generates 3 D point clouds of good quality (GSD 8 cm) § RMS=7. 4 cm, 19. 8[Pts/m 2] § Average of 3 investigated digital photogrammetric airborne camera systems § Partly problems using GSD 20 cm GSD § RMS= 24. 6 cm, 2. 3[Pts/m 2] 12. 03. 2021 22

ifp 4. Come-back of 2 D Image Correlation (Numerical Simulation) Ph. D Thesis F.

ifp 4. Come-back of 2 D Image Correlation (Numerical Simulation) Ph. D Thesis F. Blumrich, ifp 2009 Amount of the max error (in pixel) for sub-pixel detection Universität Stuttgart 1 D sub-pixel detector (state-ofthe-art) general 2 D linear Gaussian regression (new) The systematic error of the existing 1 D sub-pixel detectors were analytical described and validated A general 2 D linear sub-pixel detector has been developed and tested, which is very fast (real-time apps) and up to 1000 x more precise

ifp 5. Automated Processing of Virtual City Models - The ifp Approach Li. D

ifp 5. Automated Processing of Virtual City Models - The ifp Approach Li. D AR 3 D Building Unwrapping (Vectorisation) Apps 3 D Building Texturing (Rendering) 3 D Building Generalisation e. M ag Im Universität Stuttgart atc hin g 3 D Point Cloud (>4 points/m 2) Apps Ph. D Theses: § § § 3 D Building Re. Texturing (Re-Rendering) 3 D Building Structures (Features) Apps N. Haala (1996) B. Ameri (2000) C. Brenner (2001) M. Kada (2007) S. Becker (2010) M. Peter (2011) 24

5. Automated Processing of Virtual City Models 3 D Berlin, Project Overview ifp §

5. Automated Processing of Virtual City Models 3 D Berlin, Project Overview ifp § Project execution: § virtualcity. SYSTEMS § Reconstruction of building geometry § Autodesk Gmb. H § Texturing from oblique images Universität Stuttgart § Project duration: § East Berlin (March 2008 – July 2008) § West Berlin (November 2008 – February 2009) § 3 D-Reconstruction software developed by § virtualcity. SYSTEMS § Institute for Photogrammetry (ifp), University of Stuttgart

5. Automated Processing of Virtual City Models Universität Stuttgart ifp 3 D Berlin -

5. Automated Processing of Virtual City Models Universität Stuttgart ifp 3 D Berlin - Examples of Simple Shapes and their Combination to 3 D-Building Models

5. Automated Processing of Virtual City Models 3 D Berlin, Reconstruction Approach ifp Universität

5. Automated Processing of Virtual City Models 3 D Berlin, Reconstruction Approach ifp Universität Stuttgart Basic Shapes Flat Shed Connecting Shapes Gabled Hipped „Berliner“ Special Shapes

5. Automated Processing of Virtual City Models ifp Building Unwrapping, Reconstruction Approach (ifp 1998,

5. Automated Processing of Virtual City Models ifp Building Unwrapping, Reconstruction Approach (ifp 1998, 2007) § Automatic reconstruction of building geometry from Universität Stuttgart § Digital surface model (4 ppsqm for Berlin) § Ground plans (e. g. from the authoritative real estate cadastre) 1. Generate a generalized ground plan decomposition in disjoint (quadrilateral) cells 2. Classify the LIDAR points according to the orientation of their local regression plane 3. Find for each cell the shape that best fits the classified points

5. Automated Processing of Virtual City Models ifp Rendering and Image Abstraction (ifp 2009)

5. Automated Processing of Virtual City Models ifp Rendering and Image Abstraction (ifp 2009) § Hiding window contents § Detection of window locations by approaches from building reconstruction § Gray out / set to high reflectance Universität Stuttgart § Removal of objects on the façade § Detection using OCR techniques § Copy other parts of the texture § Non-photorealistic visualization § Cartoon-like textures § Recognize spatial situation hide details § Generic textures, single color § Determine number and arrangement of windows Copyright: Inst. Photogrammetry, Univ. Stuttgart

5. Automated Processing of Virtual City Models ifp Privacy-Enabling Abstraction, Geometry abstraction (ifp 2009)

5. Automated Processing of Virtual City Models ifp Privacy-Enabling Abstraction, Geometry abstraction (ifp 2009) § Geometric simplification § e. g. removed dormer § Façade typification Universität Stuttgart § Change number and arrangement of façade elements § Standard roof shapes, 3 D building symbols § Typification of building groups § Aggregation of building blocks § Hide borders of individual buildings Copyright: Inst. Photogrammetry, Univ. Stuttgart

5. Automated Processing of Virtual City Models Universität Stuttgart Data driven bottom-up § Interpretation

5. Automated Processing of Virtual City Models Universität Stuttgart Data driven bottom-up § Interpretation of measured data + § Flexible + § Accuracy and quality control §- Sensitive to erroneous or incomplete data Explicit geometric facade modelling Fully automatic Copyright: Institut fuer Photogrammetrie, Univ Stuttgart Knowledge based topdown Facade Reconstruction ifp Automated Feature Extraction - Facade Reconstruction Combined Knowledge Propagation § Formal grammars (e. g. shape rules) §+ Topological correctness §+ Completeness §- Variety of generated structures is restricted to knowledge base Quality sensitive

5. Automated Processing of Virtual City Models Automated Feature Extraction through Knowledge Propagation (Ph.

5. Automated Processing of Virtual City Models Automated Feature Extraction through Knowledge Propagation (Ph. D Thesis S. Becker, ifp 2010) ifp § Data Driven Facade Reconstruction bottom - up § § § Knowledge Inference top - down § § Detection of dominant or repetitive features and regularities Interrelationship between geometries Inference of production rules Knowledge Propagation § top - down Universität Stuttgart § Cell decomposition Extraction and modelling of facade geometries from terrestrial Li. DAR data Refinement based on facade imagery § Top-down prediction for verification and completion Generation of synthetic facades Copyright: Institut fuer Photogrammetrie, Univ Stuttgart Facade Grammar

5. Automated Processing of Virtual City Models Automated Feature Extraction, Results Ph. D Thesis

5. Automated Processing of Virtual City Models Automated Feature Extraction, Results Ph. D Thesis S. Becker, ifp 2010 ifp Schillerplatz, Stuttgart Terrestrial Li. DAR (Leica HDS 3000) Facade imagery (NIKON D 2 x) Universität Stuttgart § § Copyright: Institut fuer Photogrammetrie, Univ Stuttgart Lindenmuseum, Stuttgart § § Terrestrial Li. DAR (Street. Mapper) Facade imagery (NIKON D 2 x)

6. Virtual City Models and Image Understanding Universität Stuttgart ifp Stuttgart, Schillerplatz Copyright: Institut

6. Virtual City Models and Image Understanding Universität Stuttgart ifp Stuttgart, Schillerplatz Copyright: Institut fuer Photogrammetrie, Univ Stuttgart

ifp 6. Conclusions. . . Thanks to Fritz Ackermann § Prof. Ackermann created the

ifp 6. Conclusions. . . Thanks to Fritz Ackermann § Prof. Ackermann created the world-renowned Institute for Photogrammetry at Univ. Stuttgart Universität Stuttgart § Supervised many national and international Ph. D students § Gave outstanding contributions to Analytical and Digital Photogrammetry § Was a visionary and pragmatic engineer who always combined basic research with technology transfer and applications! His heritage is our commitment!!! Copyright: Institut fuer Photogrammetrie, Univ Stuttgart 35

ifp 6. Happy Birthday Fritz Ackermann - The next Milestone: the 90 th Birthday

ifp 6. Happy Birthday Fritz Ackermann - The next Milestone: the 90 th Birthday § Prof. Ackermann created the world-renowned Institute for Photogrammetry at Univ. Stuttgart Universität Stuttgart § Supervised many national and international Ph. D students § Gave outstanding contributions to Analytical and Digital Photogrammetry § Was a visionary and pragmatic engineer who always combined basic research with technology transfer and applications! His heritage is our commitment!!! Copyright: Institut fuer Photogrammetrie, Univ Stuttgart 36