High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE...
Transcript of High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE...
![Page 1: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/1.jpg)
1
HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY
A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH Zurich, CH-8092 Zurich, Switzerland
http://www.photogrammetry.ethz.ch
JACIE Civil Commercial Imagery Evaluation Workshop, March 14-16, 2006, Laurel, Maryland, U.S.A.
https://ntrs.nasa.gov/search.jsp?R=20070038223 2020-02-05T23:55:03+00:00Z
![Page 2: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/2.jpg)
2
outline• introduction
• sensor modeling of HRSI
• DTM/DSM generation from HRSI
• performance evaluation
• 3D city modeling with HRSI
• conclusions
![Page 3: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/3.jpg)
3
• High-resolution PAN & MS imagery
+ Quickbird (0.7 m)
+ IKONOS (1.0 m)
+ SPOT (2.5-10 m)
+ ALOS / PRISM (2.5 m) …….
• More than 8-bit images, higher dynamic range
• Along- / cross-track stereo;
Possibly multiple view terrain coverage
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
• Challenge:
+ Algorithmic redesign
+ Improvements
![Page 4: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/4.jpg)
4
Original IKONOS Image
Enhanced IKONOS Image
More than 8-bit images, higher dynamic range
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 5: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/5.jpg)
5
Acquisition Time: 2003-Dec-11
Acquisition Time: 2003-Dec-25• Along- / cross-track stereo;
Possibly multiple view terrain coverage
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 6: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/6.jpg)
6
SAT-PP (Satellite Image Precision Processing) -- High-Res Satellite Imagery (HRSI): ≤ 5 m
+ New Processing Methods / Products for HRSI
+ Joint Sensor Model for IKONOS, QuickBird, SPOT, ALOS/RPISM and etc.
+ Specially Designed Image Matching Procedure for Linear Array Imagery
Input
Imagery with Supplementary Data
+ IKONOS
+ QuickBird
+ SPOT ……
GCPs
SAT-PP
+ Image Orientation
+ Image Matching
+ Feature Collection
+ 3D City Modeling
+ Image Pan-Sharpening
+ DTM & Ortho-Image Generation
…….
Products
+ DSM / DTM
+ Ortho-Images
+ Pan-Sharpened
Images
+ 3D Vector Data
+ Digital Image Maps
+ 3D City Models
…….
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 7: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/7.jpg)
7
Functionality of SAT-PPProject and data management tools, image format conversion and
pre-processing, image display / roaming in mono and stereo modes
Sensor models (RFM, affine and projective DLT model)
Orientation of single stereo models
On-line quality control and error analysis via interaction of graphics elements
GCP and tie point measurement in manual and semi-automated modes
Derivation of quasi-epipolar images for stereo mapping and feature collection
Automated DSMs generation
Generation of orthorectified images
Mono-plotting functions with DTMs
Manual and semi-automatic object extraction in mono/stereo
3D city modeling by using CyberCity ModelerTM
Pansharpening image generation. Fully automated sub-pixel image registration between multispectral and panchromatic imagery
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 8: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/8.jpg)
8
Workflow of SAT-PP
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 9: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/9.jpg)
9
Sensor Modeling and Blockadjustment• Rigorous sensor model
+ Physical imaging geometry (nearly parallel projection in along-track and perspective projection in cross-
track); high accuracy; easier for statistic analysis
- Mathematically more complicated; depends on type of sensors
• Sensor model based on RFM
+ Given (for IKONOS, Quickbird) and computed RFM parameters (RPCs)
• Blockadjustment model (Grodecki & Dial; 2003)
+ Calibrated system with a very narrow FOV; accurate a priori exterior orientation data (HRSI -- OK !)
• Other simpler sensor models
+ 3D affine; relief-corrected 2D affine; DLT …...
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 10: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/10.jpg)
10
For SPOT5 HRS / HRG
Fitting accuracy: 1/20 - 1/100 pixels
Tao & Hu; 2001
Sensor model based on RFM
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 11: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/11.jpg)
11
User interface for block adjustment
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 12: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/12.jpg)
12
Ellipse fitting mode
Line intersection mode
Ellipse fitting method GCP measurement
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 13: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/13.jpg)
13
Detailed DSM Generation
The approach uses a coarse-to-fine hierarchical solution with an effective combination of several image matching algorithms and automatic quality control.
The new characteristics provided by the IKONOS and Quickbird imaging systems, i.e. the multiple-view terrain coverage and the high quality image data, are also efficiently utilized.
It was originally developed for multi-image processing of the very high-resolution TLS/StarImager aerial Linear Array images. Now it has been extended and has the ability to process other linear array images as well.
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 14: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/14.jpg)
14
Workflow of Automated DSM Generation
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 15: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/15.jpg)
15
• Multiple image matching
+ Matching guided from object space
+ Simultaneouslly multiple images (>= 2) with
Geometrically Constrained Cross-Correlation
• Matching with multiple primitives --- points + edges
• Self-tuning matching parameters
• High matching redundancy
• Efficient surface modeling
+ TIN (from a constrained Delauney triangulation method)
• Coarse-to-fine Hierarchical strategy
Automated DSM Generation Procedure
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 16: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/16.jpg)
16
Matching guided from object space
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 17: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/17.jpg)
17
Self-tuning matching parameters
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 18: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/18.jpg)
18
RPC + 2 TranslatesGCPs(CPs) RMSE-X RMSE-Y RMSE-Z
0 (124) 2.75 2.00 1.97
1 (123) 0.48 0.35 0.90
4 (120) 0.49 0.36 0.86
124 (0) 0.45 0.33 0.81
IKONOS Triplet, Hobart, Australia
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 19: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/19.jpg)
19
RPC + 2 TranslatesGCPs(CPs) RMSE-X RMSE-Y RMSE-Z
0 (124) 2.75 2.00 1.97
1 (123) 0.48 0.35 0.90
4 (120) 0.49 0.36 0.86
124 (0) 0.45 0.33 0.81
IKONOS Triplet, Hobart, Australia
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 20: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/20.jpg)
20
RPC + 2 TranslatesGCPs(CPs) RMSE-X RMSE-Y RMSE-Z
0 (124) 2.75 2.00 1.97
1 (123) 0.48 0.35 0.90
4 (120) 0.49 0.36 0.86
124 (0) 0.45 0.33 0.81
IKONOS Triplet, Hobart, Australia
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 21: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/21.jpg)
21
RPC + 2 TranslatesGCPs(CPs) RMSE-X RMSE-Y RMSE-Z
0 (124) 2.75 2.00 1.97
1 (123) 0.48 0.35 0.90
4 (120) 0.49 0.36 0.86
124 (0) 0.45 0.33 0.81
IKONOS Triplet, Hobart, Australia
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 22: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/22.jpg)
22
Automatic DTM/DSM Generation (IKONOS, Hobart, Australia)
Raster DSM (5 m Spacing)
+ The strip encompassed buildings and suburban housing
in central and southern Hobart; accuracy 0.25 m
+ 111 GCPs as reference:
RMSE-Z: 0.9 m; Mean: -0.3 m; Absolute max.: 2.9 m
+ LIDAR DSM as reference (ca. 252000 points):
RMSE-Z: 2.7 m; Mean: -0.2 m; Absolute max.: 29.6 m
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 23: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/23.jpg)
23
RPC + 2 TranslatesGCPs(CPs) RMSE-X RMSE-Y RMSE-Z
0 (43) 23.11 25.17 75.76
1 (42) 4.69 4.38 2.26
4 (39) 4.68 4.35 2.25
43 (0) 4.63 3.66 2.20
SPOT5-HRS, Bavaria, Germany
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 24: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/24.jpg)
24
RPC + 2 TranslatesGCPs(CPs) RMSE-X RMSE-Y RMSE-Z
0 (43) 23.11 25.17 75.76
1 (42) 4.69 4.38 2.26
4 (39) 4.68 4.35 2.25
43 (0) 4.63 3.66 2.20
SPOT5-HRS, Bavaria, Germany
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 25: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/25.jpg)
25
RPC + 2 TranslatesGCPs(CPs) RMSE-X RMSE-Y RMSE-Z
0 (43) 23.11 25.17 75.76
1 (42) 4.69 4.38 2.26
4 (39) 4.68 4.35 2.25
43 (0) 4.63 3.66 2.20
SPOT5-HRS, Bavaria, Germany
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 26: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/26.jpg)
26
RPC + 2 TranslatesGCPs(CPs) RMSE-X RMSE-Y RMSE-Z
0 (43) 23.11 25.17 75.76
1 (42) 4.69 4.38 2.26
4 (39) 4.68 4.35 2.25
43 (0) 4.63 3.66 2.20
SPOT5-HRS, Bavaria, Germany
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 27: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/27.jpg)
27
Study area: Bavaria, Germany
+ Area: 120 × 60 Km2
+ Height range: ca. 1600 m
SPOT HRS stereo pair
+ Acquisition time: 1st October, 2002
+ 5m / 10m res. In along-/cross-track
Reference data:
+ 81 GPS GCPs (only 41 used)
+ 6 reference DTMs
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
Automatic DTM/DSM Generation (SPOT5-HRS, Bavaria, Germany)
![Page 28: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/28.jpg)
28
Raster DSM (25 m Spacing,120 × 60 km2)
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
Automatic DTM/DSM Generation (SPOT5-HRS, Bavaria, Germany)
![Page 29: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/29.jpg)
29
Reference DSM (5 m) SPOT5 DSM (25 m)
Automatic DTM/DSM Generation (SPOT5-HRS, Bavaria, Germany)
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 30: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/30.jpg)
30
Reference DSM (25 m) SPOT5 DSM (25 m)
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
Automatic DTM/DSM Generation (SPOT5-HRS, Bavaria, Germany)
![Page 31: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/31.jpg)
31
DSM Accuracy (All Reference Data)
DSM Accuracy (Without Trees)
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
Automatic DTM/DSM Generation (SPOT5-HRS, Bavaria, Germany)
![Page 32: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/32.jpg)
32
IKONOS (5 Images, 2 Strips), Town of Thun, SwitzerlandSub-pixel accuracy in planimetry; ca. pixel accuracy in height
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
IKONOS Images, Thun, Switzerland
![Page 33: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/33.jpg)
33
IKONOS (5 Images, 2 Strips), Town of Thun, Switzerland
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
IKONOS Images, Thun, SwitzerlandDecreasing number of GCPs doesn’t decreasing the accuracy significantly
![Page 34: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/34.jpg)
34
IKONOS (5 Images, 2 Strips), Town of Thun, Switzerland
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
IKONOS Images, Thun, SwitzerlandEven M_3DAFF could achieve similar results (for IKONOS imagery)
![Page 35: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/35.jpg)
35
Study area: town of Thun, Switzerland
+ Area: 17 × 20 Km2
+ Height Range: 1600 m
IKONOS Geo Product
Reference
+ 2m spacing LIDAR DSM as reference
accuracy: 0.5 m (1σ) for open areas;
1.5 m for vegetation & build-up areas
+ 50 GPS GCPs (only 39 used)
Automatic DTM/DSM Generation (IKONOS, Thun, Switzerland)
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 36: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/36.jpg)
36
Raster DSM (5 m Spacing) Generated from IKONOS ImagesINTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
Automatic DTM/DSM Generation (IKONOS, Thun, Switzerland)
![Page 37: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/37.jpg)
37
LIDAR DSM (2 m) IKONOS DSM (5 m)INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
Automatic DTM/DSM Generation (IKONOS, Thun, Switzerland)
![Page 38: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/38.jpg)
38
LIDAR DSM (2 m) IKONOS DSM (5 m)
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
Automatic DTM/DSM Generation (IKONOS, Thun, Switzerland)
![Page 39: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/39.jpg)
39
LIDAR DSM (2 m) IKONOS DSM (5 m)
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
Automatic DTM/DSM Generation (IKONOS, Thun, Switzerland)
![Page 40: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/40.jpg)
40
Z_diff = LIDAR_DSM_Z - Interpolation(IKONOS_DSM)
Z_diff = Matched_POINT_Z - Interpolation(LIDAR_DSM)(dense LIDAR points --> Less surface modeling errors)
+ Point number: ca. 14,327,000
+ RMSE: 3.30 m
+ Mean: -0.32 mINTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
Automatic DTM/DSM Generation (IKONOS, Thun, Switzerland)
![Page 41: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/41.jpg)
41
Semi-automated Feature Extraction with SAT-PP
An extracted building from an IKONOS stereopair. The left building is measured manually and the right one is matched automatically.
- Currently available for some kind of objects, such as points, lines and polygons
- The user only needs to measure, for example, the outlines of buildings in one image. The correspondences of building outlines in other images are computed automatically.
- User intervention is possible for editing the polygon/line nodes when mismatching occurs
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 42: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/42.jpg)
42
CyberCity Modeler approach, from stereo images and laser data
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 43: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/43.jpg)
43
Point Cloud Coding System
• Independence regarding used DPS
• Reduction of manual labour
CyberCity-MODELERTM
Roof modeling
3D City Model
Visualization(Formats: FLT, VT etc.)
CAD Systems(Formats: DXF etc.)
Geometry improvement
Roof texturing
Digital Photogrammetric
Station
Point cloud (x, y, z)
ArcGIS 9(Shape, PGDB, ArcSDE)
+ DTM
Qual
ity C
ontro
l
Facade texturing
Wall generation
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 44: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/44.jpg)
44
TEXTURING
Generic Texturing
• Texture library
• Not realistic
• Regional texture types
• Automatic
Automatic Texturing
• (Oblique-) Aerial Imagery
• Realistic
• Automatic
Terrestrial Texturing
• Digital Photographs
• Realistic / High resolution
• Manually applied
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 45: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/45.jpg)
45
3D Object Extraction From IKONOS ImageryInput & Data Pre-processing
Left stereo Right stereo
Date, time (local) 16/7/2000,09:53
16/7/2000,09:53
Sensor azimuth (°) 136.7 71.9
Sensor elevation (°) 61.4 60.7
Sun azimuth (°) 38.2 38.3
Sun elevation (°) 21.1 21.0
IKONOS Melbourne Stereopair7x7 km areaelevation range of less than 100 m32 GPS-surveyed ground measured semi-automatically by ellipse-fitting method
orientation was based on the supplied RPCs parameters (from Space Imaging) plus additional 6 affine transformation parameters in image space.the RMSEs of orientation are 0.4 meters in planimetry and 0.9 meters in height.
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 46: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/46.jpg)
46
3D Object Extraction From IKONOS Imagery
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 47: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/47.jpg)
47
3D Object Extraction From IKONOS ImageryMeasurement area overview
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 48: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/48.jpg)
48
Generated 3D city model
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
3D Object Extraction From IKONOS Imagery
![Page 49: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/49.jpg)
49
• Quickbird stereo images over Phoenix, USA
- Acquired on 9 April 2004
- Along track stereo images
- GSD: 75cm (mean)
- Viewing angles: 29º, - 27º
®DigitalGlobe
forward backward
3D City Modeling from Quickbird
• Detail in Quickbird images
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 50: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/50.jpg)
50
Facade textures from library
3D City Modeling from Quickbird
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 51: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/51.jpg)
51
3D City Modeling from Quickbird
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 52: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/52.jpg)
52
3D City Modeling from Quickbird
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 53: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/53.jpg)
53
Conclusions
SAT-PP: sophisticated image pre-processing algorithms, a set of sensor models, an image matching approach for DSM generation and feature extraction from HRSI.
Sensor modeling and block adjustment:
Basically three types of sensor-model orientation concepts at our disposal:
a) rigorous/physical sensor model
b) Rational Functional Model (RFM) with given RPCs
c) 2D affine model, possibly with added parameters
d) 3D affine and DLT models
Precise (sub-pixel) GCP / tie point collection (LSM) in semi-automatic model
Sub-pixel orientation accuracy can be achieved for all models
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 54: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/54.jpg)
54
Automatic DSM/DTM generation:
Reproduces not only general features, but also detailed features of the terrain reliefHeight accuracy of around 1 pixel in cooperative terrain
RMSE values of 1.3-1.5 m (1.0-2.0 pixels) for IKONOSRMSE values of 2.9-4.6 m (0.5-1.0 pixels) for SPOT5 HRS
3D city modeling:
Conclusions
The manual and semi-automatic feature extraction capability of SAT-PP provides a good basis also for 3D city modeling applications with CyberCity-ModelerTM (CCM).
The tools of SAT-PP allowed the stereo-measurements of points on the roofs in order to generate a 3D city model with CCM. Additional features of CCM allow roof and facade texturing.
The results show that building models with main roof structures can be successfully extracted by HRSI. As expected, with Quickbird more details are visible.
INTRODUCTION SENSOR MODELING DSM GENERATION CASE STUDIES CONCLUSIONSCITY MODELING
![Page 55: High Accuracy 3D Processing of Satellite Imagery · 1 HIGH ACCURACY 3D PROCESSING OF SATELLITE IMAGERY A. Gruen, L. Zhang, S. Kocaman Institute of Geodesy and Photogrammetry, ETH](https://reader033.fdocuments.in/reader033/viewer/2022041616/5e3b55da7023de79ff24e8b2/html5/thumbnails/55.jpg)
55
Acknowledgements
Prof. Fraser, Department of Geomatics, University of Melbourne
Mr. Volpe, Eurimage S.p.A.
Dr. Poli and Dr. Wang, CyberCity AG
Space Imaging, U.S.A.
NPOC (National Point of Contact) - swisstopo, Bern