EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD

Transcription

1 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD D. Poli a, F. Remondino b, E. Angiuli c, G. Agugiaro b a Terra Messflug GmbH, Austria b 3D Optical Metrology Unit, Fondazione Bruno Kessler, Trento, Italy c Joint Research Center, European Commission, Ispra (VA), Italy

2 Outline Motivation Pleiades sensor description Data description Radiometric analysis Geometric processing DSM quality assessment Conclusion 22/5/2013 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD 2

3 Motivation Evaluation of 3D information extraction from VHR spaceborne image sensors on Trento testfield Analysis started with WorldView-2 and GeoEye-1 stereopairs (Melbourne, 2012) Pleiades triplet available through evaluation program organized by ASTRIUM GEO Few publications on Pleiades 22/5/2013 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD 3

4 Trento testfield Area: ca. 15x15km Min height: 150m Max height: 2181m Trento Population: inh. Height: ca. 200m 22/5/2013 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD 4

5 Trento testfield Space- and airborne imagery overview Pleiades triplet WorldView-2 stereopair GeoEye-1 stereopair Quickbird, Quickbird single scenes Aerial images (white) UAV images SPOT 22/5/2013 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD 5

6 Trento testfield other data DSM/DTM derived from a LiDAR flight in 2006/7. Cell size 1m (2m on mountainous areas). Reference dataset for the geometric analyses. 230 digital orthophotos (2006 and 2009), GSD 50 cm. Aerial flight with RMK TOP15 analogic camera (2009) over the city of Trento. GSD 12 cm. 180 UAV images (2012), acquired by a Microdrones MD4-200 mounting a Pentax Optio A40 digital camera (overlap: 80% along track, 40% across track). GSD 3 cm 70 ground points (2011), acq. by GNSS 12 ground points (2012), acq. by total station 22/5/2013 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD 6

7 Pleiades mission Part of the French-Italian ORFEO system Military and civil use Pleiades system: Pleiades 1A (launched in December 2011) Pleiades 1B (launched in December 2012) Orbit: height 694km near-polar (98.2 ) sun-synchronous 180 satellite phase 22/5/2013 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD 7

8 Pleiades images HiRi CCD cameras (1 camera/satellite) GSD 0.7m PAN (nadir) GSD 2.8m MS (nadir) Revisit time: 1 day Acquisition modes (one pass): Target, Strip Mapping, Tri-stereo Corridor and Persistent Surveillance Stereo acquisition modes: Stereo Tri-stereo 22/5/2013 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD 8

9 Pleiades images 22/5/2013 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD 9

10 Pleiade images over Trento Image 1 Image 2 Image 3 Along-track ( ) Across-track ( ) GSD (m) 0.72x x x0.74 Size (columns) Size (rows) Acquisition time 10:22:15 10:23:01 10:22:06 Solar Azimuth ( ) Solar Elevation ( ) Cloud cover /5/2013 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD 10

11 Pleiade images over Trento Image 1 Image 2 Image 3 22/5/2013 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD 11

12 Radiometric analysis Image quality and suitability are strongly affected by sensor viewing angle, sun acquisition angle and atmospheric conditions. Although not possible to compensate the effects induced by each component, through radiometric and DN analysis we can evaluate and distinguish some problems. Even if data is available at 12-bit, the main information is distributed between 0 and 10 bit. 22/5/2013 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD 12

13 Noise estimation With the term noise, we refer to all the anomalies and non-homogeneity that can be found in the image Spilling effects not present in the images Non-homogeneity noise effects ( filth analysis of homogeneous areas) Image divided into 5 bins (0-255, , , , ) For each bin i, normalized standard deviation computed for samples j in 5x5 pixel window 22/5/2013 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD 13

14 Noise estimation Channel Image Pan Pan Pan Red N/A N/A Red N/A N/A Red N/A N/A Green Green Green N/A N/A Blue Blue 2 N/A N/A N/A Blue N/A N/A NIR NIR NIR /5/2013 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD 14

15 Geometric processing Data acquisition Pleaides triplet Ground points Image orientation RPC model (RPC provided in metadata file) RPC correction with affine model Sub-pixel accuracy in check points Image matching Least-square matching Measurement of few seed points DSM generation Triangulazion and rasterization No editing Output: three DSMs (1m grid) From triplet From two pairs (1-2 and 2-3) Orthorectification 22/5/2013 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD 15

16 OpticalSCAPE ENVI module, recently released Developed by sarmap SA (Switzerland) and 4DiXplorer AG (Switzerland) Functionalities of SAT-PP software for geometric processing, plus DEM fusion (i.e. optical/sar) Advantage of ENVI environment 22/5/2013 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD 16

17 DSM quality analysis Visual inspection Quantitative comparison of DSM from triplet with Lidar Comparison between triplet and stereo pairs Comparison with WV2 and GE1 DSMs 22/5/2013 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD 17

18 DSM quality analysis DSM successfully generated from triplet and stereo pairs 22/5/2013 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD

19 DSM quality analysis DSM successfully generated from triplet and stereo pairs 22/5/2013 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD

20 DSM quality analysis DSM successfully generated from triplet and stereo pairs 22/5/2013 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD

21 Shape of the terrain in flat, hilly and mountain areas is correctly modelled, even in case of rapid height changes Urban areas: the road network, building blocks and large single buildings can be clearly recognised in the surface model, as well as the shape of large buildings Surface well modelled on homogeneous areas, like rural fields DSM quality analysis 22/5/2013 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD 21

22 Comparison with Lidar Quantitative analysis in 3 areas Computed height differences and statistics within areas (min, max, avg, median, stdev, RMSE) Trento center Historical city centre. Completely built up area, irregular building shapes, narrow streets, nearly no vegetation Trento railway station Large regular buildings, larger streets with some trees Trento Fersina Residential area. Regular geometries, courtyards and streets with trees. 22/5/2013 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD 22

23 Comparison with Lidar Trento center min: mean: 1.38 max: Median: 0.06 RMSE: 6.12 Lidar Pleiades DSM Height differences Lidar above the DSM Lidar below the DSM 22/5/2013 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD 23

24 Comparison with Lidar Trento station min: mean: 1.05 max: Median: 0.22 RMSE: 6.52 Lidar Pleiades DSM Height differences Lidar above the DSM Lidar below the DSM 22/5/2013 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD 24

25 Comparison with Lidar Trento Fersina Lidar Pleiades DSM Height differences Lidar above the DSM Lidar below the DSM min: mean: 0.47 max: Median: 0.25 RMSE: /5/2013 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD 25

26 Comparison with Lidar height profiles Profiles: Lidar, Pleiades 22/5/2013 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD 26

27 Comparison between triplet and pairs No significant differences in the statistics of the height differences with respect to the reference Lidar DSM Advantage of triplet: reduction of local blunders, mainly in presence of occlusions 22/5/2013 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD 27

28 Comparison with WV2 and GE1 Data and processing in (Agugiaro et al., ISPRS Melbourne) WorldView-2 (WV2) stereopair: Acquisition on: 22 Aug 2010, 10:40 am (GMT) PAN + MS8 (R, G, B, NIR, coastal, yellow, red edge, and NIR-2), avg GSD 0.5 m (PAN), 2 m (MS) Processing level: Stereo 1B Provided with RPCs FS, in-track view. angle 15.9, RS, Cloud cover <10% GeoEye-1 (GE1) stereopair: Acquisition on: 8 Sep 2011, 10:20 am (GMT) PAN + MS4 avg GSD 0.5 m (PAN), 2 m (MS) Processing level: GeoStereo Provided with RPCs RS, in-track view. angle 15, RS, -20 Cloud cover <4% 22/5/2013 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD 28

29 Comparison with WV2 and GE1 DSM quality analysis (1m size) Min Max mean RMSE Trento centre Pleiades WV GE Trento train station Pleiades WV GE Trento Fersina Pleiades WV GE Mean GSD: WV2 0.51m, GE1 0.51m, Pleiades 0.75m 22/5/2013 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD 29

30 Conclusions Trento testfield is update with last VHR satellite images, mainly for 3D investigations. Pleiades triplet analyzed radiometrically and geometrically Good radiometric quality Triplet and stereo DSMs (no edited) show good quality, both visually and after quantitative comparison to Lidar Similar behavior of WV2 and GE1, despite lower GSD Problematic areas are occlusions, but triplet acquisition reduces local blunders 22/5/2013 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD 30

31 Conclusions Triplet geometry of current triplet not optical. Better results are expected with nadir scene + two off-nadir Courtesy: Astrium Future work: Radiometric comparison with WV2 and GE1 Comparison of Pleiades DSM with Elevation1 product 22/5/2013 EVALUATION OF PLEIADES-1A TRIPLET ON TRENTO TESTFIELD 31

32 Thank you for your attention Credits: ASTRIUM GEO-Information Services Provincia Autonoma di Trento, Italy 3M project (co-founded Marie-Curie Actions FP7 PCOFOUND GA , acronym Trentino Project )