Geometric Analysis of DMC II 14 Karsten Jacobsen Leibniz Universität Hannover jacobsen@ipi.uni-hannover.de
DMC II 14 Geometry determined by panchromatic camera Panchromatic camera: focal length: 92.52 mm Image format: in flight direction 112 pixel * 7.2µm = 8.64mm across flight direction 1296 pixel * 7.2µm = 87.91mm r max = 59.345mm - for 1µm accuracy in position at image corner flatness (or knowledge about height variation) of 1.6µm required field of view in flight direction: 47.33 across flight direction: 5.66 image base (p=6%) = 32.256mm height to base relation (p=6%): 1:.356 = 2.85 -one homogenous CCD-array no problems by stitching
flight over test field Aalen, Germany 5.7cm GSD end lap 65%, side lap 65% + crossing flight with same overlap 9.5cm GSD end lap 62%, side lap 62% + crossing flight with same overlap Test field Aalen 71 targeted GCPs SX=SY=SZ= 2cm - 3cm h=74m 2.2cm GSD end lap 8%, side lap 83% + crossing flight with same overlap
Test flight 5.7cm GSD hg=728m footprint 64m x 691m Overlay of all image points to one image Overlap of images 144 images 96 up to 328 points/image, in average 232 points/image In average 7.6 images / object point 6 5 4 3 2 Images/point 24 1 1, 3, 5, 7, 9, 11, 13, 15, 17, 18, 21, 23,
Self calibration by additional parameters Hannover program system BLUH: additional parameters with physical justification 12 basic additional parameters for all types of images + camera specific parameters for stitched images Additional parameters 81 88 for determination and correction of corner effects often seen at mid-format CCD-cameras parameter 81 radial parameter 88 tangential Only use of required additional parameters Typical systematic image errors of mid-format CCD-camera (not Intergraph) maximal: 11µm
Bundle block adjustment of whole block 5.7cm GSD Overlay and local averaging of all 33 457 residuals with self calibration RMS =.14µm, max=.62µm remaining systematic image errors negligible Overlay and local averaging of all 33 457 residuals without self calibration RMS =.14µm, max=.62µm Indicating very small size of systematic image errors µm µm add. parameter,81-88 8 GCP, 48 check points at check points: σo.98µm.97µm.97µm RMSX 2.9cm 2.9cm 2.9cm.13pixel.51GSD RMSY 2.6cm 2.6cm 2.6cm RMSZ 4.6cm 2.2cm 2.1cm.46GSD.37GSD SX, SY < SZ Accuracy limited by check point accuracy
Systematic image errors - block 5.7cm GSD Systematic image errors minus radial symmetric distortion RMS =.1µm maximal:.31µm Systematic image errors RMS =.27µm maximal: 1.3µm Systematic image errors extremely small Radial symmetric distortion up to 1.µm
flight 5.7cm GSD p=q=65% (double blocks) Only East-West flight lines, p=q=65% 72 images, in average 4.8 images/object point Only North- South flight lines, p=q=65% 72 images, in average 4.5 images/object point additional parameter,81-88 additional parameter,81-88 8 GCP, 48 check points at check points: σo.89µm.89µm.89µm σo.88µm.88µm.88µm RMSX 3.cm 2.9cm 2.9cm RMSX 3.3cm 3.1cm 3.2cm RMSY 3.2cm 3.1cm 3.1cm RMSY 2.6cm 2.6cm 2.6cm RMSZ 4.9cm 2.4cm 2.4cm.12pixel.51GSD.54GSD.42GSD 8 GCP, 48 check points at check points: RMSZ 3.9cm 3.6cm 3.8cm.12pixel.56GSD.46GSD.67GSD
flight 5.7cm GSD p=65% q=37% 36 images (single blocks) 6 6 Only North- South flight lines, every 2 nd line, p=65%, q=37% 36 images, in average 2.8 images/object point Only North- South flight lines, every 2 nd line, p=65%, q=37% 36 images, in average 2.8 images/object point additional parameter,81-88 9 GCP, 37 check points at check points: additional parameter σo RMSX RMSY RMSZ,81-88 1 GCP, 45 check points at check points: σo.82µm.82µm.82µm.87µm.86µm.86µm RMSX 2.9cm 2.7cm 2.7cm.11pixel.47GSD 3.2cm 3.1cm 3.2cm.12pixel.56GSD RMSY 2.8cm 2.8cm 2.9cm 2.6cm 2.6cm 2.7cm RMSZ 3.9cm 2.5cm 3.cm.51GSD.53GSD 4.1cm 4.2cm 3.7cm.47GSD.65GSD3 Even with single coverage at check points in sub-pixel accuracy
4-times coverage double coverage 5.7cm GSD overview [GSD] single coverage GSD.9.8 RMSZ at check points.7.6.5.4.3.2.1 2 p=q=65% EW NS NS p=64% p=q=65% q=37% Left columns = without self calibration Center columns = additional parameters Right columns = additional parameters, 81-88 RMSX, RMSY ~.5 GSD (=2.8cm) independent upon coverage and self calibration = limited by check points RMSZ with self calibration.4 up to.7 GSD = excellent
Test flight 9.5cm GSD hg=123m footprint 16m x 1145m Overlap of images 68 images 91 up to 329 points/image, in average 232 points/image In average 6.6 images / object point base = 421m corresponding t=7sec for speed 12 knots 18 9.5cm GSD end lap 62%, side lap 62% + crossing flight with same overlap
Systematic image errors - block 9.5cm GSD Systematic image errors minus radial symmetric distortion RMS =.1µm maximal:.53µm Systematic image errors RMS =.2µm maximal: 1.5µm Systematic image errors extremely small Radial symmetric distortion up to.6µm
Bundle block adjustment of whole block 9.5cm GSD end lap 62%, side lap 62% + crossing flight with same overlap 6 GCP, 28 check points at check points: add. parameter,81-88 σo RMSX RMSY RMSZ.91µm 3.5cm 3.cm 6.9cm.9µm 3.5cm 3.1cm 4.6cm.9µm 3.5cm 3.1cm 4.6cm.12pixel.37GSD.33GSD.48GSD Influence of self calibration limited to Z-component 9.5cm GSD end lap 62%, side lap 62% + crossing flight with same overlap
flight 9.5cm GSD p=q=62% (double block) Only East-West flight lines, p=q=62% 36 images, in average 3.5 images/object point additional parameter,81-88 6 GCP, 26 check points at check points: σo.93µm.91µm.91µm RMSX 4.cm 4.5cm 4.4cm RMSY 2.7cm 4.4cm 4.4cm RMSZ 6.9cm 4.9cm 4.6cm.13pixel.46GSD.46GSD.48GSD flight lines + GCP
flight 9.5cm GSD p=62% q=24% 18 images (single blocks) 7 GCP, 2 check points 1 st sub-block at check points: Only East-West flight lines, every 2 nd line, p=62%, q=24% 18 images, in average 2.5 images/object point 6 additional parameter,81-88 σo.9µm.89µm.89µm RMSX 4.2cm 4.3cm 4.3cm.12pixel.45GSD RMSY 2.6cm 3.8cm 3.7cm RMSZ 1.1cm 5.7cm 5.7cm.39GSD.6GSD 7 GCP, 21 check points 2 nd sub-block at check points: additional parameter,81-88 σo.93µm.91µm.91µm RMSX 4.8cm 5.7cm 5.8cm.13pixel.61GSD RMSY 3.2cm 3.4cm 3.3cm RMSZ 8.2cm 7.cm 6.9cm.35GSD.73GSD
9.5cm GSD overview [GSD] 4-times coverage double coverage single coverage RMSX, RMSY ~.4 -.6 GSD slightly depending upon coverage and self calibration = limited by check points RMSZ with self calibration.4 up to.7 GSD = excellent Left columns = without self calibration Center columns = additional parameters Right columns = additional parameters, 81-88
Test flight 2.2 cm GSD hg=2585m footprint 2266m x 2447m 36 images 169 up to 367 points/image, in average 273 points/image In average 8. images / object point Overlap of images 32 end lap 8%, side lap 83% + crossing flight with same overlap
Systematic image errors - block 2.2cm GSD Systematic image errors minus radial symmetric distortion RMS =.2µm maximal:.85µm Systematic image errors RMS =.62µm maximal: 1.5µm Systematic image errors extremely small Radial symmetric distortion up to 2.µm
Bundle block adjustment of whole block 2.2cm GSD 6 GCP, 22 check points at check points: add. parameter,81-88 σo RMSX RMSY RMSZ 1.68µm 6.6cm 14.4cm 17.9cm 1.68µm 6.7cm 14.4cm 16.6cm 1.68µm 6.7cm 14.3cm 16.6cm.23pixel.33GSD.71GSD.82GSD No improvement by self calibration Larger σo-value as other flying heights -Problems with identification of targeted control and check points 2cm GSD too large for test field Aalen
flight 2.2cm GSD only East-West flight lines 6 GCP, 22 check points at check points: additional parameter σo RMSX RMSY RMSZ 1.4µm 8.1cm 13.3cm 12.cm 1.39µm 8.3cm 14.9cm 13.7cm,81-88 1.39µm 8.3cm 14.9cm 13.9cm 16 images, p=68%,q=83%.19pixel.41gsd.74gsd.69gsd 6 GCP, 19 check points at check points: additional parameter σo RMSX RMSY RMSZ 1.52µm 8.5cm 13.7cm 13.9cm 1.51µm 9.cm 15.cm 14.7cm,81-88 1.51µm 9.cm 15.cm 14.7cm 8 images, p=6%,q=83%.21pixel.45gsd.74gsd.73gsd
8-times coverage 2.2cm GSD overview [GSD] 4-times coverage double coverage RMSX, RMSY ~.4 -.7 GSD slightly depending upon coverage and self calibration = limited by check point identification RMSZ with self calibration.6 up to.8 GSD Influence of too small size of control and check points, nevertheless good result Left columns = without self calibration Center columns = additional parameters Right columns = additional parameters, 81-88
Overview all ground resolutions [GSD] RMSZ=.37GSD 5.7cm GSD: RMSX, RMSY limited by accuracy of control and check points 9.5cm GSD: real accuracy achievable by DMC II 14 2.2cm GSD: not optimal block configuration, limited by identification of targets in images
Conclusion Unusual small size of systematic image errors of DMC II 14 images, typical corner effects of large size CCD-arrays not significant, special additional parameters 81-88 for compensating corner effects not required Small radial symmetric image errors, slightly depending upon flying height (<1.µm for 5.7cm GSD, <.6µm for 9.5cm GSD, < 2.µm for 2.2cm GSD) Systematic image errors without radial symmetric effects: 5.7cm GSD: RMS=.1µm, maximal.31µm 9.5cm GSD: RMS=.1µm, maximal.53µm 2.2cm GSD: RMS=.2µm, maximal.85µm Reached accuracy at independent check points: for 5.7cm GSD and 9.5cm GSD RMSX, RMSY:.3.5 GSD RMSZ: depending upon multiple coverage.4.66 GSD Relation vertical to horizontal accuracy better as height/base ratio of 2.85 ( accuracy of check and control points cannot be neglected) Excellent image geometry and excellent object point accuracy by DMC II 14 -images