Criteria for Digital Camera Image Quality
Recent Developments in Digital Airborne Cameras 11th Annual Z/I Imaging® Camera Conference Keystone, September 21, 2007 Prof. Ralf Reulke, Dr. Andreas Eckardt
Motivation
New possibilities & available technology becomes important for digital camera development and applications Digital sensors (VIS, IR, LIDAR, RADAR, hyperspectral systems) Direct geo-referencing (determination of EO for each camera frame or line) Direct coreferencing and alignment of high resolution stereo channels with the multispectral channels allows combined stereo & remote sensing
Folie 2 > Image Quality > Reulke
Motivation
Absolute radiometry (remote sensing) Geometric & radiometric lab-calibration of the whole sensor (as function of air pressure & temperature) Real time processing capabilities Wireless data transmission New applications – Virtual glob
Folie 3 > Image Quality > Reulke
3D globe viewer 3D globe viewer with elevations and satellite images (Virtual Globe) Major players in the Virtual Globe arena are Google Earth, Microsoft’s Virtual Earth and NASA World Wind Search for locations through queries and user-interface controls Add data onto the map, like roads, political boundaries and basic image overlays Find out information about local businesses, driving directions and other interesting thinks through this concept Through API’s and XML-based interfaces, developers and advanced users can create new functions and data products (http://charlotte.utdallas.edu/mgis/ClassFiles/gisc6383/techassess_2005/VG_Report.doc)
Folie 4 > Image Quality > Reulke
Outline
Sensors Requirements Detector Arrays, Resolution (Improvement), Technology Terrestrial, Airborne and Space Borne Imaging Systems Quality assessment for imaging systems
New Applications Sensor & Data Fusion Radiometry / Classification
Conclusions
Folie 5 > Image Quality > Reulke
Digital Camera-Systems
Discrete (digital) Image Discretize element
Continues (analog) filtered image + EO EO determination GPS & INS
Platform
Motion of the camera
Optical system (& filter)
Digital photogrammetric camera-system Continues (analogues) Object Folie 6 > Image Quality > Reulke
Requirements (Geometry) Number of pixel in each image direction Area to cover and object resolution E.g. A=2 ·2 km² & Δ=20 ·20 cm² ↔ 10,000 ·10,000 pixel = 100 MPixel
Resolution = ground sampling distance (GSD) Sampling theorem, well suited optics
Optical system requirement from Pixel distance (PD) Optics resolution [lp/mm] = 1000 / (2 · Pixel Distance [µm]) E.g. Δ=10µm → 50 lp/mm
Image smear (optics, pixel size, airborne platform movement) can be described by Gaussian σ≈0.5..1 PD Airplane speed ≈ 70 m/s, GSD=10cm → 1ms integration time= 0.7·Δ smear
Folie 7 > Image Quality > Reulke
System Parameters Field of View (Degree) Swath width (km) Instantaneous FOV (micro radian) Number of spectral bands and spectral ranges Quantization (Dynamic Range) The Dynamic Range of the Detector is defined by the ratio of saturation output to RMS noise in the dark. The Dynamic Range for all pixels is in the range of 12-14 Bit
MTF (Modulation Transfer Function) curves Weighted MTF for the Sensor at Nyquist frequency and at a fixed readout frequency
Folie 8 > Image Quality > Reulke
Optics Parameter Effective Focal length (mm) Aperture - F number Wavelength range (Nanometer) Aberration (spherical, coma, field curvature, astigmatism, distortion, lateral color) PSF / MTF
Folie 9 > Image Quality > Reulke
Focal plane / detector subsystem Detector types, number of pixels Effective Pixel distance (pitch) and pixel size (micron) Spectral sensitivity, spectral bands Charge saturation (electron) quantum efficiency (electron/photon), Scale factor, responsivity Pixel rate and line rate Shutter options and readout timing. Focal plane dimensions Temperature of Operation
Folie 10 > Image Quality > Reulke
Focal plane / detector subsystem Quantization levels and dynamic range (number of effective bits) Linearity Blooming (%) The sensor system shall have an anti blooming provision in across track direction Smear (%)
Folie 11 > Image Quality > Reulke
Sensor Design Detector array CCD-Matrix, CCD-line
Optics Assembling Number of focal plane How to fill the gaps
Folie 12 > Image Quality > Reulke
High Resolution Detector-Arrays, CCD-Matrices KAF-31600
6496 x 4872 6.8µm square pixel 7216 x 5412
31.6Mpixel
Kodak
39Mpixel
Kodak
33Mpixel
DALSA
????
5kx6.6k 7.2µm square pixel 10.56x10.56k
111Mpixel
DALSA
CCD 595
9kx9k
81Mpixel
Fairchild Imaging
KAF-39000 FTF5066
High resolution matrices (> 100 MPixel) are available http://www.dalsa.com/news/news.asp?itemID=252 Commercial high resolution photo systems in a price range up to 10 T€ have > 20 Mega-Pixel Matrices ≈100 Mpixel matrices seem to expansive for standard applications Folie 13 > Image Quality > Reulke
Digitale Backs eyelike Jenoptik Hasselblad (former: Imacon) (http://www.hasselblad.com/) Leaf Leaf Aptus 75, 33 Mio. Pixel (Dalsa?)
PhaseOne (http://www.phaseone.com/) PHASE ONE’S 39 Mio. Pixel (Kodak?)
Sinar (http://www.sinar.ch/) Sinarback eMotion75 with 33 million pixels
Digital backs for airborne systems
Folie 14 > Image Quality > Reulke
New Sensor Technologies The Feature: Sensors with very high frame rate Fairchild Imaging has produced 9k x 9k sensors for many years 2 f/s The Challenge: 8k x 8k @ 1000 f/s Most challenging requirement is 67 Gigapixel data rate Fairchild Imaging 1000 f/s sensors produced for many years Fairchild Imaging CCD 595 9k x 9k CCD Folie 15 > Image Quality > Reulke
New Sensor Technologies Hybrid Low Light Level Applications
Xinqiao (Chiao) Liu, Boyd A. Fowler, Steve K. Onishi, Paul Vu, David D. Wen, Hung Do, and Stuart Horna, CCD / CMOS Hybrid FPA for Low Light Level Imaging, Fairchild Imaging, Inc., & U.S. Army Night Vision and Electronic Sensors Directorate Combines CCD imaging characteristics (e.g. high quantum efficiency, low dark current, excellent uniformity, and low pixel cross talk) with High speed, low power and ultra-low read noise of CMOS readout technology http://www.fairchildimaging.com/main/documents/CCD_CMOS_Hybrid_FPA_for_Low_Light_Level_Imaging.pdf
Folie 16 > Image Quality > Reulke
New Sensor Technologies Hybrid Low Light Level Applications Microscopy Live cell fluorescence Fixed cell
QE[400nm]:>75%
Confocal
X-ray Imaging Radiography Fluoroscopy X-ray crystallography
Astronomy & Space Research Adaptive optic wave front sensor Startrackers Environmental sensing
Night Vision Near Term -- aircraft, vehicle, fire control Medium Term -- manportable Folie 17 > Image Quality > Reulke
New Sensor Technologies High Resolution RGB - Sensors In difference to film typical RGB-sensors has a filter raster on the chip
Foveon X3 image sensors have three layers of pixel sensors Sigma SD9 (2268 x 1512 pixel)
Folie 18 > Image Quality > Reulke
http://www.foveon.com/X3_tech.html
dIGIcam-K14 Geometrische Auflösung Colour red
green
blue
22cm
28cm
Parameter Hight a.g. 550m Pixel size 8μm Focal legth 28mm Scale
19000
GSD (theor.) 16cm
31cm
Bayer pattern Colour interpolation Undersampling: full detail/colour not attained Blur filters to reduce aliasing artifacts
Bayer pattern decomposition Folie 19 > Image Quality > Reulke
High Resolution Detector-Arrays, CMOS-Detector
Lower power usage Integration of additional circuit on-chip Lower system cost Direct pixel access Non-linear response characteristics
Folie 20 > Image Quality > Reulke
(Geometric) Resolution Improvement Subsempling techniques, only for still imaging JenScan- Camera (Kontron-Progress-Camera) Heimann Biometric Systems Typical CCD-matrix with much smaller pixel size, moving in a sub-pixel range From 3x3 up to 6x6 sub-sampling steps Problems: SNR & MTF / moving objects
Using more than one array Problem: Filling the gaps
Folie 21 > Image Quality > Reulke
Resolution Improvement The JVC Camcorder GC-QX3U uses AIS (Accurate Image Shift). Lens features high quality optics including two aspherical lenses Image-shift technology that doubles the image data UXGA (1600 x 1200) 1.92 Megapixel Digital Stills with Pixel Shift Technology
See staggered array – SPOT 5 Folie 22 > Image Quality > Reulke
High Resolution Detector-Arrays, CCD-Lines
Manufacturer
Model
Photopixel
Size [µm²]
Atmel Atmel EEV
TH7834 customise CCD21-40
12000 2×12000 12288
6.5×6.5 6.5×6.5 8×8
KODAK
KLI-10203
3×10200
7×7
KODAK
KLI-14403
3×14204
5×5
Fairchild Imaging
CCD194
12000
10×8.5
SONY
ILX734K
3×10500
8×8
Folie 23 > Image Quality > Reulke
EyeScan – a High Resolution Panoramic Camera (KST & DLR)
Number of pixel
3∗10200 (RGB)
Radiometric dynamic / resolution
14 bit / 8 bit per channel
Shutterspeed
4 ms ... 512 ms
Datarate
2.5 Msamples/s/channel 15 Mbyte/s
Datavolume for 360° scan
3 Gbyte
Acquisition time
4 min
Power supply
12 V
Folie 24 > Image Quality > Reulke
High resolution panoramic camera Example
Auckland, NZ Folie 25 > Image Quality > Reulke
Low Light Panoramic Camera (DLR) Designed for basic research of TDI systems: • Synchronization issues • Geometric calibration of TDI sensors • DSNU & PRNU issue in dependency to the temperature • MTF measurement • SNR measurements
Folie 26 > Image Quality > Reulke
Digital Photogrammetric Airborne Camera Systems (Matrix based)
Commercial Systems DMC (Intergraph) UltraCamD Vexcel Imaging Austria ADS40 (Leica Geosystems) HRSC other
Folie 27 > Image Quality > Reulke
Digital Mapping Camera System (DMC) 4 high-resolution 7k x 4k PAN camera heads 4 multispectral 3k x 2k camera heads Camera electronic unit 3 FDS units, each with 576 GB disk space for the storage of 4,400 images total
http://www.intergraph.com/dmc/default.asp Folie 28 > Image Quality > Reulke
Pre-processing of DMC-image data DMC pre-processing relative orientation of the 4 camera heads is known geometric & radiometric correction (mosaicking) Image with (quasi-) central perspective Fusion with colour channels
© Z/I-Imaging, 2001
Folie 29 > Image Quality > Reulke
DMC, example images Left: (digitized) film - Right: digital camera
©Z/I
©Z/I
Vaihingen, test flight, mai 2003 Folie 30 > Image Quality > Reulke
Other Systems ROLLEI AIC modular LS, AIC integral CCD-chip with 4080 x 4076 pixels (16 Mpixel) 4080 x 5440 pixels (22 MPixel)
RS232 connection for remote control IEEE1394 data interface Burstrates up to 2.5 sec. / frm.
Folie 31 > Image Quality > Reulke
dIGIcam-K14 Geometrische Auflösung
Geometric resolution targets (Siemens star, bar pattern)Folie 32 > Image Quality > Reulke
dIGIcam-K14 Geometrische Auflösung Colour red
green
blue
22cm
28cm
Parameter Hight a.g. 550m Pixel size 8μm Focal legth 28mm Scale
19000
GSD (theor.) 16cm ADS40pattern high-end sensor Bayer Height 500m Colour a.g. interpolation Undersampling: Pixel size 6.5μm full detail/colour not Focal length 62.77mm attained scale 8000 Blur filters to reduce GSD (theo.) 5.5cm aliasing artifacts
31cm
Bayer pattern decomposition Resolution Pan 6.6cm
Folie 33 > Image Quality > Reulke
Airborne-Camera Alpa (CH)
Airborne-camera with two lenses and two digital 33-MPBacks ACN = AirCamNetwork
http://www.alpa.ch/files/knowledgebase/18/ALPA_ACN.pdf Folie 34 > Image Quality > Reulke
CCD-line Stereo Cameras Airborne cameras 3 line Principle for along track cameras Derenyi, University of New Brunswick, Canada 1970 Hofmann about 1985
Space borne cameras
Folie 35 > Image Quality > Reulke
Airborne CCD-Line Scanner WAOSS (Wide Angle Optoelectronic Stereo Scanner) WAAC (Wide Angle Airborne Scanner) HRSC (High Resolution Stereo Camera) ADC / ADS40 (Airborne Digital Camera / Sensor) MEOSS (Monocular Electro-Optical Stereo Scanner) MOMS-02 (Modular Optoelectronic Multispectral Stereo Scanner ) DPA (Digital Photogrammetric Assembly) TLC
Folie 36 > Image Quality > Reulke
ADC-EM2, Reichstag, 23.4.99, h=3 km, Δ≈25cm
Folie 37 > Image Quality > Reulke
Digital Photogrammetric Airborne Camera Systems High Resolution Frame Cameras with ≥100 Mpixel are availible Minimum GSD < 10 cm Radiometric dynamic ≈ 12 bit Direct georeferencing is possible
Folie 38 > Image Quality > Reulke
High-resolution Imaging Sensors on Satellite Platforms High-resolution mapping means a pixel distance between 0.5 m to 3 m. Spot 0.5 to 2 m systems Ikonos and Eros, QuickBird, OrbView 3 Governments (France, Japan, China-Brazil, India) have announced and launch high-resolution systems http://www.ipi.uni-hannover.de/html/publikationen/2006/paper/KJ_Damaskus.pdf
Folie 39 > Image Quality > Reulke
High-resolution Imaging Sensors on Satellite Platforms (IKONOS) Digital camera systems was designed and built by Eastman Kodak Company, Rochester, NY., Collect panchromatic (gray-scale) imagery with one-meter resolution, and multispectral data (red, green, blue, and near infrared) with four-meter resolution First image 1999 (Washington memorial) Altitude
681 kilometers
Inclination
98.1 degrees
Speed
7 km/s
Orbit time
98 minutes Sunsynchronous
Orbit type
Folie 40 > Image Quality > Reulke
IKONOS - Space Imaging: Rom Petersplatz
Folie 41 > Image Quality > Reulke
Quick Bird - Digital Globe: Rom Petersplatz
Folie 42 > Image Quality > Reulke
CEU Development KompSat 3 [EADS Astrium GmbH & DLR] Nr. of Pixels PAN PAN-Sensor Line Rate PAN CCD Output Rate Data Rate MS-Sensor Line Rate MS CCD Outp. Rate/Colour Data Rate Pitch PAN Pitch MS Anti Blooming Operating temperature Image Plane dx Dynamic Range PRNU DSNU SNR-PAN SNR MS Orbit Focal Length F-# PAN NIR RED GREEN BLUE
24.000 2 x 12.080-TDI 10 kHz +5/-50 % 16 x 15MPixel/s 3,84 Gbit/s 8 x 6.000-TDI 2,5 kHz +5/-50 % 2 x 7,5 MPixel/s 4 x 240 Mbit/s 8,75 µm 2 x 17,5 µm yes 10°-25°C 22 cm 14 Bit yes yes >200 >200 685 km 8,6 m 12 450 nm-900 nm 760 nm-900 nm 630 nm-690 nm 520 nm-600 nm 450 nm-520 nm
DLR is responsible for The Focal Plane and FEE Development
Folie 43 > Image Quality > Reulke
Determination of Sensor Parameter – Spatial and Radiometric Resolution Measurement principles PSF / MTF, spatial resolution Signal calculation, radiometric resolution
Folie 44 > Image Quality > Reulke
Image Quality – Spatial Image Resolution How to Determine Resolution Data? PSF & MTF evaluation u
Point source in the object space
Imaging
Gray level v distribution in the image space
y‘
Film MTF(KODAK AEROCOLOR III Negative Film 2444)
y
x‘
x
Resolution charts
Folie 45 > Image Quality > Reulke
Spatial Resolution, Rayleigh – Criterion Diffraction Limit of a Telescope α= 1,22 λ / D (in rad)
dPixel = 1,22 λ F# (in m)
Folie 46 > Image Quality > Reulke
Image Quality – Radiometric Image Resolution Dynamic range > 12 bit (4096 gray levels) Sensor electronics noise 0.1 ... 0.5 % from dynamic range Signal calculation E = π/4 f# · cos4θ ∫dλ Tλoptics · Tλup · Lλtarget λ Signal depends from s = ηDV ⋅ηV ⋅ηel ⋅τ⋅ A ⋅ ⋅ E = τ⋅ R DN ⋅ E [ DN] hc (spectral) illumination Athmosphere & clouds Spectral transmission of the otical system (lenses & filter) Aperture Pixelsize / -area Responsivity of the detector array Integration time
Only integration time and aperture (?) are variable Spectral limitations (e.g.for RGB) increase radiometric problems Folie 47 > Image Quality > Reulke
SNR Model
SNR =
ne ne + σ CCD _ rms 2 + σ channel 2 + σ ADC 2
ne : Photon Noise σCCD_rms : CCD rms Noise σchannel : Electronics channel ADC σADC :
Folie 48 > Image Quality > Reulke
Electronically increasing of the aperture
1
GSD
2
n
Optik
TDI Principe
t1[µs] dx µm
1
t2[µs]
2
tn[µs]
n
Advantage: Photon _ SNRTDI = Photon _ SNR ∗ Nr _ TDI _ Steps
Folie 49 > Image Quality > Reulke
Radiometric Resolution
Reichstag, Berlin
Sonnenbereich
Schattenbereich
Folie 50 > Image Quality > Reulke
System Quality Criteria Depends from application Standards? German standard DIN 18740-4 “Anforderungen an digitale Luftbildkameras und an digitale Luftbilder” NATO STANAG 3769 (Minimum Resolved Object Sizes and Scales for Imagery Interpretation) http://cartome.org/min-rez.htm National Image Interpretability Rating Scales (NIIRS) (http://www.fas.org/irp/imint/niirs.htm) General Image-Quality Equation: GIQE (http://adsabs.harvard.edu/abs/1997ApOpt..36.8322L)
Folie 51 > Image Quality > Reulke
NATO STANAG 3769 (Minimum Resolved Object Sizes and Scales for Imagery Interpretation) Definitions Detection: In imagery interpretation, the discovering of the existence of an object but without recognition of the object. Recognition: The ability to fix the identity of a feature or object on imagery within a group type, ie, tank, aircraft. Identification: The ability to place the identity of a feature or object on imagery as a precise type. Technical Analysis: The ability to describe precisely a feature, object or component imaged on film.
Folie 52 > Image Quality > Reulke
Spatial Resolution
Folie 53 > Image Quality > Reulke
NATO STANAG 3769 (Minimum Resolved Object Sizes and Scales for Imagery Interpretation)
Object
Detection
Recognition
I Identification
Technical Analysis
~ 800 m
90 m
3m
0.75 m
Urban Areas
60 m
15 m
3m
0.75 m
Urban Areas
30 m
6m
1.5 m
0.4 m
Surface Ships
15 m
4.5 m
0.15 m
0.04 m
Coast and Landing Beaches
15 m
4.5 m
1.5 m
0.4 m
Bridges
6m
4.5 m
1.5 m
0.3 m
Airfield Facilities
6m
4.5 m
3m
0.15 m
Aircraft
4.5 m
1.5 m
0.15 m
0.04 m
Vehicles
1.5 m
0.5 m
0.15 m
0.04 m
Terrain
Folie 54 > Image Quality > Reulke
National Image Interpretability Rating Scales (NIIRS) (http://www.fas.org/irp/imint/niirs.htm) The aerial imaging community utilizes the National Imagery Interpretability Rating Scale (NIIRS) to define and measure the quality of images and performance of imaging systems Through a process referred to as "rating" an image, the NIIRS is used by imagery analysts to assign a number which indicates the interpretability of a given image The NIIRS concept provides a means to directly relate the quality of an image to the interpretation tasks for which it may be used NIIRS provides a systematic approach to measuring the quality of photographic or digital imagery, the performance of image capture devices, and the effects of image processing algorithms Folie 55 > Image Quality > Reulke
General Image-Quality Equation: GIQE A regression-based model relating aerial image quality, expressed in terms of NIIRS, to fundamental image attributes GIQE treats three main attributes: scale, expressed as the ground-sampled distance; sharpness, measured from the system modulation transfer function; and the signal-to-noise ratio
The GIQE can be applied to any visible sensor and predicts NIIRS ratings with a standard error of 0.3 NIIRS The image attributes treated by the GIQE are influenced by system design and operation parameters The GIQE allows system designers and operators to perform trade-offs for the optimization of image quality Jon C. Leachtenauer, William Malila, John Irvine, Linda Colburn, and Nanette Salvaggio, General Image-Quality Equation: GIQE, Applied Optics, Vol. 36, Issue 32, pp. 8322-8328 Folie 56 > Image Quality > Reulke
Fusion Sensor fusion with complementary and competitive Sensors Limitations from measurement principles (e.g. with CCD-camera have a limited spectral range) Different types of cameras: Day & night vision: VIS, RGB – TIR Different types of measurement principles (Camera & laser scanner
Competitive Sensors More than one camera of the same type observe the same area (camera node) – improvement of accuracy
References: L. Klein; Sensor & DataFusion, SPIE press http://www2.informatik.hu-berlin.de/cv/ (Anforderungen an geometrische Fusionsverfahren, Workshop, 20. November 2006)
Folie 57 > Image Quality > Reulke
Fusion „Data Fusion is a formal frame work in which are expressed means and tools for the alliance of data originating from different sources. It aims at obtaining information of greater quality; the exact definition of ‚greater quality‘ will depend upon the application.“ Wald, L. (1999): Some terms of reference in data fusion. IEEE Transactions on Geosciences and Remote Sensing, 37(3), pp. 1190-1193
Folie 58 > Image Quality > Reulke
Fusion Fusion of high resolution image data with Data from different spectral channels Laser scanner Radar Hyperspectral systems
http://www2.informatik.huberlin.de/cv/index.php?seite=3&sub=4&site=./conferences/Fusion_Workshop_11.2006/index. html
Folie 59 > Image Quality > Reulke
Airborne Laserscanner http://www.photogrammetry.ethz.ch/summerschool/pdf/08_Brenner_aeri al_scanner.pdf
Airborne laser scanning principles Laser Semiconductor lasers or Nd:YAG lasers pumped by semiconductor lasers, emits at λ = 1064 nm (near infrared) 810 nm (ScaLARS), 900 nm (FLI-MAP), 1540 nm (TopoSys) Pulsed laser ↔ Continuous laser operation
Folie 60 > Image Quality > Reulke
Resolution of LIDAR data „Typical“ LIDAR data Row data 1-2m point distance DSM 1m Raster DEM 10m, 5 m, 1m Raster Image data 25 cm GSD Artifacts in the ortho image due to weak LIDAR data resolution
Folie 61 > Image Quality > Reulke
The Mount Etna Case Study: A Multisensor View
German Aerospace Research Establishment (DLR), Germany R. Horn, K. P. Papathanassiou, A. Reigber, R. Scheiber (Institut für Hochfrequenztechnik) P. Hausknecht, P. Strobl, R. Boehl (Institut für Optoelektronik) M. Scheele, R. Reulke, W. Baerwald (Institut für Weltraumsensorik) G. Puglisi, M. Coltelli (Istituto Internazionale di Vulcanologia), G. Fornaro (Istituto di Ricerca per L’Elettromagnetismo e i Componenti Elettronici), Consiglio Nazionale delle Ricerche (CNR), Italy
Folie 62 > Image Quality > Reulke
The experiment objectives The generation of a high precision DEM using airborne microwave and optical sensors The retrieval of surface parameters using radar polarimetry and optical spectrometry The investigation of the synergy potential of optical and microwave sensors The exploration of the limits of airborne differential SAR interferometry
Folie 63 > Image Quality > Reulke
Results
Folie 64 > Image Quality > Reulke
Results
Folie 65 > Image Quality > Reulke
Results
Folie 66 > Image Quality > Reulke
Results
Folie 67 > Image Quality > Reulke
Results Geometry - Layover, foreshortening, shadowing. Critical issues for radar. GPS - Positioning accuracy depends on GPS signal quality. This is crucial for geocoding to be performed as required. Weather - Clouds influence the optical part of multisensor systems, radar is more flexible. Synergy - The information delivered by optical and microwave sensors is complementary.
Folie 68 > Image Quality > Reulke
Radiometrie / Classification New business with derived products RapidEye: http://www.rapideye.de RapidEye is a commercially funded provider of geospatial information and customized solutions Offer geospatial management information customized to the needs of clients in the Agriculture, Forestry, Oil & Gas, Environmental and Governmental markets Based on five-band multispectral image data sets acquired by RapidEye satellites. Products can be radiometrically and geometrically corrected Derived data are vegetation index, image classification, etc. Folie 69 > Image Quality > Reulke
Atmospheric Correction Radiation components for the @sensor radiance Reflected radiation (LReflected) from a certain pixel Path radiance (LPath) - photons scattered into the sensor’s instantaneous field-of-view, without having ground contact ρ ⋅ Eg L = LPath + LRe flected = LPath + τ ⋅ = c0 + c1 ⋅ DN π τ, ρ and Eg are the ground-to-sensor atmospheric transmittance, surface reflectance, and irradiance on the ground Or for the surface reflectance ρ =π ⋅
(c0 + c1 ⋅ DN ) − LPath τ ⋅ Eg
Folie 70 > Image Quality > Reulke
Lab-Measurement Concept based on Ulbricht-sphere → L (radiance) Absolute accuracy of the signal of the Ulbricht-sphere is about 2 % Integration time: 7 ms Aperture 56 -> 5.6 80 -> 8.0 110 -> 11.0 160 -> 16.0
TDI – 0, 2, 5, 10
Folie 71 > Image Quality > Reulke
Results Atmospheric model: urban Visibility is unknown Flight 01, visibility = 25 Flight 01, visibility = 25 Calculate calibration with ATCOR4 (Richter, DLR)
Folie 72 > Image Quality > Reulke
Image Classification Pixel-based classification Based on spectral information of pixels Minimum user input
1. Defining number of classes 2. Defining training samples 3. Examining validity of training samples 4. Selection of classification algorithm
Unsupervised classification Image -ISODATA Classification - Kmeans
Supervised classification - Parallelepiped -Minimum distance -Mahalanobis distance -Maximum likelihood
Post classification Accuracy Assessment
Object- oriented classification Based on spectral & spatial information of pixels 1. 2.
Multi resolution segmentation Knowledge-based classification of segments Folie 73 > Image Quality > Reulke
Supervised Classification Outputs derived from DMC-Data Original image
Parallelepiped
Folie 74 > Image Quality > Reulke
Maximum Likelihood
Conclusions
New developments in high-resolution imaging Technology opens new opportunities, different camera geometry and (absolute) multispectral capabilities Image quality and quality of derived products are important for the customers Digital systems make new applications possible (and necessary), because of a new kind of information, new products and data fusion with products from other data sources New products are possible from radiometric calibrated data Radiometric calibration in the lab is a challenging task
Digital camera systems have the potential to substitute film cameras and have the ability to generate new products Folie 75 > Image Quality > Reulke
