Future Imaging Systems András Jung
How can we describe an imaging system for remote sensing applications?
The Resolutions Spatial the smallest object that can be resolved by the sensor Spectral the specific wavelengths the sensor can record Radiometric the number of possible brightness values Temporal how often a sensor can detect the same object
Quality parameters for imaging systems Spatial resolution Spectral resolution Spectral misregistration (smile effect) Spatial misregistration (keystone effect) Stray light Polarization dependence Signal to noise ratio (SNR) Noise equivalent radiance (NER) and sensitivity Acquisition speed, integration time, readout mode
Who is who? System Spatial (m) Spectral (#) Radiometric (bit) Temporal (days) 1000 4 11 daily 250/500/100 36/7 12 daily 30 8 8 16 15/30/90 15 8 16 30 10 12 16 2.5-20 15 16 3-5 23.5 15 8 16 30 244 14 23 30 196 16 16 1-4 4 11 5 0.61-2.44 4 11 5 6.5 5 12 1-2 0.55 1 11 1.7-5.9 <1< 498 12-14 3-10 416 12-14 <1< 126 16 Source: A.M.Melesseet al. 2007: Remote Sensing Sensors and Applications in Environmental Resources Mapping and Modelling, Sensors, 7, 3209-3241 & www.enmap.org, www.hyspex.no, www.hyvista.com, www.specim.fi
Who is who? System Spatial (m) Spectral (#) Radiometric (bit) Temporal (days) 1000 4 11 daily 250/500/100 36/7 12 daily 30 8 8 16 15/30/90 15 8 16 30 10 12 16 2.5-20 15 16 3-5 23.5 15 8 16 30 244 14 23 30 196 16 16 1-4 4 11 5 0.61-2.44 4 11 5 6.5 5 12 1-2 0.55 1 11 1.7-5.9 <1< 498 12-14 3-10 416 12-14 <1< 126 16 Source: A.M.Melesseet al. 2007: Remote Sensing Sensors and Applications in Environmental Resources Mapping and Modelling, Sensors, 7, 3209-3241 & www.enmap.org, www.hyspex.no, www.hyvista.com, www.specim.fi
Who is who? System Spatial (m) Spectral (#) Radiometric (bit) Temporal (days) AVHRR 1000 4 11 daily MODIS 250/500/1000 36/7 12 daily Landsat-7 ETM+ 30 8 8 16 ASTER 15/30/90 15 8 16 ALI 30 10 12 16 SPOT 1-4 2.5-20 15 16 3-5 IRS-1C 23.5 15 8 16 EnMap 30 244 14 23 Hyperion 30 196 16 16 IKONOS 1-4 4 11 5 QUICKBIRD 0.61-2.44 4 11 5 RAPID EYE 6.5 5 12 1-2 WORLDVIEW 0.55 1 11 1.7-5.9 AISA Dual <1< 498 12-14 HySpex 1-3 416 12-14 HyMap <1< 126 16 Source: A.M.Melesseet al. 2007: Remote Sensing Sensors and Applications in Environmental Resources Mapping and Modelling, Sensors, 7, 3209-3241 & www.enmap.org, www.hyspex.no, www.hyvista.com, www.specim.fi
Typical remote sensing data acquisition Quasi-Point Image Image 400 km 4 km 0.4 m continuously consistently over the entire globe, Foto: C. Sandow, 2009 Foto: www.fun-with-pictures.com explorativ expeditiv individual/human local... ad-hoc complex semi-predictive sub-regional...
Typical remote sensing data acquisition Quasi-Point Image Image 400 km 4 km 0.4 m continuously consistently over the entire globe ad-hoc complex semi-predictive sub-regional... Foto: C. Sandow, 2009 SPATIAL SPECTRAL RADIOMETRIC TEMPORAL Scanner explorativ expeditiv individual/human Field local spectrometer Scanner Foto: www.fun-with-pictures.com SPATIAL SPECTRAL RADIOMETRIC TEMPORAL (wh) SPATIAL SPECTRAL RADIOMETRIC TEMPORAL (wh)
Changes in resolution approach Customization Usability resolution Weight resolution (extra light) On-Demand resolution (temporal) Financial resolution Flexibility resolution
Present changes in data acquisition Quasi-Point Image Image 400 km 4 km 0.4 m Field spectrometer continuously consistently over the entire globe Foto: C. Sandow, 2009 SPATIAL SPECTRAL RADIOMETRIC TEMPORAL Scanner Scanner Foto: www.fun-with-pictures.com SPATIAL SPECTRAL RADIOMETRIC TEMPORAL (wh) SPATIAL SPECTRAL RADIOMETRIC TEMPORAL (wh)
Present changes in data acquisition Quasi-Point Image Image 400 km 4 km 0.4 m Field spectrometer SPATIAL SPECTRAL RADIOMETRIC TEMPORAL Mini scanner Foto: C. Sandow, 2009 Field Scanner Scanner Field Scanner Scanner Foto: www.fun-with-pictures.com SPATIAL SPECTRAL RADIOMETRIC TEMPORAL SPATIAL SPECTRAL RADIOMETRIC TEMPORAL
UAV with hyperspectral Scanner Hyperspectral mini scanners (< 1 kg) 400-1000 nm realistic GPS is applicable Nice tool for high resolution and flexible data collection Micro-Hyperspec/Headwall Pablo J. Zarco Tejada quantalab.ias.csic.es PIKA Airborne Resonon www.resonon.com
Field hyperspectral scanner (terrestrial stationer) Vertical imaging Horizontal imaging Scanning by synchronized motor movement Full range mostly with two cameras (VNIR+SWIR) HySpex Modelle 400-2500 nm www.hyspex.no Specim Modelle 400-2500 nm www.specim.fi Headwall Modelle 400-2500 nm www.headwallphotonics.com
New demands Flexibility Independency Easy mobility Rapid data access More data (4R)
Future remote sensing scenarios Quasi-Point Image Image 400 km 4 km 0.4 m SPATIAL SPECTRAL RADIOMETRIC TEMPORAL Mini scanner Field Foto: C. Sandow, 2009 spectrometer Field Scanner Scanner Scanner Foto: www.fun-with-pictures.com SPATIAL SPECTRAL RADIOMETRIC TEMPORAL SPATIAL SPECTRAL RADIOMETRIC TEMPORAL
Future remote sensing scenarios Quasi-Point Image Image 400 km 4 km 0.4 m SPATIAL SPECTRAL RADIOMETRIC TEMPORAL Mini scanner Field Foto: C. Sandow, 2009 spectrometer Field Scanner Scanner Spectral Video Camera Scanner Foto: www.fun-with-pictures.com SPATIAL SPECTRAL RADIOMETRIC TEMPORAL SPATIAL SPECTRAL RADIOMETRIC TEMPORAL
Future remote sensing scenarios Image Image Image 400 km 4 km 0.4 m SPATIAL SPECTRAL RADIOMETRIC TEMPORAL Mini scanner Field Foto: C. Sandow, 2009 Spectral spectrometer Field Scanner Video Camera Scanner Scanner Foto: www.fun-with-pictures.com SPATIAL SPECTRAL RADIOMETRIC TEMPORAL SPATIAL SPECTRAL RADIOMETRIC TEMPORAL
Future remote sensing scenarios Image Image Image 400 km 4 km 0.4 m SPATIAL SPECTRAL RADIOMETRIC TEMPORAL Mini scanner Field Foto: C. Sandow, 2009 Spectral spectrometer Field Scanner Video Camera Scanner Scanner Foto: www.fun-with-pictures.com SPATIAL SPECTRAL RADIOMETRIC TEMPORAL SPATIAL SPECTRAL RADIOMETRIC TEMPORAL
Future remote sensing scenarios Image Image Image 400 km 4 km 0.4 m SPATIAL SPECTRAL RADIOMETRIC TEMPORAL Mini scanner Field Foto: C. Sandow, 2009 Spectral spectrometer Field Scanner Video Camera Scanner Scanner Autonomous Spectral Camera Foto: www.fun-with-pictures.com SPATIAL SPECTRAL RADIOMETRIC TEMPORAL SPATIAL SPECTRAL RADIOMETRIC TEMPORAL
Imaging spectrometers Hyperspectraler Scanner - either the Scanner (Airborne, field) - or the sample moves (Lab). - successive framing (delay in time) Hyperspectrale Frame -Camera - one shot one spectral datacube - behaves like a digital camera - simultaneous imaging (real-time) - hyperspectral video frames
Hyperspectral Movie Cubert GmbH, Ulm, Germany 22
Spectral video camera Cubert GmbH, Ulm, Germany Full frame imaging Integration time pro frame: 10µs-100ms Spectral range: 400-1100nm Channels: 140 Spectral resolution: 4 nm 25 images/sec 23
Flexibility in data acquisition
Image data from a non-scanning camera Cubert GmbH, Ulm, Germany
Camera above the forest (40 m)
Camera above the forest
Hyperspectral images above moving forest canopy
New demands Flexibility Independency Easy mobility Rapid data access More data (4R)
Future demands Flexibility I can measure any time Independency Short decision path Easy mobility Take and measure Rapid data access Fast post-processing More data (4R) HQ data (repeat-stat)
New demands Flexibility I can measure any time Independency Short decision path Easy mobility Take and measure Rapid data access Fast post-processing More data (4R) HQ data (repeat-stat)
New demands Flexibility I can measure any time Independency Short decision path Easy mobility Take and measure Rapid data access Fast post-processing More data (4R) HQ data (repeat-stat)
Easy mobility Take and measure Rapid data access Fast post-processing More data (4R) HQ data (repeat-statis.)
Easy mobility Take and measure Rapid data access Fast post-processing - real time - 24/48 More data (4R) HQ data (repeat-statis.) - more retakes - more representative data - more control on data production
Focus on high resolution (HR)data Spatial Spectral Radiometric Temporal Focus on local issues
Setup matrix in the field during data acquisition Sensor fixed Sensor moving Object fixed present present Object moving future future
data acquisition
What is about data evaluation?
Time between data acquisition and data evaluation is too long!
Time between data acquisition and data evaluation is too long! To reach higher temporal resolution post-processing should be faster
Data acquisition Data post-processing Data evaluation Action plan Science Industry Delay Real time Data acquisition Data post-processing Data evaluation Action plan
Higher temporal resolution = better classification The higher the variance the higher the accuracy of the classification for the same object. Training set samples must be selected to span all sources of sample variance J.Burger, 2012.
The multi-angular hyperspectral observation capability may be one of next important steps in the field of hyperspectral remote sensing Prasad S. Thenkabail Research Geographer, U.S. Geological Survey (USGS) USGS EROS Data Center, Sioux Falls, SD, USA. August 16-18, 2011
Two tricks to increase resolution What can scanner users do to increase spatial resolution? What can frame camera users do to increase spatial resolution?
Higher spatial resolution by shifting Super-resolution concept
Higher spatial resolution by shifting for line scanners Super-resolution concept Original Resolution Horizontally shifted Enhanced resolution Vertically shifted Overlapping S. Pi queras et al., Relevant aspects of quantification a nd s ample heterogeneity i n hyperspectral image resolution, Chemometr. Intell. La b. Syst. (2012), doi:10.1016/j.chemolab.2011.12.004
Higher spatial resolution by angular micro-movements for frame cameras Chance for shaking hands
Higher spatial resolution by angular micro-movements for frame cameras Chance for shaking hands
Higher spatial resolution by angular micro-movements for frame cameras Chance for shaking hands Jung et a l. Unpublished experiment, 2012
Higher spatial resolution by angular micro-movements for frame cameras Chance for shaking hands Maybe hyperspectral stereoscopy? Jung et a l. Unpublished experiment, 2012
Conclusions High demand for hyperspectral image data higher temporal resolution autonomous imaging systems mobile imaging systems New challenges in HR spectral-spatial data processing in post-processing in hardware development
In the next years field and low-altitudeairborne spectroscopy will be extended by intelligent and mobile imaging spectrometers. Intelligent: post-processing & on-demand (evaluation real time ) Mobile: hand-held & UAV
Vielen Dank!