Outline • Introduction • Method • Experiment results • Summary & future work
Introduction • Definition of image simulation • generates synthetic images based on the analysis and understanding of imaging acquisition • Application • Evaluation of system specifications • Test of processing facilities • Test-bench for future algorithm development • Cost-versus-quality trade-offs
Introduction • Simulation tools • DIRSIG (The Digital Imaging and Remote Sensing Image Generation Model) • Spectral range: • 0.3 - 20 μm region • Types of imagery: • multi- and hyper-spectral passive systems, • polarimetric imagery, • radiative transfer in littoral waters, • and active LIDAR systems source：http://dirsig.blogspot.com/2011/02/ scene-building-with-blender.html
Introduction • Simulation tools • EeTes (EnMAP end-to-end Simulation) • Spectral range: VNIR & SWIR • PICASSO (Parameterized Image Chain Analysis & Simulation SOftware) • Spectral Range: visible to near-infrared(VISNIR) & TIR • Summary • Image simulation in mid-infrared regions is rarely discussed, especially the absorption bands.
Introduction • Applications of mid-infrared regions (3-5 μm) • Sensitive to high temperature objects(fire, active volcanoes etc.) • Mid-infrared absorption bands • Fundamental research on these two special band to make preparation for mid-infrared simulation.
Introduction • Image simulation chain • Surface scene simulation is basis for other two processes. • Solar radiation is absorbed and less will reach the ground and be reflected. • Question • whether the reflected part of surface radiance can be neglected ? • what factors affect the surface radiance composition ? • study bands: 2.7 &4.3 μm Surface scene simulation Sensor hardware simulation Atmospheric simulation
Method • ground radiance simualtion • atmospheric transfer model MODTRAN (MODerate resolution atmospheric TRANsmission) • MODTRAN can simulate the absorption effects of atmospheric molecules to the solar radiation. • Simulation outcome • Total surface radiance (represented by Rt) • Reflected radiance ( represented by Rr) • Emitted radiance (represented by Re) • Rt = Rr +Re • Evaluation index: Rr / Re
Method • Input parameters
Method • surface features • assume all features are lambert in simulation.
Method • Spectral reflectance (from JHU spectral library) The reflectance of soil is relatively higher than vegetation and water
Experiment results • Rr/Re near 2.7μm in summer and winter Temperature& reflectance have impacts on surface radiance compositon in mid-infrared absorption bands
Experiment results • Rr/Re near 4.3μm in summer and winter The result is similar to that in 2.7 μm regions
Experiment results • Ratio of Rr to Re of the band • assumption: square-wave spectral response function • Response equals 1 within the band • Response equals 0 outside the band
Experiment results • Rr_b/Re_b in 2.7 & 4.3 band The result in bands is consistent with that in wavelengths.
Summary & future work • Summary • Temperature and reflectance of surface features both contribute to the surface radiance composition. • Whether the reflected radiance can be neglected in surface scene simulation relates to the expected accuracy of simulation. For example, if a 10 percent of error is allowed, the reflection of soils, water and vegetation can all be neglected.
Summary & future work • Further work • More factors need to be involved: water vapor contents, BRDF, etc. • Reflectance data of surface features should be expanded. • In-situ validation: field measurements of reflected and emitted radiance. • Simulation is working with the sensor. Since the proportion changes with the wavelength, for specific sensor, the surface composition analysis also depends on the bandwidth.