Reconstruct high fidelity image and/or 3D geometry and/or materials using the angle, frequency, time, and polarization diversity afforded by existing and postulated sensors or sensor combinations. The reconstruction can take place over multiple time intervals and sensor trajectories. Of particular interest are reconstruction strategies that include error bounds or posteriors on the parameters being estimated rather than point estimates. Further, it is likely that the diversity and sampling density desired will not be available with practical sensing strategies so reconstruction quality as a function of the completeness of the sensor information is also of interest. In addition to incomplete sensor information challenges each sensor will have its own technical challenges. Sensors of interest include LIDAR, video, WAMI, SAR, and hyperspectral.
Challenges for SAR 3D reconstruction include specular scattering (as opposed to diffuse scattering for EO imagery), 3D layover effects, multipath, and interactions with clutter. Advantages for SAR include that it is active and coherent. These properties facilitate interferometric processing, full polarization matrix reconstruction due to multiple coherent apertures, waveform diversity, and of course, since SAR is active and has a longer wavelength, the variability due to sun angle and atmospheric effects is not an issue as it is with EO passive sensing.
| Topic | Data on SDMS | SPIE Paper (PDF) |
|---|---|---|
| Data Dome: Full k-space sampling for high frequency radar research | SDMS Data | SPIE Paper |
| Challenge Problem for 2D/3D Imaging of Targets from Volumetric Data Set in an Urban Environment | SDMS Data | SPIE Paper |
| Civilian Vehicle Radar Data Domes | SDMS Data | SPIE Paper |
| A Challenge Problem for SAR-based GMTI in Urban Environments | SDMS Data | SPIE paper |
Estimate the posterior probabilities of the presence of an object given an observation and provide means to evaluate performance of the decision outcome as well as the relative or absolute (geo-location referenced) position of the object. Performance should explicitly depend on object and environment characteristics and on time, angle, frequency, and polarization diversity afforded by existing and postulated sensors or sensor combinations. Use of time diversity (change detection) is of particular importance since most objects of interest are dynamic in some time scale. Sensors of interest include RADAR (including MTI), SAR, LIDAR, video, WAMI, and hyperspectral.
Environmental challenges for Detection from SAR include scene complexity and obscurations. Sensor-based challenges include limited resolution, image-processing artifacts, small or scintillating object signature, and small object-to-background signal ratios.
| Topic | Data on SDMS | SPIE Paper (PDF) |
|---|---|---|
| Standard SAR ATR evaluation experiments using the MSTAR public release data set | SDMS Data | SPIE Paper |
| A Challenge Problem for SAR Change Detection and Data Compression | SDMS Data | SPIE Paper |
| A Challenge Problem for SAR-based GMTI in Urban Environments | SDMS Data | SPIE Paper |
| A Challenge Problem for Detection of Target in Foliage | SDMS Data | SPIE Paper |
Estimate the relative or absolute (geo-location referenced) spatiotemporal evolution of an object using temporal diversity afforded by existing and postulated sensors or sensor combinations. ''Tracks'' should be contiguous but may present as labeled disjoint intervals such as when observations are denied. Of particular interest are tracking strategies that include error bounds or posteriors rather than singular sequential position estimates, and the influences of environment characteristics, traffic densities, object dynamics (or lack thereof), and revisit rate on predicted performance, and ratios of process-to-real time requirements. Sensors of interest include MTI RADAR, SAR, video, WAMI, and hyperspectral.
Environmental challenges for SAR tracking include obscurations, shadows, traffic intersections, the presence of similar objects in proximity, and high traffic densities. Sensor-based challenges include low revisit rate, limited resolution, image-processing artifacts, small or scintillating object signature, small object-to-background signal ratios.
| Topic | Data on SDMS | SPIE Paper (PDF) |
|---|---|---|
| A Challenge Problem for SAR-based GMTI in Urban Environments | SDMS Data | SPIE Paper |
Recognition is fundamentally a special case of reconstruction via a mapping of the estimates (posteriors) onto a discrete decision space defined by exemplars or models representing the objects of interest. Thus, analogously, Recognition is the process of estimating posterior probabilities of an image and/or 3D geometry and/or materials using the angle, frequency, time, and polarization diversity afforded by existing and postulated sensors or sensor combinations with the means to evaluate the performance of a recognition decision outcome. Of particular interest are recognition strategies that explicitly exhibit error bounds or posteriors in the decision space rather than decision outcomes. Performance should explicitly depend on object-library and environment characteristics. Sensors of interest include LIDAR, video, WAMI, SAR, and hyperspectral.
As with reconstruction, challenges for SAR recognition include specular scattering (as opposed to diffuse scattering for EO imagery), 3D layover effects, multipath, and interactions with clutter.
| Topic | Data on SDMS | SPIE Paper (PDF) |
|---|---|---|
| Standard SAR ATR evaluation experiments using the MSTAR public release data set | SDMS Data | SPIE Paper |
| Challenge Problem for 2D/3D Imaging of Targets from Volumetric Data Set in an Urban Environment | SDMS Data | SPIE Paper |
| Civilian Vehicle Radar Data Domes | SDMS Data | SPIE Paper |
| Wide Angle SAR Data for Target Discrimination Research | SDMS Data | SPIE Paper |
This data product is available for immediate DOWNLOAD.