Monthly Technical Report December 1998 SRI Project Number: ECU-1515 Contract Number: NMA202-97-C-1004 DARPA Order Number: E645 Prepared by: Aaron J. Heller, Sr. Computer Scientist Prepared for: Michele Motsko, Physical Scientist National Imagery and Mapping Agency Reston, VA SUMMARY OF ACTIVITIES: 1. Integration of Spur detection and following into FICS 2. Further refinement of 2-D linear delineation system 3. Working with NIMA MUSE and DTOP 4. Integration of Vexcel REX into BOS/RCDE system. 5. Evaluation of full-frame NTC result. DETAILED DESCRIPTION: 1. Integration of Spur detection and following into FICS (Connolly) 2. Further refinement of 2-D linear delineation system (Fischler,Wolf) The algorithm now produces a model that is 96% correct and 89% complete (with respect to our baseline reference model). Using a model integration technique discussed below, these performance numbers can be traded to provide 96% complete and 81% correct (using just the original Ft. Benning Ortho-photo). 2) Since can be adjusted to provide an almost error-free result, we can integrate the results obtained from multiple views, resolutions, and sensors by simply OR-ing the models and eliminating redundancies. I have been working on an algorithm to perform this merging operation. It is almost completed, and I hope that when I add more imagery to the mix (e.g., the new Ft Benning coverage), the completeness score will be close to 100% without any additional significant loss in correctness. In time for demo, I expect to be able to provide an example of street modeling for some small area in Ft, Hood, and an example of the use of a prior road model to detect changes in the scene being modeled. 3. Working with NIMA MUSE and DTOP (Burns,Heller) Learned how road junctions should be represented in VPF. Read through the document called "Military Standard Vector Product Format" and found that: 1. Roads should be represented as line features 2. Line features are associated with edge-type primitives (p 57) 3. Networks of line features should be represented using the "winged-edged" method and cross-index fields in the edge data structure provide for this method. (Appendix B and table 7 p 32) Did an initial test to determine if DTOP was using this representation and putting real junction data in the fields (not just arbitrary values). I looked at the road features on the Santiago CD returned by the query "vpfimpcl -feat G:,dtopn,t100cu01,trans,Transportation,roadl.lft,*,,,,,,,,,, -ext -90 19 -10 21". I counted the number of edges associated with these roads that indexed other road edges in the query set via their winged-edge cross-index fields. I also counted the number of edges that did not just cross-index themselves (i.e. a road just joined to itself) -- which would be a syntactically correct use of the fields, but not useful. Results: The total number of "road" edges in the query set: 1830 Of those, the number whose "leftedge" and "rightedge" fields pointed to edges in the query set: 1668 Of those, the number whose "leftedge" or "rightedge" fields that did not just point to edge itself: 1662 In sum, it seems like DTOP and the Santiago CD are exploiting the VPF data structures for representing road networks and not filling them up with arbitrary values. A more thurough check could be done, without too much work, to see if the topology is correct by traversing the chains of cross-indicies and see if they cycle about the nodes associated with the junctions. 4. Integration of Vexcel REX into BOS/RCDE system. (Heller) 5. Evaluation of full-frame NTC result. (Wolf,Heller) --------------------------------------------------------------------------- Automatic Population of Geospatial Databases Monthly Report to SRI for December Bob Wilson Vexcel Corporation 6 January 1999 1. MAJOR TECHNICAL ACCOMPLISHMENTS 1.1 Extraction of Buildings from Dense DEMs Chris Lesher continued to work on software to refine and "clean up" building edges extracted from dense DEMs. For test data he has been using DEMs of downtown Denver and Ft. Hood. The software now supports computing straight edges, squaring off corners, and computing local preferred orientations from road/street networks and averaging building edges. I presented some very preliminary results at the September '98 IPR. Chris Lesher is just about done with this development. He has written software to generate wireframe models in VRML. We have not yet done any formal evaluation, but the results look very nice and clean. We hope to show off some of the results at the demo this month. 1.2 Sent REX to SRI Vexcel sent stand-alone (not tied to PCI) versions of REX for Sun and SGI platforms to SRI via ftp. 1.3 Received May '98 Images of Ft. Benning from SRI The images, which had been scanned and processed by Max Roos at TEC, were sent to us on 8 mm tape. We loaded them and began evaluating their suitability for dense DEM extraction. 2. ACCOMPLISHMENTS VIS-A-VIS STATEMENT OF WORK 2.1 Refine the BOS architecture 2.2 Survey automated model extraction techniques 2.3 Develop feature extraction and consistency enforcement algorithms 1.1 and 1.2 2.4 Refine the design of and implement the dynamic database 2.5 Produce, maintain, and distribute data sets and ground truth 1.3 2.6 Develop evaluation metrics and perform evaluations 2.7 Interface to FRE contractors 2.8 Develop and perform demonstrations 2.9 Transfer technology 2.10 APGD program management weekly conference calls, monthly report, and phone conversations. --------------------------------------------------------------------------- GDE Systems APGD Monthly Report December 10, 1998 Technical Section Summary The current increment of program funding has been exhausted. Accordingly most work has been halted. A small amount of corporate risk funding has been approved, in anticipation of the final program funding increment, to prepare for the January 27 final demonstration. Detailed Work Description (by SOW item): 1. Architecture Refinement No activity 2. Algorithm Survey No activity. 3. Algorithm Development No activity. 4. Multi-Sensor Registration No activity 5. Persistent Store. No activity 6. Dataset Production & Distribution: Work on the Fort Hood feature extraction has been halted. The latest results will be transmitted to the prime. 7. Evaluations: No activity 8. Interface to FRE Contractors: No activity. 9. Demonstration Scenarios: Work is underway, on company risk funding, to prepare a building extraction demo for January 27. 10. Technology Transfer No activity 11. Option Years No activity 12. Program Management We keep in close contact with team members via weekly conference calls and e-mail communications as appropriate.