January 24th and 25th, 2007, Marriott Inn & Conference Center, College Park, MD
Sponsor: US Army Research Office
Current and future battlefield strategies envision increasingly electronic warfares equipped with large arrays of advanced imaging devices, generating vast amounts of image data. Analysis and understanding of these images, amidst noise, uncertainty and dangers of battlefields, is beyond individual soldiers and one needs automated systems that can efficiently extract relevant patterns in the observed images. Multispectral imaging devices and image analysis, termed IMage INTelligence (IMINT), are at the core of Army's doctrine for urban battlefields, with applications in surveillance, reconnaissance, combat support, and terrain mapping. Consequently, ARO has been a prime sponsor of image analysis and understanding research at US universities for several years now.
Although past research has led to isolated successes, significant advances have been limited for several reasons. We believe that an important reason has been the lack of free access to Army-related (and Army-relevant) image databases to the university researchers. Additionally, a lack of active dialog between Army labs and academia has also played a role. Recent availability of affordable imaging devices, such as laser scanners, infrared imagers, electro-optics, and hyperspectral imagers, is promising to address the data access issue.
Image analysis and computer vision community has developed several broad approaches for image analysis, for example in the areas of medical image analysis and biometrics, and an important issue is to adapt them to battlefield needs. Past research has also outlined a number of specific challenges to image exploitation, namely variability in target signatures, presence of clutter, target articulation and motion, incompleteness of learning/training data, limited network bandwidth and limited computational power.
At this workshop we propose to outline important subareas that merge academic strengths with Army needs. The specific issues we would like to discuss are:
- Can the successes in medical image analysis and biometrics be translated into Army-relevant image analysis ?
- Amongst different challenges to image analysis, what are especially relevant to Army needs ? How would Army engineers and researchers prioritize the main challenges?
- What is the potential for multiple-spectral and hyperspectral imagers in Army scenarios?
- What genre of image understanding and pattern recognition approaches -- discriminative, descriptive, or generative -- are suitable for Army applications?
- Finally, how should Army researchers help validate applications of academic research to Army problems?
- Dr. Huey Anderson , AMRDEC, AMRDEC Interests in ATR
- Dr. Greg Arnold , AFRL, Exploiting Persistence and Layered Sensing
- Prof. Rama Chellappa , University of Maryland, College Park, Pattern Recognition and Anamoly Detection Using Mulit-Sensor Data
- Prof. John Fisher , MIT, Dynamic Factorization Tests: Application to Multi-Modal Data Association
- Prof. Stuart Geman , Brown University, Parsing Images with Context/Content Sensitive Grammars
- Prof. Al Hero , University of Michigan, Prim Diffusions for Clustering, Classification, and Anamoly Detection
- Prof. Michael I. Miller/ Dimitri Bitouk , Johns Hopkins, Biometric Identification of Speaking
- Prof. Randy Moses , Ohio State University, Wide-Angle Radar Imaging
- Dr. Nasser Nasrabadi , Army Research Lab, Use of Kernels for Hyperspectral Target Recognition
- Prof. Pietro Perona , Caltech, Unsupervised Learning of Perceptual Categories
- Dr. James Ratches , NVESD, Status of ATR in Military Applications
- Prof. Guillermo Sapiro , Minnesota, On a theory for 3D Shape Recognition
- Prof. Sudeep Sarkar , University of South Florida, Gait Recognition
- Prof. Shankar Sastry , UC Berkeley, Embedded Intelligence: Beyond Sensor Webs
- Prof. Wesley Snyder , North Carolina State University, A Strategy for Shape Recognition
- Prof. Stefano Soatto UCLA, 3D Reconstruction of Complex Scenes from Video
- Prof. Anuj Srivastava Florida State University, Intrinsic Bayesian Active Contours for Shape Estimation
- Prof. Allen Tannenbaum , Georgia Tech., Geometric Methods on Visual Tracking
- Prof. Laurent Younes , Johns Hopkins., Parallel Translations on Groups of Diffeomorphisms and Shape Spaces
- Mr. Nikolay Balov, FSU
- Dr. Sankar Basu, National Science Foundation
- Dr. Alex Chan, ARL
- Dr. Liyi Dai, Army Research Office
- Dr. Daniel DeMenthon, National Science Foundation
- Dr. Jim Eilbert, Lockheed Martin
- Dr. Jim Hilger, NVESD
- Dr. Barry G. Ives, Lockheed Martin
- Mr. Shantanu Joshi, FSU
- Dr. Lance Kaplan, ARL
- Dr. Dalton Rosario, ARL
- Dr. Clare Walters, NVESD
The direction for the hotel can be found at Directions
The hotel is about 20 miles from Reagon National airport and 30 miles from BWI airport. The supershuttle (www.supershuttle.com ) will cost about $35 per person from either airport to the hotel.
