The Center for Geospatial Intelligence (CGI) was established only recently (February, 2004). CGI researchers have since submitted a number of proposals to NGA, NSF, NASA, and the Army. This page will be used to provide summaries of all ongoing projects as they are funded.
Project Title: Content-Based Information Mining and Visualization for Exploitation of Multi-Modal Databases
Principal Investigator: Chi-Ren Shyu
Co-Investigators: Curt Davis and K. Palaniappan
Funding Source: National Geospatial-Intelligence Agency
Period of Performance: 8/04 - 7/07 (excluding 2 option add on years)
Research Summary: The current size of intelligence datasets far exceeds what manual intelligence analysis can handle to correlate findings among heterogeneous or multi-modal databases. This will only get worse moving forward as future imagery collection systems (FIA, WorldView, SBR, etc.) and other intelligence methods come online to deal with the post 9-11 requirements of more ubiquitous and persistent surveillance.
Searching for relevant knowledge across multi-modal intelligence databases requires an extensive knowledge of the semantic meaning of images, a keen eye for its visual patterns, efficient strategies for collecting and analyzing geospatial data, and a good understanding of related tradecrafts (signal intelligence, measurement intelligence, etc.). For example, "What are possible annotations for this building complex given the construction methods and materials used and the trace measurement of these chemical signatures?" Answering this question requires an extensive knowledge of the perceptual categories used for image annotation and the ability to draw information from one tradecraft modality to others.
In this project we will leverage our multi-disciplinary research experience in high-resolution earth image feature extraction, content-based image retrieval, semantic modeling and knowledge exchange, and visualization.
There are four major objectives for this proposed project:
(1) Develop, test, and integrate efficient and relevant image feature extraction algorithms appropriate for multi-modal (optical, radar, etc.) medium and high-resolution earth images:
(2) Build a hybrid query system for image-based and text-based information retrieval.
(3) Build a semantic and knowledge sharing information mining hub for analysts.
(4) Develop a distributed visualization system for displaying and hierarchically browsing multi-modal database retrieval results and incorporate relevance feedback for query focus.
Project Title: Discrimination Mode Processing for Handheld Landmine Detection
Principal Investigator: Dominic Ho
Funding Source: US Army Night Vision Lab, RDECOM CERDEC NVESD
Period of Performance: 4/05 - 6/07
Research Summary: The handheld mine detector developed by the US Army with algorithms developed in our previous research has proven to be a success, and it is currently used in Afghanistan and other countries. The focus of the handheld mine detection effort has recently been shifted to humanitarian applications to clear up mine fields left after wars. The objective of this project is to develop automatic signal processing algorithms for ground penetrating radar to improve the performance of the existing handheld mine detection system, especially for humanitarian demining.
Landmine detection using handheld GPR units is a challenging task. First of all, the detector is expected to detect both anti-tank and anti-personnel landmines which present a wide range of size and metal content. Secondly, the detector is expected to operate in different environments and terrain regions and therefore the background can be highly non-stationary. Signal processing is therefore a crucial component to improve landmine detection.
Handheld GPR mine detection has two operating modes: search mode and discrimination mode. The search mode determines an alarm location that potentially contains a landmine target, and the discrimination mode examines the alarm location to confirm if the alarm location contains a landmine.
This research concentrates on the development of automatic discrimination mode algorithms to differentiate between landmine target and clutter object. The algorithm under development will use the consistency of the detection information, and frequency domain features to improve the discrimination. Blind test results in the previous phase of research have shown that the frequency domain features can reduce the probability of false alarms by a factor of 2 at 90% probability of detection, compared to the results from the best human expert operator.
The work will be done in collaboration with the Univ. of Florida where the fusion of ground penetrating features and metal detector features will be done to increase the overall detection performance.
Project Title: Spectral Analysis and Processing for Landmine Detection with Downward-Looking and Forward-Looking Ground Penetrating Radar
Period of Performance: 4/04 - 3/07
Research Summary: The US Army is currently developing a vehicle based landmine detection system that uses ground penetrating sensors for detecting anti-tank landmines. The CGI is assisting the US Army by developing algorithms to improve the detection of weak scattering plastic landmines, and to enhance the discrimination between the landmine and clutter objects.
The PI along with Dr. Paul Gader (Univ. of Florida) recently proposed the use of frequency domain features extracted from the Energy Density Spectrum (EDS) to improve landmine detection. The motivation for this approach comes from the fact that landmine targets and clutter objects often have different shapes and/or composition, yielding different EDS that may be exploited for their discrimination. Preliminary results based on the processing of calibration lane data measurements indicate that the EDS approach can discriminate mines with some clutter objects such as pieces of wood, rocks, and metal debris.
In this project, we will examine different methods to generate the EDS, investigate features to be extracted from the EDS that can provide the best results, and then fusing these with other detection algorithm outputs. The focus will on off-road performance where a large amount of clutter objects will be encountered. This work will be performed in collaboration with the Univ. of Florida by developing fusion techniques that combine the spectral features and geometric features to further improve detection accuracy.
The project will also involve the support to BAE, a contractor to build the hardware vehicle system, in transferring the developed algorithm into real-time processing for field evaluations. We will also participate in the field tests for further algorithm development and improvement efforts.
Project Title: Enterprise Architecture Development in NASA Earth Sciences Application Development Activities
Principal Investigator: Tim Haithcoat
Funding Source: National Aeronautics and Space Agency
Period of Performance: 5/04 - 4/05
Research Summary: The National States Geographic Information Council (NSGIC) provides a unified State voice on geographic information and technology issues, advocates state interests, and supports its membership in their individual initiatives. The Council actively promotes prudent geographic information integration and systems development. NSGIC reviews legislative and agency actions, promotes positive legislative actions, and provides advice to public and private decision-makers. NSGIC members are actively involved in the application of geospatial technologies in their member States. They are often at the forefront of GIS and information technology innovation.
NSGIC works closely with similar purpose organizations, including the Federal Geographic Data Committee (FGDC), the Local Leaders in GIS (LLGIS) and the four local government organizations that compose it, the University Consortium for Geographic Information Science (UCGIS), the Western Governor's Association (WGA), and the National Association of State Chief Information Officers (NASCIO). The NSGIC plays a key role in the dissemination of information and activities related to the development and integration of NASA's National Applications across the Nation.
In this project, NSGIC will provide support for the development of geospatial information architecture components as well as track the results from the missions and application development of NASA's Application Division. The objective of these activities is the implementation of predictions and measures resulting from the modeling and analysis of the EOS data stream. NSGIC provides a key dissemination mechanism to achieve this broad application. NSGIC members coordinate geospatial activities in their own state and are involved in day-to-day decision-making on programs identified by Earth Sciences Enterprise as national applications including, emergency management, energy forecasting, agriculture, transportation, infrastructure, smart growth, public health, invasive species, and air/water management and conservation. These tasks will compliment and support NASA programs and projects with partnering federal agencies by helping to initiate, document, and disseminate this information at the state and local government levels.
The primary task will focus on the development of an enterprise architecture. This will aid the NSGIC and NASCIO enterprise architecture initiative in the development and documentation of the components that result from the ongoing development of NASA's national applications and the continuing data and information streams resulting from EOS. It will also enable Code YO to fully participate in this architecture effort and respond to the development of the Federal Enterprise Architecture (FEA) efforts across the Federal Government. By placing NASA's information within the adaptive enterprise architecture, it will be positioned at enterprise levels and be more discoverable to potential users. This will increase the subsequent utility of these resources at state and local levels of government.