Automated methods for Scene Understanding/Situation Awareness

8:15 – 10:15 Scene Understanding And Situation Awareness

This Master Class is designed to help participants understand the latest developments in automated and semi-automated methods for Scene Understanding and Situation Awareness for military, intelligence, and civil applications. The intended audience includes users that are tasked with developing, using or evaluating techniques and systems for scene and situation understanding and for detecting, assessing, tracking, and prosecuting threat activity. This is also for military or industry representatives involved in policy-making who have a need to learn the basic concepts, issues, and realistic capabilities of tools and methods for image analysis, situation and threat assessment.

What will be covered :

• Template- and model-based target recognition concepts and techniques
• Methods for adaptive evidence accrual, context display, and exploration
• Algorithms for scene/situation hypothesis generation/evaluation/selection
• Performance metrics for target recognition, scene and situation characterization

How you will benefit:

• Gain an understanding of concepts, current techniques, algorithms, and performance metrics
• Learn the application of performance metrics for characterizing objects’ imagery by combining object literal and induced features together with contextual information, both within the image and in collateral information (to include other sensed or descriptive data)

Session leader:

Dr. Alan Steinberg
Georgia Tech Research Institute

Understanding underlying information transfer processes during image fusion

10:30 – 12:30 In-Depth Objective Evaluation Of Image Fusion

Image fusion has already gained acceptance as a useful tool in night and improved vision applications where multiple sensors are available with a wealth of different algorithms proposed for the task. But how can we really know which algorithms to choose for each application and what to expect from them when they encounter real data. This workshop will focus on in-depth fusion evaluation that would allow us to make informed decisions based on objective qualification as well as quantification of algorithm performance based on the desired visual information transfer sought through fusion.

What will be covered:

• Theory of visual information transfer during the image fusion process and its breakdown into tractable categories.
• Measures of various information transfer processes taking place during fusion in order to qualify its performance
• Creation of customized metrics to evaluate specifically desired outcome of image fusion

How you will benefit:

• Gain an understanding of the underlying information transfer processes taking place during image fusion
• Understand how more informed qualifications of fusion performance can be achieved
• Learn how to construct objective metrics for specific image fusion goals

Session leader:

Dr. Vladmir Petrovic
Research Associate, Imaging Science
University of Manchester (UK)

1:30pm - 3:30pmEnd-User Perspective

Image Fusion from the Users Perspective

This in-depth focus session will provide a true end-user perspective. Learn about the distinct details on current imagers currently used in the field and problems and concerns with them. It will also give a users perspective on image fusion from ground , air, and mobility platforms.

Session Leader:

Charles R. Greer
Crane Division, Naval Surface Warfare Center (NSWC Crane), Electro-Optic Technology Division
Code JXQL

Augmenting traditional sensor systems

3:45 – 5:45 Human-Centered Multi-INT Fusion

The traditional role of data fusion has involved the use of physical sensors to observe physical targets, in effect trying to characterize the physical landscape for situational awareness. In this workshop we will explore how rapid changes in non-traditional warfare have changed the focus of fusion systems to try to assess the human landscape as well as the physical landscape.

What will be covered?

• Humans acting as new sources of information
• Human analysts supporting the analysis process (advanced visualization and sonification interfaces)
• Humans acting as an ad hoc community of analysts (“crowd-sourcing” of the analysis process)

How you will benefit?

• Understand the dynamics of ad hoc community of observers
• Ascertain how providing inputs augment traditional sensor systems

Session Leader:

COL. (ret) James Graham
Professor of Practice, College of Information Sciences and Technology, Deputy Director, NC2IF
Penn State University