Infrared sensor modeling for human activity discrimination tasks in urban and maritime environments

Area of Science:

• Military technology
• Sensor performance modeling
• Human-computer interaction

Background:

• Infrared (IR) sensors are crucial for military observation, but current models inadequately predict performance for human activity discrimination.
• The existing NVTherm model, a standard for target acquisition, relies on static imagery and human observer calibration, which is insufficient for dynamic scenarios.

Purpose of the Study:

• To enhance the NVTherm model's capability for a priori determination of IR sensor effectiveness in human activity discrimination tasks.
• To address the challenges posed by dynamic scenes and motion cues in activity discrimination.

Main Methods:

• Conducted a series of studies to calibrate the NVTherm model specifically for human activity discrimination.
• Developed new processing methods and standards for defining target metrics in complex, dynamic imagery.
• Performed experiments to support the calculation of new calibration parameters for the model.

Main Results:

• Established new calibration parameters for the NVTherm model tailored to human activity discrimination.
• Demonstrated improved methods for representing dynamic scenes and incorporating motion cues into sensor performance analysis.
• Validated new standards for target metrics applicable to complex IR imagery.

Conclusions:

• The updated NVTherm model is significantly closer to being validated for discriminating human activity using IR sensors.
• These advancements will improve the U.S. Army's ability to predict and optimize IR sensor performance for complex surveillance tasks.