Statistical detection of resolved targets in background clutter using optical/infrared imagery

Area of Science:

• Optical engineering
• Signal processing
• Image analysis

Background:

• Traditional automatic target detection (ATD) methods assumed unresolved targets.
• Advanced focal plane technology now enables highly resolved target imaging.
• Resolved targets obscure background clutter, posing a challenge for detection.

Purpose of the Study:

• To develop a general solution for detecting resolved targets in background clutter.
• To derive an optimum detection algorithm for such scenarios.
• To analyze the key factors influencing detection performance.

Main Methods:

• Derivation of an optimum detection algorithm based on a test statistic and threshold.
• Analysis of detection performance dependence on apparent contrast and signal-to-noise ratio (SNR).
• Investigation of the impact of the background clutter to common system noise ratio.
• Validation through computer simulations of theoretical detection and false alarm probabilities.

Main Results:

• Detection performance is primarily governed by apparent contrast, not SNR.
• Performance is highly sensitive to the background clutter to common system noise ratio.
• Targets can be detected even when apparent contrast approaches zero, if background clutter exceeds system noise.

Conclusions:

• The developed algorithm provides a general solution for resolved target detection in cluttered backgrounds.
• Apparent contrast and the clutter-to-noise ratio are critical parameters for electro-optical/infrared (EO/IR) sensor design.
• Findings are applicable to engineers and scientists designing EO/IR sensors for target detection in complex scenes.