Aerial reconnaissance film screening using optical matched-filter image-correlator technology

Area of Science:

• Computer Vision
• Image Processing
• Artificial Intelligence

Background:

• High costs of human photo interpreters in aerial reconnaissance drive demand for automated screening.
• Optical matched-filter image-correlator (OMFIC) systems offer potential for automated image analysis.

Purpose of the Study:

• To investigate the performance of an OMFIC system for automated screening in aerial reconnaissance.
• To project the effectiveness of OMFIC systems with expanded filter memory banks.

Main Methods:

• Detailed investigation of an OMFIC system architecture.
• Simulation and projection of system performance using holographic lens-matched filter memory banks.
• Evaluation of detection probability and false alarm rates under varying image conditions.

Main Results:

• The OMFIC system demonstrated a high probability of detection (98%) with a low false alarm rate (1.4%).
• Performance was maintained despite significant image scale variations (+/-9.5%).
• A memory bank of 23 orientational views of an M-60 tank was sufficient for high accuracy.

Conclusions:

• OMFIC systems are highly capable for automated aerial reconnaissance screening.
• The system achieves excellent detection rates with minimal false positives, even with scale variations.
• Expanded filter banks enhance the OMFIC system's efficiency and accuracy for complex scenes.