Detection and Confirmation of Multiple Human Targets Using Pixel-Wise Code Aperture Measurements.

Area of Science:

• Computer Vision
• Signal Processing
• Machine Learning

Background:

• Conventional target detection requires full video reconstruction, which is time-consuming and can lead to information loss.
• Compressive sensing offers potential for efficient video data handling by reducing bandwidth and storage requirements.

Purpose of the Study:

• To apply a novel compressive sensing approach for direct human target detection and classification within the measurement domain.
• To evaluate the feasibility of detecting human targets using raw videos from a pixel-wise code exposure (PCE) camera.

Main Methods:

• Utilized a pixel-wise code exposure (PCE) camera to capture condensed video frames.
• Employed a combination of You Only Look Once (YOLO) and Residual Network (ResNet) deep learning algorithms for detection and confirmation.
• Tested the framework on optical and mid-wave infrared (MWIR) videos from the SENSIAC database.

Main Results:

• The proposed framework successfully demonstrated feasible human target detection up to 1500 meters.
• The method operates directly on compressive measurements, bypassing the need for full reconstruction.

Conclusions:

• Direct detection in the compressive domain is a viable strategy for efficient target identification.
• Further research is needed to enhance the target confirmation capabilities of the proposed framework for practical applications.