Infrared target tracking with kernel-based performance metric and eigenvalue-based similarity measure

Area of Science:

• Computer Vision
• Signal Processing
• Infrared Imaging

Background:

• Infrared target tracking is crucial for various applications.
• Existing methods face challenges in maintaining accuracy and robustness.
• Online performance evaluation and adaptive correction are needed.

Purpose of the Study:

• To develop a comprehensive infrared target tracking framework.
• To enhance tracking accuracy through integrated performance evaluation and position correction.
• To demonstrate the framework's efficacy on real-world data.

Main Methods:

• Utilized mean shift tracking for initial target acquisition.
• Implemented a kernel-based metric for online tracking performance evaluation.
• Developed an eigenvalue-based similarity measure for position correction using weighted covariance matrices.

Main Results:

• The proposed framework successfully tracked targets in infrared image sequences.
• The integrated performance evaluation and correction module improved tracking robustness.
• Experimental results validated the efficacy of the eigenvalue-based similarity measure.

Conclusions:

• The presented framework offers an effective solution for infrared target tracking.
• The combination of mean shift, performance evaluation, and position correction enhances reliability.
• The novel similarity measure contributes to improved tracking accuracy and stability.