Parallel Three-Branch Correlation Filters for Complex Marine Environmental Object Tracking Based on a Confidence...

Area of Science:

• Computer Vision
• Robotics
• Marine Technology

Background:

• Marine object tracking is vital for search and rescue (SAR).
• Complex marine environments (variable light, waves, occlusion) challenge existing tracking systems.
• Need for accurate, real-time, and robust visual trackers in marine settings.

Purpose of the Study:

• To propose an efficient dynamic visual tracker for complex marine environments.
• To enhance the robustness and accuracy of marine object tracking.
• To prevent model drift and improve overall tracking performance.

Main Methods:

• Developed a parallel three-branch correlation filter approach.
• Incorporated a confidence mechanism for robust tracking.
• Utilized weighted fusion of response maps for accurate object localization.
• Employed Gaussian-triangle joint distribution in the training phase.
• Integrated a confidence metric for filter update verification.

Main Results:

• The proposed tracker demonstrated enhanced robustness against environmental challenges.
• Accurate object localization achieved through weighted fusion of response maps.
• Effective prevention of model drift via confidence-based filter updates.
• Outperformed existing trackers in complex marine environments.

Conclusions:

• The parallel three-branch correlation filter tracker is highly effective for marine object tracking.
• The confidence mechanism significantly improves tracking accuracy and robustness.
• The proposed method offers a promising solution for SAR operations in challenging marine conditions.