Enhanced robust spatial feature selection and correlation filter learning for UAV tracking.

Area of Science:

• Computer Vision
• Machine Learning
• Signal Processing

Background:

• Spatial boundary effects degrade discriminative correlation filter (DCF) model performance.
• Extracting features from wider regions introduces background noise, reducing discrimination power.

Purpose of the Study:

• To propose an innovative method, EFSCF, for robust visual tracking.
• To effectively handle spatial boundary effects and suppress background noise in DCF models.

Main Methods:

• Jointly sparse feature learning to address boundary effects and background noise.
• Utilizing the ℓ_2,1-norm for robustness against training outliers and non-Gaussian noise.
• Implementing a jointly sparse feature selection scheme for simultaneous row and column regularization of the filter.

Main Results:

• EFSCF demonstrates enhanced robustness against noise and appearance changes.
• The proposed method achieves superior tracking performance compared to state-of-the-art trackers.
• Experiments on UAV datasets validate the effectiveness of EFSCF.

Conclusions:

• EFSCF offers an effective solution for visual tracking challenges, particularly boundary effects and noise.
• The novel application of structural sparsity in filter learning advances the field.
• The method provides a significant improvement in tracking accuracy and robustness.