A novel Gravity-FREAK feature extraction and Gravity-KLT tracking registration algorithm based on iPhone MEMS mobile...

Area of Science:

• Computer Vision
• Augmented Reality
• Sensor Fusion

Background:

• Mobile devices have limited computing power and memory, hindering advanced AR experiences.
• Traditional feature description algorithms like FREAK face challenges in resource-constrained environments.
• Augmented reality (AR) requires robust and efficient feature extraction and tracking for seamless integration of digital information.

Purpose of the Study:

• To develop a computationally efficient feature description algorithm for AR on mobile devices.
• To leverage Micro-electromechanical Systems (MEMS) sensors to enhance feature extraction and tracking.
• To improve the accuracy and robustness of AR registration under mobile conditions.

Main Methods:

• Proposed Gravity-FREAK algorithm using gravity projection vector for feature orientation.
• Implemented adaptive and generic corner detection for Gravity-FREAK matching purification.
• Utilized a Gravity Kaneda-Lucas Tracking (KLT) algorithm based on MEMS sensors for target tracking.

Main Results:

• Gravity-FREAK reduces computational cost by eliminating neighborhood gray gradient calculations.
• Improved accuracy in feature extraction compared to traditional methods.
• Enhanced robustness of tracking registration algorithms in mobile environments through MEMS sensor integration.

Conclusions:

• Gravity-FREAK offers a viable solution for efficient AR on mobile devices.
• MEMS sensor integration significantly improves feature description and tracking performance.
• The proposed methods enhance the overall AR interaction experience by overcoming hardware limitations.