Distinctive accuracy measurement of binary descriptors in mobile augmented reality.

Area of Science:

• Computer Vision
• Mobile Computing
• Image Processing

Background:

• Mobile Augmented Reality (MAR) demands real-time, robust visual descriptors.
• Floating-point descriptors (e.g., SIFT, SURF) are too slow for MAR.
• Binary descriptors offer efficiency but their suitability for MAR is unclear.

Purpose of the Study:

• To evaluate and compare the performance of four leading binary image descriptors for MAR applications.
• To identify the most efficient and robust descriptor for real-time MAR.

Main Methods:

• Comparative analysis of BRIEF, ORB, BRISK, and FREAK binary descriptors.
• Performance evaluation using the Mikolajczyk and ALOI datasets.
• Assessment based on computation time and robustness to brightness, scale, and rotation.

Main Results:

• FREAK demonstrated the shortest computation time among the tested descriptors.
• FREAK exhibited superior robustness against variations in scale, rotation, and brightness.
• The other binary descriptors showed limitations in efficiency or robustness for MAR.

Conclusions:

• FREAK is the most suitable binary descriptor for real-time Mobile Augmented Reality applications.
• Its efficiency and robustness make it ideal for MAR's dynamic viewing conditions.
• This finding guides the selection of optimal descriptors for enhanced MAR experiences.