Directional intensified feature description using tertiary filtering for augmented reality tracking.

Area of Science:

• Computer Vision
• Augmented Reality
• Image Processing

Background:

• Augmented Reality (AR) applications across diverse fields like engineering, medicine, and gaming rely heavily on robust image tracking.
• Current image tracking systems often lack the necessary robustness and efficiency for seamless virtual-physical world integration.
• Developing robust tracking algorithms is a significant challenge in implementing effective AR experiences.

Purpose of the Study:

• To enhance user experience in Augmented Reality by improving image tracking robustness.
• To introduce a novel feature descriptor, Directional Intensified Features with Tertiary Filtering (DITF), for more reliable image tracking.
• To validate the robustness of the DITF algorithm against various image transformations.

Main Methods:

• Image description using Directional Intensification using Tertiary Filtering (DITF).
• Utilizing Tri-ocular, Bi-ocular, and Dia-ocular filters to intensify features in multiple directions.
• Performance analysis and validation using the Oxford dataset.

Main Results:

• The DITF model achieved high repeatability scores: 100% for illumination variation, 100% for blur changes, and 99% for view-point variation.
• Comparative analysis demonstrated that DITF outperforms state-of-the-art descriptors including BEBLID, BOOST, HOG, LBP, BRISK, and AKAZE in precision and recall.
• The DITF algorithm exhibits enhanced robustness against image transformations.

Conclusions:

• The DITF algorithm significantly improves image tracking robustness, crucial for enhancing AR user experiences.
• DITF offers a more reliable and efficient solution compared to existing feature descriptors for AR applications.
• The proposed method addresses key shortcomings in current image tracking systems, paving the way for more advanced AR implementations.