Approximate matching of digital point sets using a novel angular tree

Area of Science:

• Computer Vision
• Pattern Recognition
• Digital Geometry

Background:

• Digital point sets represent patterns in computer vision.
• Existing real-domain data structures need modification for digital applications.
• Efficient Approximate Point Set Pattern Matching (APSPM) in the digital plane is challenging.

Purpose of the Study:

• To propose a novel data structure for digital APSPM.
• To enable efficient and error-controllable circular range queries in the digital plane.
• To bridge the gap between real-domain and digital-domain APSPM algorithms.

Main Methods:

• Introduction of the "angular tree" data structure.
• Utilizing the farthest pair of points for initial correspondence.
• Integration of computational geometry and digital geometry principles.
• Development of an APSPM algorithm leveraging the angular tree.

Main Results:

• The angular tree facilitates efficient circular range queries.
• The proposed APSPM algorithm demonstrates effectiveness in the digital plane.
• The data structure supports error-controllable matching.
• Implementation and testing on various point sets confirm performance.

Conclusions:

• The angular tree is a versatile and efficient data structure for digital APSPM.
• This novel approach enhances pattern recognition in digital environments.
• The method provides a robust solution for matching point set patterns digitally.