Related Experiment Videos
A flexible similarity measure for 3D shapes recognition
1Departamento de I.E.E. y Automática, Universidad de Castilla La Mancha, 13071 Ciudad Real, Spain. Antonio.Adan@uclm.es
IEEE Transactions on Pattern Analysis and Machine Intelligence
|November 4, 2004
Summary
This study introduces a new 3D object recognition method using Modeling Wave (MW) topology and Cone-Curvature (CC) features for flexible similarity measurement, proving effective even with partial data, noise, and occlusion.
Area of Science:
- Computer Vision
- 3D Object Recognition
- Computational Geometry
Background:
- 3D object recognition is crucial for many applications.
- Existing methods struggle with partial data, noise, and occlusion.
- Spherical models and topology-based features offer potential for robust recognition.
Purpose of the Study:
- To present a novel strategy for 3D object recognition.
- To introduce a flexible similarity measure based on Modeling Wave (MW) topology.
- To develop a robust method for recognizing objects from partial information.
Main Methods:
- Utilized Modeling Wave (MW) topology for n-connectivity in 3D meshes.
- Introduced Cone-Curvature (CC) features derived from MW topology.
- Developed a similarity measure based on CC features for object comparison.
- Tested the method on range data of diverse 3D shapes.
Main Results:
- Demonstrated successful 3D object recognition using the proposed CC features and similarity measure.
- Showcased the method's adaptability to partial object information.
- Validated the robustness of the approach against noise and occlusion in complex scenes.
Conclusions:
- The new strategy based on MW topology and CC features provides a robust and adaptable approach to 3D object recognition.
- The method shows significant potential for real-world applications, including those with incomplete or degraded data.