Related Experiment Videos
A model for global symmetry detection in dense images
F Labonté1, Y Shapira, P Cohen
1Department of Electrical and Computer Engineering, Pavillon André-Aisenstadt, Ecole Polytechnique de Montréal, Québec, Canada.
Spatial Vision
|January 1, 1995
Summary
This study introduces a new computational model for detecting bilateral symmetry in images. The model, inspired by human perception, uses grouping to efficiently find symmetry in complex visual scenes.
Area of Science:
- Computer Vision
- Image Processing
- Computational Psychology
Background:
- Bilateral symmetry detection is crucial for image analysis and understanding.
- Previous models often face computational challenges with dense local features.
- Human symmetry perception involves pre-attentive grouping mechanisms.
Purpose of the Study:
- To propose a novel computational model for bilateral symmetry detection in images.
- To leverage psychophysical findings on grouping and symmetry perception.
- To reduce the computational load in symmetry detection algorithms.
Main Methods:
- A three-stage computational model: grouping, symmetry detection, and symmetry subsumption.
- Incorporation of a preliminary grouping stage to optimize processing.
- Implementation and evaluation of the proposed model.
Main Results:
- The model effectively detects bilateral symmetry in images with dense local features.
- The preliminary grouping stage significantly reduces computational requirements.
- Model performance shows strong agreement with human symmetry perception.
Conclusions:
- The proposed model offers an efficient and perceptually relevant approach to bilateral symmetry detection.
- Grouping is a key factor in facilitating and improving symmetry detection.
- This model advances the field of computational vision by integrating psychological insights.