Symmetry detection on shapes and objects in 2D images using feature points

Abstract

Symmetry is a worldwide phenomenon that exists in all forms and shapes in the world. It is a crucial part of geometry, nature, and shapes. It helps to organise our world conceptually by creating patterns that we can analyse and study. With the recent resurging interest of computer vision and computer graphics, it is no doubt that the study of the detection of symmetry will be useful in developing new algorithms or refining existing methods. Potential applications like camera tracking and object pose estimation can benefit from the findings of symmetry detection. This project presents an existing method to detect symmetry in an image, by grouping feature points based on their underlying symmetry and characteristics. With the method, the project aims to analyse how symmetry detection is done on the lowest level and evaluate the effectiveness of the detection, also stating the limitations of the current method.