Abstract
In this paper, we introduce a new approach for fitting of a bounding rectangle to closed regions. In this approach the coordinates of the vertices are computed directly using a closed-form solution. This approach is based on simple coordinate geometry and uses the boundary points of regions. Using a least-square approach we determine the directions of major and minor axes of the object, which gives the orientation of the object. The four vertexes of the bounding rectangle are computed by pair wise solving the four straight lines. Examples from synthetic data and some real-life data show that the approach is both accurate and efficient.
Original language | English (US) |
---|---|
Pages (from-to) | 1981-1989 |
Number of pages | 9 |
Journal | Pattern Recognition |
Volume | 40 |
Issue number | 7 |
DOIs | |
State | Published - Jul 2007 |
Keywords
- Least-square method
- Major axis
- Minimum-bounding box
- Minor axis
- Segmentation
- Shape features
ASJC Scopus subject areas
- Software
- Signal Processing
- Computer Vision and Pattern Recognition
- Artificial Intelligence