Abstract
This paper presents an image up-scaling based approach to improving the quality of image segmentation that uses linked mean-shift vectors [6]. While already effective, the segmentation quality of [6] can be further improved by strengthening the approximation process it uses. Making up for the weakness, the proposed algorithm remarkably outperformed existing competitor algorithms in segmentation quality by Probabilistic Rand Index [8] and Segmentation Covering [9] for Berkeley Segmentation Dataset [7].
Original language | English |
---|---|
Pages (from-to) | 1176-1179 |
Number of pages | 4 |
Journal | Digest of Technical Papers - SID International Symposium |
Volume | 45 |
Issue number | 1 |
DOIs | |
State | Published - Jun 2014 |
Keywords
- image segmentation
- mean-shift algorithm
- single instruction multiple data