Combination of Multiple Segmentations by a Random Walker Approach
In this paper we propose an algorithm for combining multiple image segmentations to achieve a final improved segmentation. In contrast to previous works we consider the most general class of segmentation combination, i.e. each input segmentation can have an arbitrary number of regions. Our approach is based on a random walker segmentation algorithm which is able to provide high-quality segmentation starting from manually specified seeds. We automatically generate such seeds from an input segmentation ensemble. Two applications scenarios are considered in this work: Exploring the parameter space and segmenter combination. Extensive tests on 300 images with manual segmentation ground truth have been conducted and our results clearly show the effectiveness of our approach in both situations.
Unable to display preview. Download preview PDF.
- 1.Cingue, L., Cucciara, R., Levialdi, S., Martinez, S., Pignalberi, G.: Optimal range segmentation parameters through genetic algorithms. In: Proc. of 15th Int. Conf. on Pattern Recognition, vol. 1, pp. 474–477 (2000)Google Scholar
- 2.Cour, T., Benezit, F., Shi, J.: Spectral segmentation with multiscale graph decomposition. In: Proc. of CVPR, pp. 1124–1131 (2005)Google Scholar
- 3.Felzenszwalb, P., Huttenlocher, D.: Efficient graph-based image segmentation. Int. Journal of Computer Vision, 167–181 (2004)Google Scholar
- 7.Martin, D., Fowlkes, C., Tal, D., Malik, J.: A database of human segmented natural images and its application to evaluating segmentation algorithms and measuring ecological statistics. In: Proc. 8th Int. Conf. Computer Vision, vol. 2, pp. 416–423 (2001)Google Scholar