Multiresolutional Cluster Segmentation Using Spatial Context
A multiresolutional cluster/relaxation image segmentation algorithm is described. A preliminary split-merge procedure generates variable-sized quadtree-blocks. These multiresolutional units are used in the subsequent clustering. A probabilistic relaxation procedure conducts the final labeling. A large reduction in data processing is attained by processing blocks rather than pixels, while still yielding good segmentation results.
KeywordsActive Block Spatial Context Final Segmentation Current Cluster Image Segmentation Algorithm
Unable to display preview. Download preview PDF.
- Nagin P.A., “Segmentation Using Spatial Context and Feature Space Cluster Labels,” Univ. of Massachusetts, COINS Techn. Report 78–8, May 1978.Google Scholar
- Gerbrands J.J., Backer E., “Split-and-merge Segmentation of SLAR-imagery: Segmentation Consistency,” Proc. 7ICPR, Montreal, Canada, 1984, pp. 284–286.Google Scholar
- Ball G.H., Hall D.J., “ISODATA — A Novel Method of Data Analysis and Pattern Classification,” Techn. Report SRI, California, 1965.Google Scholar
- Zucker S.W., Hummel R.A., Rosenfeld A., “An Application of Relaxation Labeling to Line and Curve Enhancement,” IEEE Trans. Comp., Vol, C-26, No. 4, April 1977, pp. 394–403.Google Scholar