You are here

Multivariate image segmentation using semantic region growing with adaptive edge penalty

TitleMultivariate image segmentation using semantic region growing with adaptive edge penalty
Publication TypeJournal Article
Year of Publication2010
AuthorsQin, K., and D. A. Clausi
JournalIEEE Transactions on Image Processing
Volume19
Pagination2157 - 2170
ISSN1057-7149
Keywordsadaptive edge penalty, Algorithms, Automated, Computer-Assisted, image enhancement, image interpretation, image segmentation, iterative methods, Markov processes, Markov random field, MRF spatial context model, Multivariate Analysis, multivariate image segmentation, multivariate iterative region growing, Pattern Recognition, region-level k -means based initialization, Reproducibility of Results, semantic region growing, Semantics, Sensitivity and Specificity
Abstract

Multivariate image segmentation is a challenging task, influenced by large intraclass variation that reduces class distinguishability as well as increased feature space sparseness and solution space complexity that impose computational cost and degrade algorithmic robustness. To deal with these problems, a Markov random field (MRF) based multivariate segmentation algorithm called #x201C;multivariate iterative region growing using semantics #x201D; (MIRGS) is presented. In MIRGS, the impact of intraclass variation and computational cost are reduced using the MRF spatial context model incorporated with adaptive edge penalty and applied to regions. Semantic region growing starting from watershed over-segmentation and performed alternatively with segmentation gradually reduces the solution space size, which improves segmentation effectiveness. As a multivariate iterative algorithm, MIRGS is highly sensitive to initial conditions. To suppress initialization sensitivity, it employs a region-level k -means (RKM) based initialization method, which consistently provides accurate initial conditions at low computational cost. Experiments show the superiority of RKM relative to two commonly used initialization methods. Segmentation tests on a variety of synthetic and natural multivariate images demonstrate that MIRGS consistently outperforms three other published algorithms.

DOI10.1109/TIP.2010.2045708