COMPARISON OF SEVERAL APPROACHES FOR THE SEGMENTATION OF TEXTURE IMAGES

In this paper, several approaches including K-Means, Fuzzy K-Means (FK M),Fuzzy Adaptive Resonance Theory (ART2) and Fuzzy Kohonen Self-Organ izing Feature Mapping (SOFM) are adapted to segment the texture image. In our tests, five features, energy, entropy, correlation, homogeneit y, and inertia, are used in texture analysis, The K-Means algorithm ha s the following disavantages: (i) slow real-time ability, (ii) unstabi lity. The FKM algorithm has improved the performance of the unstabilit y by means of the introduction of fuzzy distribution functions. The Fu zzy ART2 has advantages, such as unsupervised training, low computatio n, and great degree of fault tolerance (stability/plasticity). Fuzzy o perator and mapping functions are added into the network to improve th e generality, The Fuzzy SOFM integrates the FKM algorithm into fuzzy m embership value as learning rate and updating strategies of the Kohone n network. This yields automatic adjustment of both the learning rate distribution and update neighborhood, and has an optimization problem related to FKM. Therefore, the Fuzzy SOFM is independent of the sequen ce of feed of input patterns whereas final weight vectors by the Kohon en method depend on the sequence. The Fuzzy SOFM is ''self-organizing' ' since the ''size'' of the update neighborhood and learning rate are automatically adjusted during learning, Clustering errors are reduced by Fuzzy SOFM as well as better convergence. The numerical results sho w that Fuzzy ART2 and Fuzzy SOFM are better than the K-Means algorithm s. The images segmented by the algorithms are given to prove their per formances.