Hierarchical image segmentation using adaptive pattern sizes
K. Ohkura, Yücel Uğurlu, Hidezaku Nishizawa, Takashi Obi, Akira Hasegawa, Masahiro Yamaguchi, Nagaaki Ohyama · 2003
In this paper we propose a method for unsupervised image segmentation, which is suitable for finding the features contained in medical images. The method is based on the hierarchical clustering method in multi-dimensional pattern vector space. We consider to change the size of pattern vectors adaptively to explore useful image features which can be used in medical diagnosis. We have tested our method on the simulation image, which is generated by the Markov Random Field (MRF) model, and the real medical images, photomicrographs of colon tumor, and its effectiveness is confirmed.