Functional transforms in MR image segmentation
Andrea Gavlasova, Aleš Procházka, Jaroslav Poživil, Oldřich Vyšata · 2008
Image segmentation, feature extraction and image components classification form a fundamental problem in many applications of multi-dimensional signal processing. The paper is devoted to the use of watershed transform for image segmentation in connection with wavelet transform allowing image de-noising and image components feature extraction. Proposed methods are applied for biomedical image analysis and processing. The study of MR image segmentation devoted to the detection of its specific components results in the proposal of the appropriate image preprocessing to reduce problems of its oversegmentation. Resulting algorithms include the use of wavelet transform and gradient methods in the preprocessing stage. Proposed algorithms are verified for simulated images and applied for a selected MR biomedical images containing different structures.