基于形态学重建滤波的脑部磁共振图像分割

Brain Magnetic Resonance Image Segmentation Based on Morphological Reconstruction Filter

在阐述形态学基本原理、形态学重建滤波原理和方法的基础上,提出了形态学交变序列重建滤波,并应用于脑部磁共振图像分割。试验结果表明,该方法能有效地滤除无用的信息而保留图像原始形状不变。再应用形态学梯度运算、测地距离和流域变换方法就能准确地分割出脑部磁共振图像的头盖骨和脑膜。

Based on basic theories and methods of mathematical morphology and morphological reconstruction filter, morphology alternating sequential filter by reconstruction is put forward. It is applied to segment brain magnetic resonance image. The experimental results indicate that useless information in the brain magnetic resonance is filtered effectively and the original shape is left unaltered by the proposed method. After applying the morphological gradient operator, geodesic distance and watershed transformation to the processed image, the cranium and meninges of brain magnetic resonance image are segmented precisely.