中文摘要：

为解决木材细胞纤维图像分割中的某些图像分割不连续的现象，引入了基于形变模型（Deformable Models）的水平集（Level Set）方法对木材细胞图像进行分割，并用Matlab实现了基于该形变模型的窄带（Narrow Band）快速算法。对针叶材和阔叶材的显微图片进行仿真试验表明，该方法适合于对具有分支、突触以及拓扑结构变化的木材细胞图像进行快速精确分割，不但具有全局优化的能力，而且可以检测出模糊或离散状边界，对噪声也有一定的鲁棒性。

英文摘要：

Level set method based on deformable models was applied to image segmentation of wood cells, and a fast level set method called Narrow Band algorithm is realized with Matlab aiming to solve the discontinuous phenomenon occurred in the image segmentation of wood cells. Simulation experiments were performed on microscopic images of wood cells from coniferous and deciduous trees. Results showed that the algorithm with goal optimization capacity is quick and precise in segmenting images with significant protrusions or transformable topology, which can examine fuzzy and scattered boundaries, and also has certain level of noise robustness.