摘要
针对传统的CT图像脑瘤分割方法往往需要先验知识指导的弊端,且脑瘤CT图像中肿瘤特征点与周围背景对比度较小,计算机自动提取这些特征点具有一定难度。为能清晰观察肿瘤具体位置,提出了一种基于数学形态学Top-Hat变换的脑瘤CT图像分割方法。利用形态学变换可得到分割迫切需要的谷峰值点、高低曲率点,再将变换结果与原始图像做"异或"运算就可以达到有效分割出肿瘤。实验结果表明,方法在无需先验知识指导情况下可使分割效果明显优于传统方法,具有很好的实用性。
The traditional methods are deficient for segmenting brain tumors in CT images because they need priori knowledge. What is more,they are difficult to automatically extract feature points because of low contrast between feature points and the background. In order to overcome these shortcomings,this paper proposes a method based on mathematical morphology. It mainly includes several steps. Firstly,Top-Hat Transformation and its duality are used to get the peak points,the valley points and the high-low curvature points that are urgently necessary for segmention. Secondly,transformed image with the original to ＂XOR＂ operator can achieve the purpose of effectively segment the tumor. The result shows that without the guidance of the priori-knowledge,this method is obviously better than traditional methods and has a good usability.
出处
《计算机仿真》
CSCD
北大核心
2010年第9期249-252,共4页
Computer Simulation
基金
西北工业大学研究生创新种子基金(Z200944)
关键词
数学形态学
脑瘤分割
图像分割
Mathematical morphology
Brain tumor segmentation
Image segmentation
