基于活动轮廓模型的脑部图像分割方法

Brain Image Segmentation Method Based on Active Contour Model

脑部组织的复杂性和磁共振成像机制的影响,造成脑部核磁共振数据带有灰度不均衡性和噪声干扰,使得传统分割模型得不到满意的结果。为此,提出一种活动轮廓分割模型,首先将带有加性偏移的概率函数融合到统计分割模型的能量函数中,降低由图像中灰度不均衡性造成的影响;其次采用截断高斯核函数构造局部邻域项,使模型有较好的抗噪声性能。实验结果表明,新模型对带有噪声和灰度不均衡性的图像能够得到较准确的分割结果。

The complexity of the brain tissue and the influence of the magnetic resonance imaging mechanism cause brain magnetic resonance imaging data with gray imbalance and noise interference.This makes the tradition segmentation model can not get satisfactory results.This paper presents an active contour segmentation model.First, fusing probability function with additive migration to the energy function of the statistical model, it reduces the impact of gray imbalance on the segmentation result.Secondly, constructing local neighborhood term with a truncated gaussian kernel function, and the model has good noise resistance.The experimental results show that new model can obtain the accurate segmentation results on images with noise and gray imbalance.