基于形状约束的 LBF模型的水平集分割

摘要

LBF model is a typical local region model because it fits local regional information inside and outside through Gaussian kernel function.Moreover, the medical images are complicated and the segmentation often appears the objects similar to specific shape, it is for fas-ter finding the local regional feature shape ( such as circle and ellipse) .In the paper we propose that to incorporate the difference between the initial level set and the elliptic equations as the term of shape energy into LBF model, in the process of evolution it does not need to update corresponding shape through affine transformation, but only to carries out iteration calculation on the difference energy function of the initial contour and shape level set.In experimental results, we choose four eyes pictures to do the comparative analyses on LBF model and the LBF model with the shape items added respectively.The results show that the proposed model is able to detect the elliptic objects in the image ac-curately, and the segmentation accuracy reaches 96%, this also solves the problems of image greyscale inhomogeneity and being sensitive to noise.%由于LBF模型是通过高斯核函数来拟合局部内外的区域信息，是典型的局部区域模型。此外医学图像的复杂性且分割经常出现类似特定形状的目标。为了更快地找到局部的区域特征形状（如圆或椭圆），提出将初始水平集与椭圆方程的差异作为形状能量项融入到LBF模型中，在演化过程中无需经过仿射变换来更新相应的形状，只需对初始轮廓和形状水平集差异能量函数进行迭代计算。在实验结果中，选取4幅眼睛图片分别对LBF模型和加入形状项的LBF模型进行比较分析。结果表明，所提出的模型能够准确地检测图像中椭圆状的目标，且分割准确率达到了96％，也解决了图像灰度不均匀、对噪声敏感的问题。