C-V方法及其在体表损伤图像分割中的应用

法医体表损伤种类的多样性及损伤图像灰度图多峰值、边界模糊等特点决定了精确判定损伤部位的边界十分困难。C-V方法是一种变分水平集分割方法,它结合区域信息使得分割结果全局最优,同时采用水平集函数的表达方式,可以非常自然地处理轮廓线拓扑结构的变化。本文改进了C-V方法中符号距离函数的快速生成算法,并把它用于体表损伤图像分割,实验证明,C-V方法在体表损伤图像的分割中取得了较理想的效果,它可以较清晰地把损伤部位的边缘分割出来。特别是对边缘比较复杂的图像,C-V方法相比传统的分割方法,优势更加明显。

C-V Method and Its Application to Body-surface Trauma Image Segmentation

The variety of body-surface trauma,multiple peaks of its image and the fuzzy boundaries determine that it is so difficult to judge the boundaries of the trauma accurately.C-V is a method using level sets to easily deal with the topological changes,which can globally and optimally segment an image.We improve its signed Distance Function.The experiment shows that the result of the application of C-V to the segmentation of the body-surface trauma image is satisfactory.It can determine the edge of body-surface images more clearly than other traditional methods,especially when C-V is used to segment the images which edges are more complex.