多尺度框架下基于C-V模型的肺纹理分割

摘要

使用水平集法对CT图像进行肺纹理分割时存在运算时间长、噪声影响大等问题.为此,提出一种多尺度框架下基于C-V水平集法的三维肺纹理分割方法.根据小波变换理论,对低频图像进行三维体数据的C-V水平集分割,参照演化结果初始化高频图像水平集,逐步得到原胸肺CT三维图像的分割结果.实验结果表明,该方法能缩短运算时间,并减小噪声对分割结果的影响.%In chest CT images analysis, for the reasons of CT noise and long calculating time of 3D lung marking segmentation using Level-set, this paper proposes a method that combines 3D C-V Level-set model with multiple scales analysis to segment lung marking. According to the theory of wavelet transform, volume data of C-V model processing starts from the lowest scales to the highest scales step by step. Experimental result shows that the new method can reduce processing time drastically and weaken influence of noise.