基于结构连续先验的CT图像序列自适应分割算法  

Adaptive Segmentation Algorithm for CT Image Sequence Based on Structure Continuity as Prior Information

作　　者：王苏恺 潘晋孝[1] 陈平[1] Wang Sukai Pan Jinxiao Chen Ping(Shanxi Provincial Key Laboratory of Signal Capturing ＆ Processing, North University of China, Taiyuan, Shanxi 030051, China)

机构地区：[1]中北大学信息探测与处理山西省重点实验室,山西太原030051

出　　处：《激光与光电子学进展》2016年第11期125-130,共6页Laser & Optoelectronics Progress

基　　金：国家自然科学基金(61227003;61301259;61471325;61571404);山西省自然科学基金(2015021099)

摘　　要：在土层计算机断层成像(CT)图像序列中,由于分割目标的灰度值分布不一致,为目标区域的提取带来一定的难度。为解决此类问题,建立了基于结构先验的自适应分割方法。该方法通过设定初始切片的阈值,利用图像序列之间同一结构表征对象的连续性,以目标区域的面积变化作为约束条件进行递归分割,实现CT图像的自适应分割。对分割后的目标区域进行三维重构,以实现三维可视化。结果表明,该方法适用于分割目标灰度值不一致的图像序列,计算量小且能够精确地分割成千上万层图像中的目标。Soil computer tomographic （CT） images sequence are difficult to extract the target because the gray of the segmented object has no uniform distribution. In order to solve the problems, we come up with an adaptive segmentation algorithm of CT images sequence based on structure continuity as prior information. We choose the initial image and determine the most appropriate threshold. Based on the continuity of the same object structure between the images sequence, the areal change of the object region is used as the prior information to perform recursive segmentation. The adaptive segmentation of CT images are achieved, and 3D visualization is realized by 3D reconstruction on the segmented object area. The results show that the proposed algorithm is applicable to CT images sequence with non-uniform gray distribution in the object regions, it results in small amount of calculation, and can precisely segment objects in thousands of target images.

关 键 词：图像处理 CT图像 自适应 先验结构信息 递归分割 三维可视化 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]