摘要： 提出一种基于切片图像的体素标号测量算法。通过对CT图像序列作预处理建立点云模型，由外向内对体素进行序数标号，采用局部极值搜索和体心精简得到球粒体体心，由体心标号值完成统计。实验结果表明，随机点噪声范围在35%以内时，算法能准确统计出存在粘连的球粒体的数目与空间分布，且体积估算误差小于5%。

关键词: 空间球粒体, 统计测量, 体素标号, CT切片图像, 点云模型

Abstract: A measurement algorithm of voxels marking number based on slice images is presented. After preprocessing the CT image sequence, a point cloud model is built. The number mark for voxels is signed orderly from outside to center. Through local extremum search, the body-centers are got after downsizing. Statistics are completed using mark values of the body-centers. Experimental results show that the count and distribution of space spherules existed adhesion can be accurately measured by this method, when the random point noise probability is within 35%. And the volume estimation error is less than 5%.

Key words: space spherules, statistical measurement, voxels marking number, CT slice image, point cloud model

中图分类号: 

• TP391