中文摘要：

针对虚拟内窥镜中心路径提取算法时间效率不高的问题,通过改进边界距离变换,提出了中心路径的快速提取算法。首先建立最小距离场,在边界内推过程,仅扫描并处理与边界点面连接的点并对其设标记属性,每次向里剥离一层体素标记属性随之增加,直到对内部点集都置到边界的最小边界距离值（DFB）;然后找到DFB值不小于其所有面邻接点DFB值的体素点,即3D局部最大值;最后用最短路径相连3D局部最大值,形成中心线。实验结果表明,改进的中心路径提取算法的时间效率较传统距离变换算法有很大提高。

英文摘要：

Aiming at the problem that the time efficiency of centerline extraction algorithm of virtual endoscopy is not high,this paper presented fast centerline extraction through improved boundary distance transformation. Firstly,the method established the minimum distance field. In the process of boundary recursion inward,it scanned the face-connecting points connected with boundary,processed and labeled with attribute. The labeled attribute increased with stripping a layer of voxels inward every time. Until every point of internal point set had minimum distance from boundary（ DFB）. Then,the voxel whose DFB value was not less than the DFB value of the point which was face-connecting connected with it was called as 3D-local-maximum. Finally,the method connected all of 3D-local-maximums each other through the shortest path to generate the centerline. The experimental result show that the time efficiency of the improved centerline extraction algorithm has greatly improved compared with the traditional distance transformation.