Distance contained centerline for virtual endoscopy

Collision detection and flight path planning are difficult problems in virtual navigation in complex environments, especially in endoscopic navigation inside human organs. We solve these problems in elongated environment through a combination of shape keeping centerline and the distance from each internal point to boundary. Distance contained centerline (DCC) algorithm is presented to extract a line-shaped skeleton and compute the distance from skeleton to each erosion layer. In addition, we combine voxel-coding to remove spurious branches adaptively and reorganize topology accurately. Experiments show that the centerline obtained through DCC represents the complex shape of elongated objects fairly well and guides virtual navigation effectively.