Viewpoint selection of treelike structures

Abstract

We present a novel viewpoint selection of treelike object which reveals the branch patterns during the progressive exploration. In the intervention procedure, the catheter(or stent) advance to an unintended vessel, because some viewpoints can generate the collapsed spatial information between adjacent vessels in the branch configuration. Unlike other methods, our views focus on the inlets of adjacent vessels and the structures of the configuration, which are characterized by the branching graph and the exploration contexts. In addition, we choose the sample views which shows two adjacent branch structures as the main focus. Moreover, to evaluate a viewpoint, we introduce our viewpoint quality measure based on the sample views called the average of contribution(AC). Rather than sampling over the viewing sphere, to reduce searching cost, we attain several viewpoint candidates over the sample views and we will show the highest scored viewpoint from candidates is close to the optimal viewpoint in the sample space.