浮雕显著性信息熵驱动的三维模型最优视点选择

Best Viewpoint Selection Driven by Relief Saliency Entropy

在三维模型最优视点选择中,将模型的浮雕显著性信息度量引入模型视点互信息的信息熵分析,提出一种结合视觉显著性的模型视见描述子刻画方法.针对模型视点球的各个视点计算模型可见面片浮雕显著性信息熵驱动的视见描述子,并从中选择视见描述子值最大的视点作为模型的最优观察视点.实验结果表明,由该方法获取的最优视点得到的视图能够反映和展示三维模型的视觉显著特征.

To automatically select an optimal viewpoint of 3D models,a novel view descriptor driven by relief saliency entropy is presented,which is defined by incorporating the relief saliency measure of a 3D model into the entropy analysis of viewpoint mutual information.The view descriptor with saliency information entropy is calculated for each pre-sampled viewpoint on a viewing sphere bounding the underlying 3D model.The viewpoint of the highest value of view descriptor is the one that defines the optimal view.The experimental results demonstrate that the optimal views selected by our proposed approach can always show the visually salient features of the underlying models.