基于相似性学习的三维模型最优视图选择算法

摘要

Focusing on the problem that the existing optimal view measurement is difficult to measure different kinds of 3D objects, a new selection algorithm of optimal view for 3D object based on similarity learning is proposed in this paper.In this method, the optimal view samples for all kinds of 3D objects in the model library will be set up based on the use's knowledge firstly.Then through similarity learning, the optimal view of the input 3D object can be gotten.Here, the AdaBoost algorithm is used on the shape feature of input 3D object for similarity learning in order to get the optimal view sample of this object.Then the rendering views of the 3D object coming from different viewpoints are compared and analyzed with the optimal view sample on shape similarity matching.The rendering view with highest similarity will be regarded as the optimal view of the 3D Object.The experiments show that the optimal view selected can not only effectively close the user's knowledge, but also have good stability.%针对已有最优视图度量难以适用于不同类型的三维模型,提出基于用户知识、在模型库中为各类模型建立最优视图样例,并在此基础上进行相似性学习,根据相似性度量获得输入模型最优视图的选择算法.首先采用AdaBoost算法对输入三维模型形状特征进行相似性学习,得到该模型的最优视图样例；然后将输入模型从不同视点得到的渲染视图和最优视图样例进行形状相似性分析,以相似度最高者作为输入模型的最优视图.实验结果表明,采用文中算法得到的最优视图不仅可以有效地逼近用户选择结果,而且具有较好的稳定性.