改进的三维模型形状分布检索算法

针对传统D1距离形状分布函数获取采样点计算复杂、模型内容描述不充分和检索速率低下等问题提出了一种改进方法。该方法的关键点是：首先采用平移和缩放对模型进行标准化处理，用于减少面片之间的差异，使得采样点均匀地落在模型的表面；其次采用三角面片的索引号进行随机数的生成，并且利用三角面片的重心和质心进行有效的计算，以便用于缩短模型的处理时间和提高检索速率。利用普林斯顿大学三维模型数据库中的部分模型作为实验数据，实现结果表明：改进的方法不会降低模型的检索性能，并有效地减少了模型查询和处理时间。

Improved shape distribution retrieval algorithm of 3D models

This paper proposed an improved method for traditional D1 shape distribution function with complex computation which obtains large amount of samples and can not fully describe the model content and also has low search rate.There are key points of this method as follows: first,it normalizes the model by using translation and zoom so that it can reduce the difference between the triangular patches and make the sampling points uniformly fall on the surface of the model;secondly,it uses the index number of the triangular patches to generate the random numbers,and uses the gravity of model and centric of triangular patch for effective computation,to reduce the processing time and improve the efficiency of three-dimensional model retrieval.In addition,this paper validated the practical value of improved D1 shape distribution method by using the idea of semi-automatic model classification.Using part models of three-dimensional model database in Princeton University,the results show that: both improved methods will not decrease the precision and recall of model retrieval,but effectively reduce the processing time.