文物点云模型的优化配准算法

Optimal registration algorithm for point cloud model of cultural relics

针对带有噪声的文物点云模型,采用一种由粗到细的方法来实现其断裂面的精确配准。首先采用一种变尺度点云配准算法实现粗配准,即配准测度函数的尺度参数由大到小逐渐变化,可避免算法陷入局部极值,并获得较高精度的初始配准结果;然后采用基于高斯概率模型的改进迭代最近点(iterative closest point,ICP)算法进行细配准,可以有效地抑制噪声对配准结果的影响,实现断裂面的快速精确匹配。采用兵马俑文物碎块的配准结果表明,该优化配准算法能够实现文物断裂面的精确配准,而且在细配准阶段取得了较高的配准精度和收敛速度。因此该优化配准算法是一种快速、精确、抗噪性强的文物点云配准方法。

Aiming at the point cloud model of cultural relics with noise,this paper proposed a registration method from coarse to fine to register the fracture surfaces accurately. Firstly,it proposed a variable scale registration algorithm of point cloud model to complete coarse registration. In the coarse step,the scale parameter of registration measure function changed gradually from large to small,which could not only avoid the algorithm falling into local extreme value,but also obtained a higher accuracy of registration results. Secondly,it used an improved iterative closest point( ICP) algorithm based on Gaussian probability model to complete fine registration. The improved ICP algorithm could effectively suppress the impact of noise on registration results and achieved accurate registration of fracture surfaces. The registration results of Terracotta Warriors blocks show that the optimal registration algorithm can complete accurate registration of cultural relics and get high registration accuracy and convergence rate in fine registration step. So the proposed optimal registration algorithm is a fast,accurate and high anti-noise point cloud registration method of cultural relics.