【摘要】 符合人类认知的三维网格模型自动分割结果可应用于许多领域研究，准确的分割也更有利于三维测量。针对现有三维网格分割结果不稳定、易受形状、姿态干扰等问题，本文采用一种基于相互可见性和表面偏移的形状直径函数来改善三维模型自动分割的结果。首先，输入网格模型，利用谱方法进行过分割；其次，基于相互可见性迭代，合并过分割块；最后，利用表面偏移的形状直径函数合并得到分割结果。利用本文分割算法对人体模型进行分割，结合传统的切片法和椭圆拟合法对分割后的部位进行围度测量，通过对比多种三维分割方法，分析本文测量结果和交互式测量软件测量的结果，证明本文分割方法效果较好，测量结果具有一定可靠性。更多还原

【Abstract】 The automatic segmentation results of 3 D mesh model in line with human cognition can be used in many fields, and accurate segmentation is also more conducive to 3 D measurement. For the problems that the existing 3 D mesh segmentation results are unstable and easy to be disturbed by shape, We use mutual visibility and the shape diameter function based on surface offset to improve the automatic segmentation results. Firstly, we input the mesh model, segmented it by spectral method, then merged the over cut blocks based on mutual visibility, and finally merged the segmented blocks by the shape diameter function of surface offset. We use the traditional slice method combined with the ellipse fitting method to measure the circumference of the segmented parts. In this paper, several 3 d segmentation methods are compared, and the results show that the proposed method is effective. The human body is segmented by the segmentation algorithm, and the circumference of the segmented parts is measured by combining the traditional slice method and ellipse fitting method. By comparing and analyzing the measurement results in this paper with those measured by interactive measurement software, it is proved that the measurement results in this paper have certain reliability.更多还原