基于球面图像的三角网格模型分割

为了从局部形状的角度实现对三角网格模型的管理和重用,提出了一种基于球面图像的三角网格模型分割方法.通过球面参数化及球面划分,将三角网格模型的表面属性信息映射到球面图像中;利用成熟的区域生长、区域合并图像分割算法对球面图像进行分割;将球面图像的分割结果转换为三角网格模型的分割结果.实验结果表明:该分割方法可以对不均匀的低分辨率三角网格模型进行有效分割,降低了几何属性估算对分割结果的影响,不会发生过分割现象,不需要进行分割的后续处理.

Triangular Mesh Segmentation Based on Spherical Images

In order to manage and reuse triangular mesh models with local shapes,a segmenting method for triangular mesh models based on spherical images was proposed.With this method geometric signals of a triangular mesh model are mapped into spherical images through spherical parameterization and partition of a sphere.Then the spherical images are segmented by sophisticated image processing algorithms of region growing and region merging.Finally,the segmentation result of the spherical images is mapped into the related triangular mesh model.Experimental results show that the proposed method is effective to non-uniform,low-resolution triangular mesh models,the impact of accuracy of estimated geometric signals on segmentation results is reduced,and the segmentation result does not need to be post-processed.