基于区域生长的网格模型分割技术

Region Growth Approach for Mesh Model Segmentation

针对三角形网格模型简化中保持细节特征的要求,提出了以顶点和三角形作为种子的区域生长策略,并采用可变阈值的生长方法.首先将与顶点相关的三角形面积的均值作为顶点的属性值,在区域生长过程中重复选择具有极大属性值的图元作为种子;在预分割后,区域要进行合并和平滑处理.实验表明,该方法快速有效,能够满足网格模型简化中保持局部区域特征的需要.

Focusing on detail preserving in triangle mesh model simplification, matching of the local average size of meshes surrounding a vertex or a triangle is taken as the criterion of region growth, and the matching threshold is lowering down in proportion to the increment of mesh density in a region. At first, the area of each mesh and their average value surrounding each vertex are computed and the process of region growth is started repetitively from a seed of the largest mesh. After completion of preliminary segmentation, merging of compatible separate regions and smoothing of their boundaries are undertaken. Experiment shows that this approach is fast and effective, meeting the requirement of keeping local features in mesh simplification.