摘要
研究并实现了一种新的基于三角网格模型的区域分割算法,该算法首先计算网格模型中每个三角网格的法矢和面积,通过检测三角网格的面积及其法矢夹角进行区域生长,找到全部的边界网格,从而实现区域分割。在此基础上,应用了相邻块整体融合、块分解融合以及块边界光滑等方法,优化了分割结果。通过实验验证了算法的有效性和稳定性。
This paper proposed a new approach of border detection based on region growing for robust region segmentation of discrete point data implemented in reverse engineering system. In the proposed method, triangle meshes, firstly, were generated from the input discrete point data. The normal and the area of generated meshes were detected to seek boundary meshes making use of angle deviation criterion and area criterion based on a region-growing technique so that region segmentation is achieved.Furthermore, the post-processing techniques were applied so as to farther optimize the results of segmentation, including amalgamation between holistic segments, amalgamation between parts of segments, and smoothness of segment boundaries. Experimental results, finally, demonstrate that this region segmentation approach based on edge detection is effective and possesses better flexibility and stability.
出处
《中国机械工程》
EI
CAS
CSCD
北大核心
2005年第9期796-801,共6页
China Mechanical Engineering
基金
高等学校优秀青年教师教学科研奖励计划资助项目
江苏省青年科技基金资助项目(BQ2000004)
航空科学基金资助项目(01H52051)
关键词
逆向工程
区域分割
面积法矢准则
区域生长
reverse engineering
segmentation
area and normal criteria
region growing
