摘要
为了提高航空发动机管路测量数据的反求建模效率,提出了一种区域增长分割算法。该方法主要是利用管路表面在曲面索引系数映射、主曲率映射、高斯映射上的特性,并基于均值漂移算法和遗传算法提出了区域分割算法中种子区域的选择策略。然后利用区域增长分割方法实现了对管路测量数据的分割。经过仿真和实测数据验证,所提管路分割算法具有较好的分割质量和效率。
To improve the reverse modeling efficiency of aeroengine pipelines, a region-growing segmentation algorithm for aeroengine pipeline point cloud data is proposed based on geometric attributes. The method maps respectively the geometric attribute value of measurement points to surface shape index paps of Koenderink, principal curvature coordinate system and gaussian sphere. Searches are performed for recognizing seed regions of point cloud data based on the characteristics of pipeline feature surfaces in the above geometric attribute mappings. The main tools that the recognition process uses are the mean shift algorithm and genetic algorithm. Point cloud data is segmented into many data regions and characterized as corresponding pipeline feature surfaces at the same time. Applications show that the proposed algorithm deals with pipeline point cloud data accurately and effectively.
出处
《航空学报》
EI
CAS
CSCD
北大核心
2008年第2期285-291,共7页
Acta Aeronautica et Astronautica Sinica
基金
总装备部预先研究项目
河南省科技攻关项目(032425068)
关键词
航空发动机管路
区域分割
几何属性
点云
反求工程
aeroengine pipeline
region segmentation
geometric attribute
point cloud
reverse engineering