运用人工鱼群算法的3D扫描碎片重建探究

Research on 3D scan debris reconstruction by using artificial fish algorithm

针对传统三维碎片拼接匹配过程中依赖单一特征及存在误差累积的问题,提出了一种运用鱼群算法的全局最优匹配方法。该方法先对碎片点云数据进行多特征提取,结合纹理、专家经验信息对混合在一起的多种类型碎片进行粗糙集分类,之后采用鱼群算法的最优解求得最佳匹配方案。实例验证所提全局匹配方法具有能力强、与初始位置无关及较强的稳健性等特点,为三维碎片的全局匹配提供了一种有效的解决方案。

Previous approaches for reconstructing fragments rely mainly on a single characteristic and thus may cause accumulative errors. In this paper,we present a global optimal matching method for 3 D fragments by using artificial fish school algorithm. The proposed method first extracts multi-featured elements from the point cloud of the fragments. Combined with texture and expert knowledge,rough set theory is then applied to classify multiple types of fragments. The artificial fish school algorithm is subsequently adopted to achieve optimal matching results. Results indicate that the proposed method is powerful,robust,and independent of initial position. The proposed method can be a new efficient tool for the global matching of fragments.