摘要
随着三维扫描技术的快速发展,获取各类场景的点云数据已经非常简单快捷;加之点云数据具备不受光照、阴影、纹理的影响等优势,基于点云的三维物体识别已成为计算机视觉领域的研究热点。首先,对近年来面向点云数据的三维物体识别方法进行归纳和总结;然后,对已有方法的优势及缺点进行分析;最后,指出点云物体识别中所面临的挑战及进一步的研究方向。
With the rapid development of 3D scanning technology, it is convenient to obtain point clouds of different scenes. Since point clouds are not influenced by light, shadows and textures, recognizing 3D object from scene point clouds has become a research hotspot of computer vision. This paper first summarized the 3D object recognition methods from point clouds in recent years. Then the advantages and disadvantages of the existing methods were discussed. Final- ly, the challenges and further research directions of object recognition were pointed out.
出处
《计算机科学》
CSCD
北大核心
2017年第9期11-16,共6页
Computer Science
基金
国家自然科学基金(61602373
61472319
61401355)
西安市碑林区科技项目(GX1615)
陕西省自然基金(2015JZ015)资助
关键词
点云数据
三维物体识别
特征提取
图匹配
Point clouds,3D object recognition,Feature extraction,Graph matching
