Compact Descriptor for Video Sequence Matching in the Context of Large Scale 3D Reconstruction
Compact Descriptor for Video Sequence Matching in the Context of Large Scale 3D Reconstruction
复制标题
大规模 3D 重建背景下视频序列匹配的紧凑描述符
DOI：
10.1007/978-3-642-32335-5_6
复制
发表时间：
--
期刊：
影响因子：
--
通讯作者：
Andreas Schilling
中科院分区：
文献类型：
--
作者：
Roman Parys;Florian Liefers;Andreas Schilling
如何有效地计算大型数据库中的图像关系是大规模图像三维场景重建中的关键问题之一。寻找描述相同三维几何形状的图像是摄像机标定和三维重建的前提。在本章中，我们提出了一种简单紧凑的描述符，使我们能够有效地计算视频序列之间的相似度。除了提供相似性度量之外，描述符还使选择匹配在一起的各个视频帧成为可能。有了我们的描述子，这一计算可以在类似于传统SIFT算法匹配两幅图像所需的时间内完成。利用提出的描述子，我们可以在视频流或图像序列之间建立一个大的关系图。该关系图将在以后组装大型几何模型时使用。
One of the key problems in the large scale reconstruction of 3D scenes from images is how to efficiently compute image relations in large databases. Finding images depicting the same 3D geometry is the pre-requisite for camera calibration and 3D reconstruction. In this chapter we present a simple and compact descriptor that enables us to efficiently compute similarity between video sequences. In addition to providing a similarity measure, the descriptor also makes it possible to select individual video frames that match together. With our descriptors, this computation can be done in a time similar to that required by the traditional SIFT algorithm to match just two images. Using the presented descriptors, we can build a large relation graph between video streams or image sequences. This relation graph is used later in assembling a large geometric model.
Web 搜索引擎的多层架构
DOI：
--
发表时间：
2003
期刊：
Proceedings of the IEEE/LEOS 3rd International Conference on Numerical Simulation of Semiconductor Optoelectronic Devices (IEEE Cat. No.03EX726)
影响因子：
--
作者：
K. M. Risvik;Yngve Aasheim;Mathias Lidal
通讯作者：
Mathias Lidal