人工智能培训

搜索

人工智能论文:通过图像匹配实现精确的相机地理定位(Towards Accurate Camera Geopositioning by Image Matchin

[复制链接]
deshuo 发表于 2019-3-15 12:32:16 | 显示全部楼层 |阅读模式
deshuo 2019-3-15 12:32:16 523 0 显示全部楼层
人工智能论文:通过图像匹配实现精确的相机地理定位(Towards Accurate Camera Geopositioning by Image Matching)在这项工作中,我们提出了一个相机定位系统,该系统基于匹配的图像与具有全景图像的数据库。为了匹配,我们的系统使用基于卷积特征从全局图像描述符聚合的存储器向量,以便于在数据库中快速搜索。为了加速搜索,使用聚类算法来平衡地理定位和计算时间。我们使用新的离群值去除算法从查询图像中细化获得的位置。对于全景到全景匹配,获得查询图像的匹配,其中召回@ 5大于90%。我们将来自地理位置相邻的位置的可用全景图集中到单个紧凑的表示中,并且仅以很小(约3%)的召回损失为代价观察大约50%的计算增益。最后,我们提出了一种坐标估计算法,该算法将中位数定位误差减少到最多20%。
In this work, we present a camera geopositioning system based on matching aquery image against a database with panoramic images.For matching, our systemuses memory vectors aggregated from global image descriptors based onconvolutional features to facilitate fast searching in the database.To speedup searching, a clustering algorithm is used to balance geographicalpositioning and computation time.We refine the obtained position from thequery image using a new outlier removal algorithm.The matching of the queryimage is obtained with a recall@5 larger than 90% for panorama-to-panoramamatching.We cluster available panoramas from geographically adjacent locationsinto a single compact representation and observe computational gains ofapproximately 50% at the cost of only a small (approximately 3%) recall loss.Finally, we present a coordinate estimation algorithm that reduces the mediangeopositioning error by up to20%.人工智能论文:通过图像匹配实现精确的相机地理定位(Towards Accurate Camera Geopositioning by Image Matching) pyG556B55uj7r7G7.jpg
URL地址:https://arxiv.org/abs/1903.05454     ----pdf下载地址:https://arxiv.org/pdf/1903.05454    ----人工智能论文:通过图像匹配实现精确的相机地理定位(Towards Accurate Camera Geopositioning by Image Matching)
回复

使用道具 举报

您需要登录后才可以回帖 登录 | 立即注册

本版积分规则 返回列表 发新帖

deshuo当前离线
新手上路

查看:523 | 回复:0

快速回复 返回顶部 返回列表