摘要
在景点导游系统、户外广告等增强现实系统中,户外建筑物识别有较大的应用价值。本文讨论了一种户外建筑物的识别方法,其特点是把整个系统分为服务器/客户端设计。设计中,在服务器端可以利用服务器强大的计算及存储能力来计算感兴趣区域（POI）的建筑物的ORB特征;客户端通过摄像头等设备抓取视频帧并获取当前视频帧的ORB特征传递给服务器;服务器通过汉明距离匹配这两组ORB特征,得到最匹配的建筑物信息返回给客户端。为了提高识别效率,预先将POI划分成二维网格,并根据客户端的GPS信息定位在其中一个单元格中,匹配时,只需要遍历比较已定位单元格的8个邻域中的建筑物即可。
In the scenic spot guide system, outdoor advertising and other augmented reality system, outdoor building identification has great application value. This paper discuss a method for the identification of outdoor buildings, which divide the system into server / client component. In this design, on the server side can use server’s powerful computing and storage capacity to calculate the orb features of region of interest（POI） buildings; client through video cameras capture video frames and obtain the orb features of current video frame and send it to the server; the server use hamming distance matching method match the two sets of orb feature, get the best matches of the building information and send it back to the client. In order to improve the recognition efficiency, POI is divided into grids, and according to the client’s GPS information, location the client in one of the cells,do matching work, only need to traverse and compare to the 8 neighbourhood cell’s building.
作者
罗林
LUO Lin（Guangzhou University Huaruan Software College, Guangzhou 510990, China）
出处
《电脑知识与技术》
2015年第10期127-128,138,共3页
Computer Knowledge and Technology
