Design And Implementation Of Visual Positioning System Based On Hierarchical Feature Network

Augmented Reality(AR)fuses virtual information into reality.It makes the virtual information consistent with reality on terms of illumination,geometry and temporality.The technology enhances one’s perception about the real world.The location of mobile devices in reality can achieve geometric consistency of augmented reality technology.Visual localization technology has the advantages of high precision.It achieves real-time localization of mobile devices.However,it is difficult to locate accurately under the condition of strong illumination change.To solve this problem,the work of this paper is as follows:1.In order to solve the problem of retrieval failure and slow retrieval speed,the paper proposes a visual localization method using dynamic traversal.Firstly,it performs image preclustering based on the scene map.Secondly,it uses the image global descriptors to obtain the candidate frames and dynamically traverses the query image.Finally,the local feature descriptors are used to match the feature points,and the Perspective-n-Point algorithm is used to estimate the camera pose of query image.Experimental results show that it can effectively extract global descriptors and local features under the condition of strong illumination changes.2.Hierarchical feature network needs a lot of computation.In this paper,the hierarchical feature network is deployed on the cloud server.The cloud server estimates the pose of device and builds the map,the mobile terminal shows motion trajectories.The paper designs and implements a visual localization system based on hierarchical feature network.The test results show that the algorithm is designed reasonably,and the system works well.Therefore,the proposed system has theoretical significance and application value.