Mutual Positioning Method in Unknown Indoor Environment Based on Visual Image Semantics

In our society, realizing intelligent positioning in indoor environments is important to build a smart city. Currently, mutual requirements the unknown environment are growing fast. However, such environment, we can obtain neither outdoor radio signal nor images advance for online positioning. Therefore, how achieve becomes an interesting problem. this paper, propose vision-based method environment. First, two users take of use semantic segmentation network identify targets contained images, and upload generated sequence user shared database real time. Then, every time reupload due change location, it necessary retrieve whether another has uploaded same database. If retrieval successful, means that have seen scene. Finally, select target from taken based on scene establish three-dimensional coordinate system, respectively, calculate their own position coordinates realize through sharing. Experiment results show proposed successfully between while ensuring high accuracy.