Research On Key Technologies Of Indoor Positioning Based On Inertial Navigation Data And Image Landmarks

As the development of science technology and the arrival of the artificial intelligence era,the indoor positioning becomes common application in people's life.Especially in the building without GPS signal and the underground parking lot,people always confused about their location.In this paper,the key techniques of indoor positioning based on inertial navigation data and image landmarks are mainly studied.The research on indoor positioning in this paper mainly focuses on the following two aspects.First,plan the road map of the building structure based on inertial navigation data.In path planning,this paper improves the heading direction based on the PDR algorithm for pedestrian tracking.The method of k-means clustering is used to cluster headings.According to the characteristics of the building structure,an adaptive k-means clustering algorithm for the angle in the polar coordinate system confirms the heading.Based on this,a simple clustering of the altitude is used to confirm the floor distribution for the altitude distribution characteristics.The simulation results for inertial navigation data indicate that two-dimensional and three-dimensional structural maps of buildings can be planned based on the inertial navigation data.Second,confirm the initial position based on the image landmark matching.In this way,a positioning method that combines inertial navigation data and image landmarks is implemented.In image landmark matching,the image landmarks are defined as the key points of the building structure,and the image landmarks are matched by using the idea of secondary matching in positioning.The first time is based on SIFT feature vector matching,and the second time is based on text retrieval.Judging the similarity of the first matching number,if the similarity is low,you can directly do the image landmark positioning.If the similarity is high,do the second detection to locate the location of the image landmark.The image landmarks of the key points in the building are collected,and the simulation results indicate that the image landmarks match to locate the key points in the building.In the end,this paper collects the inertial navigation data and image landmark datasets of the building structure through the existing hardware system.The simulation is to structure the infographic of building structure information.When the user walks in the building again,the user position can be located according to the image landmark matching.In the following walking process,the distance and direction of pedestrian walking can be calculated based on the inertial navigation data.The indoor positioning method combined with the inertial navigation data and image landmarks is feasible.