The Improvement Of Visual Position Matching Algorithm Base On Inertial Navigation

Location has become the important need of people's daily life.Nowadays the Global Position System(GPS)has been able to provide the accurate and reliable location outside.But inside the house,the GPS does not perform well due to shelter from buildings.Therefore,the research on indoor positioning has been hot in recent years.Because of the popularity of indoor surveillance,high precision and not subject to electromagnetic interference,the positioning methods based on computer visual have better application foundation compared with other ways.There is a problem that the position information must match with someone after computer visual positioning.If the match is false,it will cause bad influence to the positioning precision.Teixeira algorithm works best now,but its performance is not good when there is large pedestrian density.Matching algorithms under the crowded situation is studied in this thesis.Matching method based on the pedestrian's direction is proposed.In the method,the matched coordinate points are deleted from the original path coordinates set,whichincreases the matching rate by eliminating the influence of these matched points for the remaining coordinate points.The line acceleration of the pedestrian which reflects the real movement direction is extracted,to provide more accurate direction information and basis for the path confusing problem.Pitching angle,flip angle and azimuth angle are obtained through the distribution of the gravity acceleration in the accelerometer coordinate system and magnetometer.And the linear acceleration obtained from the accelerometer is transfered to the world coordinate system based on the real-time mobile position,which makes the pose of the mobile phone have no effect on direction extraction.The pedestrian can carry the phone with any pose,that improves the practicability of the method.The experimental results show that the matching algorithm of this thesis,compared with Teixeira algorithm,the matching rate in 3s rises from 80%to 96%,and if the matching rate reaches 100%,time reduces from 9s to 7s in the condition of 10m×10m area and 10 pedestrians.Therefore,the algorithm proposed in this thesis has higher matching rate and shorter matching time.