Research On High Precision 3D Indoor Positioning Algorithm Based On VLC

In recent years,indoor positioning has attracted widespread attention for the rapid development of location services.Visible Light Positioning(VLP)technology uses lightemitting diode(LED)as signal sources for positioning,which has the advantages of no electromagnetic interference,environmental protection,and simultaneous lighting.VLP technology is worth studying and has good application prospects in the field of indoor positioning.Based on the investigation of VLP algorithms,this paper focuses on two common receivers—photodiode(PD)and image sensor(IS)from the following aspects: high-precision and real-time positioning algorithm,LED detection and identification(ID)recognition algorithm,etc.And experimental platforms are separately constructed to verify the proposed algorithms.The main research contents are as follows:(1)Aiming at the optimization of the PD-based VLP algorithm,a VLP system using an improved bacterial colony chemotaxis(IBCC)algorithm is proposed.First,a fitness function based on the geometric relationship between the LED projection circles is proposed,and the3 D positioning problem is transformed into a global optimization problem where the search variable is the Z coordinate of the receiver.Then,the IBCC algorithm is used to quickly search the Z coordinate of the receiver.Next,the X and Y coordinates of the receiver are obtained through the trilateral measurement equations.Finally,the simulation results verify the accuracy and real-time performance of the proposed algorithm.(2)Aiming at the problems of LED detection and ID recognition in IS-based VLP systems,a LED-ROI detection algorithm based on least squares fitting and a LED-ID fast recognition algorithm based on online sequential extreme learning machine(OS-ELM)are proposed.First,obtain the edge points of the LED by image processing techniques,and use the least squares to fit the center point and radius of the region of interest of LED(LED-ROI)to complete the detection.Then,extract the area of the LED-ROI,the number of bright stripes,and the ratio of the width of the bright and dark stripes as LED features,use OS-ELM for ID classifier training to complete fast ID recognition.Finally,the experimental results validate the effectiveness and superiority of the OS-ELM algorithm in LED-ID recognition.(3)Aiming at the static positioning problem in IS-based VLP systems,a static positioning algorithm based on fast Levenberg-Marquardt(FLM)is proposed.First,LED-ROI detection and ID recognition are performed to obtain the 2D-3D matched points.Then,the initial estimation of the pose is quickly obtained through direct linear transformation(DLT).Next,the least square method is established to optimize the reprojection error of the observation points,and the Levenberg-Marquardt algorithm is used to quickly solve the least square problem to get the optimal pose of the camera.Finally,the experimental results validate the accuracy and reliability of the proposed static positioning algorithm.(4)Aiming at the problem of the dynamic positioning problem in IS-based VLP systems,a dynamic positioning algorithm based on improved kernelized correlation filters(IKCF)and fast weighted Levenberg-Marquardt(FWLM)algorithms is proposed.First,the LED-ID recognition is performed in the first frame.Then,the IKCF algorithm is used to track the LEDROI,and an adaptive model update mechanism is used to avoid the model drift.Next,the FWLM algorithm is used for pose estimation,and a weighting matrix is used to reduce the weights of low-reliability LEDs in the least square problem.Finally,the experimental results validate the accuracy,real-time and robustness of the proposed dynamic positioning algorithm.