Research On Visible Light Indoor Positioning Technology Based On Imaging Sensor

With the rapid development of Internet technology and the gradual expansion of the Internet of Things market,location service based on location recognition plays an increasingly important role in people's lives and work.Compared with other indoor positioning technologies,visible light positioning(VLP)technology has the advantages of high positioning accuracy,no electromagnetic interference,and low cost.In this paper,VLP technology based on image sensor(IS)has been deeply studied from the following aspects,the designing of codeword and encoding,the detection and recognition of image,the accuracy of the positioning algorithm and other issues,also,an experimental platform is built to verify the proposed method.The main research contents and innovations are as follows:1)An improved Manchester encoding(ME)method with a large duty cycle is presented,which solves the problems of low average light intensity of LED beacons and uneven brightness among LED beacons after LED beacons are driven by the traditional encoding methods.A codeword structure composed of one start bit,six-character pairs and one stop bit is designed,which solves the problems of less address,too long codeword structure and inconvenient expansion that can be represented by the traditional LED-ID design method.The experimental results show that the codeword which based on an improved ME method with a 90% duty cycle can maximize the average light intensity coefficient and reach 0.8 on the basis of ensuring the decoding success rate,which is 60% higher than that before the improvement and proves the advancement of encoding and codeword design.2)A normalized pixel fringe value method is proposed to convert irregular light fringes into standard one-dimensional fringes for decoding algorithms,which solves the problem that most decoding algorithms have a low success rate in resolving LED-ID by using one column or one rectangular pixel in the LED region of interest(LED-ROI)image.The experimental results show that the decoding success rate is higher than 84% when the horizontal distance between LED beacons and IS is greater than 150 cm but less than 240 cm.3)A monocular vision positioning algorithm based on double LED beacons is proposed,and the Perspective n-Point(PnP)algorithm is innovatively used in the positioning system with two LED beacons,which solves the problems that additional sensors or marker points are required to assist and the receiver is not allowed to tilt when positioning based on double LED beacons.The positioning process is divided into two steps: feature point extraction and pose solving.In the first step,15 feature points are extracted from the image with double LED beacons by using the proposed multi-feature point extraction based on double LED beacons cooperation(MPDL)method;in the second step,PnP and Iterative Closest Point(ICP)algorithm are used to obtain 3D coordinates.A 190 cm × 190 cm × 230 cm indoor VLP experimental platform is built to verify the proposed algorithm.In the static case,when the vertical height difference between the receiver and the transmitter is 170 cm,200cm,and230 cm,the average positioning error is 5.3cm,7.5cm,and 8.8cm;in the dynamic case,when the vertical height difference between the receiver and the transmitter is 220 cm,the average positioning error is less than 10 cm,which proves the effectiveness of the positioning algorithm.