Geomagnetic navigation has become popular for autonomous navigation method with features such as autonomous, all-weather, no time-accumulated error and so on. Its accuracy mainly depends on the accuracy of geomagnetic matching algorithms. Pedestrian Dead Reckoning technology is a positioning technology that calculates the relative position of pedestrians based on sensor information but can only obtain relative position information. According to the advantages and disadvantages of the two technologies, this paper proposes a high precision indoor positioning method which uses a smartphone as a hardware platform to build a magnetometer sensor model. An improved particle filter algorithm is used to solve the problem of geomagnetic fingerprint's fuzzy solution. The mean square error criterion establishes a matching trajectory and iterative calculations achieve real-time correction of PDR cumulative error. Finally, simulation experiments are performed. The experimental results show that the fusion location algorithm proposed in this paper is 42% higher than the PDR algorithm. Compared to a single geomagnetic fingerprint matching algorithm, positioning accuracy increased by 57%.