Abstract: In order to assist blind people to travel independently by subway and solve the problems of hopping discontinuity of location points and poor real-time location in traditional indoor location methods, a method by fusing region and electronic map is proposed. The method can be divided into two stages: offline region classification and online location. In the stage of offline region classification, the best Gauss filtering template is searched for RSSI sequences generated from Bluetooth sensors by iterative refinement and the filtered sequence means are used to construct location fingerprint database, and the support vector machine model is used to classify the location fingerprint database at the first level. In the stage of online location, a sliding window is used to classify the regions to be located at the second level, a KNN algorithm based on Euclidean distance is used to calculate the location coordinates within the range of the sliding window, and the path layer data in a electronic map is used to reduce the deviation of location coordinates so as to further control the error margin and improve the efficiency of location. The tests in a subway station hall show that the filtering method can improve the accuracy of location by nearly 4% and finally the accuracy of location can reach 1.59 m by using this location algorithm.