A hybrid image retrieval method based on WiFi

The invention relates to a hybrid image retrieval method based on WiFi. The invention solves the problem that the image retrieval speed is slown and time-consuming when the traditional positioning method is used. The invention comprises the following steps: 1. selecting an indoor scene and establishing an indoor scene coordinate system; 2, selecting data sampling point at equal intervals and recording geographic position information of that sampling point; 3, the indoor scene images taken at each sampling point and the RSS values being stored in a database one-to-one correspondence; 4, selecting K images similar to that image to be matched as a coarse match result; 5, carrying out feature point matching on K image selected by rough matching and images to be matched; 6, erroneously match feature point pairs being eliminated; 7, counting the number of SURF local characteristic point pair correctly matched between the image to be matched and the K images; 8, selecting the database image with the largest numb of