The fingerprint samples are collected point-by-point in traditional fingerprint database construction approach, which is inefficient. In response to this problem, this paper proposes a database construction system based on fast acquisition of fingerprint samples and the Radial Basis Function (RBF). Firstly, several linear paths are marked in the indoor environment according to the determined sampling interval and sampling number. Secondly, according to the Pedestrian Dead Reckoning (PDR) algorithm, the coordinates are assigned to the Received Signal Strength (RSS) to form a sparse fingerprint database. Thirdly, we select the RBF to extend sparse fingerprint database, with the purpose of constructing a database with higher fingerprint granularity. Experimental results show that our proposed approach can be applied to reduce the labor cost in database construction, as well as guaranteeing high-enough accuracy performance of fingerprint database.