Inertial based pedestrian positioning is very useful in emergency response situations, such as search and rescue (SAR) missions as it is independent of external infrastructures and signals. Two practical problems emerge in inertial based pedestrian positioning: trajectory initialization and trajectory calibration. The former one denotes how to determine the initial position and heading of the pedestrian, and the latter denotes how to reduce the dead reckoning (DR) errors due to inertial drifts. This paper provides a solution to these two problems adopting a few ultra-wide band (UWB) anchors, which can provide range measurements between the anchors and the pedestrian. The foundation of the method is graph based simultaneously localization and mapping (SLAM). The nodes in the graph are the poses of the pedestrian and the edges are the range measurements between the UWB anchors and the tags mounted on the pedestrian. Both simulations and real scenario experiments are carried out to validate the proposed method. The proposed approach outperformed the particle filtering based approach in terms of robustness and processing time.