Indoor localization in complex environments is challenged by signal instability, environmental occlusion, and cumulative motion-estimation errors. To address these issues, this study proposes a fusion localization method that combines Wi-Fi fingerprint trajectories and pedestrian dead reckoning (PDR) trajectories in a library scene. A dual-device collaborative acquisition scheme is adopted, in which an ESP32-C5 board collects Wi-Fi data and a smartphone collects inertial and attitude data. Weighted K-nearest neighbor (WKNN) is used for Wi-Fi fingerprint positioning, while PDR trajectory propagation is derived from linear acceleration and a roll-derived directional signal for stepwise propagation. On this basis, a 9-dimensional augmented multi-step extended Kalman filter is constructed to fuse Wi-Fi and PDR information. Experiments were conducted in two walking scenarios, namely a U-shaped trajectory and an L-shaped trajectory, within the same library environment. Results show that the proposed fusion method consistently improves overall trajectory fitting and demonstrates good applicability under different trajectory geometries. Although the improvement in local point-to-point errors remains scenario-dependent, the proposed framework effectively combines Wi-Fi absolute position constraints with continuous PDR motion information, thereby improving trajectory localization performance in complex indoor environments.