With the rapid advancement of Vehicle Navigation and Intelligent Driving techniques, vehicle trajectories are vital for applications in navigation systems, city management systems, and intelligent driving systems. However, vehicle trajectory data contains sensitive user information, and directly releasing raw trajectory data can lead to privacy breaches. Nevertheless, current synthetic trajectory generation schemes fail to adequately account for Points of Interest (POIs) when extracting feature distributions from original trajectories, makes the generated synthetic trajectories vulnerable to location inference attacks by adversaries. To address this issue, this paper proposes a synthetic trajectory generation method designed to resist location inference attacks. The specific contributions are as follows: First, a privacy-preserving method is introduced to construct interest region combinations that satisfy K-anonymity, effectively mitigating location inference attacks. Second, the road network structure is incorporated into the synthetic trajectory framework, utilizing a quadtree-based grid partitioning approach to capture the movement patterns of the original trajectories. Finally, synthetic trajectories are generated by extracting the feature distribution of trajectory datasets that satisfy K-anonymity in the interest region combinations. Security analysis demonstrates that the proposed scheme can effectively resist location inference attacks. Experiments conducted on real-world datasets show that the proposed method significantly outperforms other comparative methods in terms of utility.