In this paper, a LiDAR-based depth estimation compensated VIO framework is proposed to address the problem of inaccurate depth estimation of UAV-mounted VIOs in high altitude, which in turn leads to reduced state estimation accuracy. In this framework, LiDAR uses accurate one-dimensional distance measurement information with external reference information between LiDAR and the camera instead of inter-frame constraint information, which avoids the problem of inaccurate feature depth estimation by triangulation in high altitude state. Meanwhile, this paper also adds the LiDAR measurement information into the back-end optimization, which achieves the simultaneous optimization of the LiDAR measurement results with the external parameters and improves the robustness of the system. Finally, in the flight experiments with maximum flight altitude of about 40 m and 100 m and travel distance of about 720 m and 1056 m, the proposed method reduces the absolute position error by 52% and 66%, respectively, compared with VINS-MONO.