In UAV vision-assisted inertial navigation system， the image data with uncertain delay cannot meet the synchronization requirements with other sensors in the UAV indoor navigation. Therefore， the distance between the vision sensor and the inertial measurement unit （IMU） is accurately estimated. Relative delay is very important. This paper proposes a method that can effectively estimate the image delay， and compensate the delay of the visual data according . Finally， it is capable to finally use the extended Kalman filter （EKF） to realize the fusion of the IMU data and the visual data to estimate the UAV Real-time pose and speed. The results of software simulation and experimental verification on the UAV platform demonstrate that the method can accurately estimate the time delay and significantly improve the positioning performance of indoor real-time navigation.