Abstract. In a UAV system, physical sensor offsets occur between an observation sensor and a GPS/IMU sensor which represents the position of UAV. The difference of angle between the GPS/IMU sensor’s axis and the observation sensor’s axis is referred to as boresight angle. The difference in physical position between the two is referred to as lever-arm. It is important to obtain accurate offset values in order to utilize UAVs in rapid mapping manner. Due to the sensor offsets, misalignment error can be caused when generating a mosaic image. Offset values can be measured by using extensive ground control points. However, this is very costly and time-consuming. In this study, we describe an estimation method for sensor offsets using tie-points and re-weighted least square estimation method. The proposed method consists of 5 steps. Firstly, a frame images for the target area were classified into image strips based on the kappa value of initial EOPs (Exterior Orientation Parameters), after which tie points were extracted between adjacent images. Secondly, strip bundle adjustment was performed to update initial EOPs using images with the same flight direction. Thirdly, tie-points between adjacent strips were extracted. Fourthly, block bundle adjustment was performed in all images, using all extracted tie-points. Finally, a mosaic image is generated using the EOPs value to which the estimated sensor offset is applied. From this study, we confirmed that the sensor offset of the platform could be estimated only with the tie-points extracted between adjacent images. And we confirmed that misalignment was adjusted when generating mosaic image. We expect that our research makes UAV system to be operated more fluently.