High-precision positioning of the ground target is an important premise for the UAV (unmanned aerial vehicle) to carry out target reconnaissance, firepower guidance, effectiveness evaluation, etc. However, the accuracy of the UAV′ s target positioning is limited by factors such as many error factors and long transmission chain. The method of multi-frame image registration method based on Kalman filter for ground target location was studied. By combining the multi-frame target image acquired by UAV, the high-precision fusion positioning of UAV to ground target was studied based on Kalman filter. Monte Carlo method was introduced to simulate the error convergence, value and distribution of multi-frame fusion location method based on Kalman filter. The influence of observation interval and line-of-sight elevation angle on error convergence was analyzed. Several suggestions for improving the positioning accuracy were proposed.