Scene matching is one of the important navigation methods in aircraft navigation. The difference in scale between the aerial image and the reference map will affect the final matching accuracy during the process of scene matching. In practice, the method of up-sampling is used to improve the final accuracy, but it is not accurate that the higher the up-sampling multiple, the better the effect. In this paper, the effect of the difference between the scale of aerial image and reference map on the final positioning accuracy is studied. First, the aerial images are up-sampled at different multiples. The SURF feature points of the aerial images and the reference map are extracted, and the optimal matching point pairs are selected by the RANSAC algorithm. Then, a multi-scale method for solving aircraft positioning is proposed. Considering the positioning accuracy and calculation amount comprehensively, the multiple of up-sampling 4 times is the optimal result.