Abstract: Aerial image has become a practical visualization product and decision-making tool because of its rich geographical information. Usually, it is necessary to splicing multiple aerial images along the seamline to cover the research areas. So how to select the optimal seamline is of research significance. This paper presents a  new method for selecting aerial image seamlines: firstly, an exponential cross-correlation method is constructed to express the dislocation degree of the corresponding area of the aerial images with high quality; secondly, the Dijkstra algorithm is optimized to find the path with the least dislocation difference in the overlapping region as the optimal seamline; finally, two sets of aerial images from the designated cities and urban areas are used to compare with the  related methods. The results show that the seamline obtained by the proposed method passes through fewer independent ground objects, and it takes less time. 

Key words: exponential cross correlation, mismatch expression function, seamline determination, aerial orthoimage, optimized Dijkstra’s algorithm