Abstract: For the digital surface model data obtained by image dense matching, an intelligent seamline extraction algorithm considering the topographic changes in the overlapping area is proposed. This method uses the elevation deviation of the DSM data overlapping area to construct the difference images. It is fusion processing between difference images and the difference images after denoising, stretching and reverse calculation. For the fusion data, the optimal path search method based on the minimum spanning tree is used to extract the optimal seamline. The experimental results show that our seamline extracted can avoid the protruding ground objects effectively with large elevation differences and ensure the integrity of the ground objects. This method can solve the problem of automatic seamline selection in the process of mosaicking DSM data. The method effectively reduces the errors brought by the traditional mosaic method, and has high reliability and stability.

Key words: digital surface model, intelligent seamline extraction, topographic changes, minimum spanning tree, path search