This paper proposes a method of image seam line based on the combination of Dijkstra algorithm and morphology. First, the method determines the differential images of the overlap part of two images. And then, morphological expansion processing is implemented based on the differential images and the eight-neighborhood sparse matrix is constructed according to the sparse matrix. Finally, the seam line with the optimal path can be obtained by automatic search of seam line on the differential images by Dijkstra algorithm. The experimental results show that compared to the Dijkstra method that is not subject to morphological processing, the improved algorithm can avoid the high brightness area as housing under the circumstance of less time (3.72s) in the process of automatic research seam line.