Continuous space ant colony algorithm for automatic selection of orthophoto mosaic seamline network

This paper proposes a novel method for selecting orthophoto mosaic seamline network using a continuous space ant colony algorithm. The method first establishes a cost function based on gray differences and a gray gradient in the overlapping area. Then, a continuous space ant colony algorithm is established to iteratively select a least cost path for the optimal mosaic seamline. This proposed algorithm is carried out in three steps: (1) generate the initial mosaic network using area Voronoi diagrams with overlap; (2) adjust the vertices (The intersection of mosaic seamline network); and (3) search the seamlines using the continuous space ant colony algorithm to obtain the globally optimal mosaic seamline network. Two groups of experiments were performed using different numbers of orthophotos. The results demonstrate that our method saves 50 s and 53 s compared with the commercial software OrthoVista in Dataset 1 and Dataset 2, and saves 19 s and 22 s compared with Dijkstra in Dataset 1 and Dataset 2, respectively. The average cost of the mosaic seamline network generated by our method is the lowest (15.56) out of OrthoVista (21.79), Dijkstra (18.35) and Qmosaic (18.68) in Dataset 1, and the lowest (19.28) out of OrthoVista (26.33), Dijkstra (22.60) and Qmosaic (23.56) in Dataset 2.

Automated summary

This paper proposes a novel method for selecting a mosaic seamline network in orthophotos using a continuous space ant colony algorithm. Experiments demonstrate that the proposed method achieves better performance compared to existing commercial software and algorithms.