Automated Seamline Detection in Orthophoto Mosaicking Using Improved Snakes

When mosaicking two adjacent orthophotos, the transition from one orthophoto to the other is cannot be seen. The requirements of such an optimal seamline include high color similarity and high texture similarity. In this paper, a novel method integrated snakes and bresenham is presented. The calculation of photometric energy in the snake model considers not only the discrete points, but all the points that the snake passes which found by bresenham. This solves the local optimum problem existed in the snake model to some extent. In a hierarchical strategy, a proper seamline is delineated fully automatically. The experiments carried out in this paper show the promising potential of this schema.