一种改进的最佳缝合线算法

An improved optimal seam algorithm

当待拼接的源图像之间存在较大色彩差异时,最佳缝合线算法的输出图像中会存在明显的接缝,影响图像质量。针对这一问题,引入多频段融合算法,提出了一种最佳缝合线的改进算法。算法采用多尺度分解方法,将图像扩展至不同尺度空间,在不同的尺度空间内采用不同权值进行图像融合,重构还原得到最终的拼接图像,并建立主客观两种评价体系,验证算法的有效性。实验表明,相较于传统最佳缝合线算法,本文提出的算法能过够有效地解决接缝问题,输出的拼接图像在质量上有着明显的提升。

When there are large color differences among the source images to be seamed,there will be obvious seams in the output image of the optimal seam algorithm,which will affect the image quality.In order to solve this problem,an improved optimal seam algorithm is proposed based on the introduction of the multi-band fusion algorithm.The algorithm employs multi-scale decomposition method to expand the image to different scale space,and uses different weights to fuse the image in different scale space.And then,the final mosaic image can be reconstructed and returned to the original condition by the image pyramid.The experiment shows that compared with traditional optimal seam algorithm,the improved algorithm can solve the seam problem existing in the previous algorithms effectively and the output mosaic image has significant improvement in quality.