An Upscaling–Downscaling Optimal Seamline Detection Algorithm for Very Large Remote Sensing Image Mosaicking

For the mosaicking of multiple remote sensing images, obtaining optimal stitching line in overlapping region is a key step creating seamless mosaic image. However, for very large computation finding seamlines involves huge amount image pixels. To handle this issue, we propose stepwise strategy to obtain pixel-level lines images via an upscaling–downscaling sampling procedure. First, resolution reduced and graph cut algorithm applied find energy-optimal seamline Then, stripe along preliminary identified from overlap area remove other inefficient nodes. Finally, nested within seek original Compared existing algorithms, proposed method produces fewer spectral differences between less-crossed features experiments. wide range involving data, new uses less than 10 percent time needed by SLIC+ method.