Seamline optimization based on ground objects classes for orthoimage mosaicking

Seamline determination is one of the key steps in orthoimage mosaicking. To avoid discontinuity in mosaic, this letter presents a seamline optimization method based on ground objects classes. The presented method classifies ground objects in overlaps of adjacent images into three classes, that is, buildings, vegetation and other classes, using normalized difference vegetation index and morphological building index. Based on ground objects classes, three types of regions, that is, obstacle regions, preferred regions and general regions, are further formed. Then, each type of region is assigned a different weight to optimize obtained pixel costs. Finally, a path finding with limiting level of maximum cost and differential cost is carried out to search the shortest path from start point to end point based on optimized pixel costs in overlaps as the final seamline. Experimental results including comparison with other methods demonstrate the potential of the presented method.