【作者】 张汉松； 陈建裕； 侯淑涛； 霍奕任；

【Author】 ZHANG Han-song;CHEN Jian-yu;HOU Shu-tao;Huo Yi-ren;College of Resources and Environment,Northeast Agricultural University;The State Key Laboratory of Satellite Ocean Environment Dynamics,The Second Institute of Oceanography,State Oceanic Administration;

【机构】 东北农业大学资源与环境学院； 国家海洋局第二海洋研究所卫星海洋环境动力学国家重点实验室；

【摘要】 针对复杂背景下遥感影像镶嵌中快速自动生成均匀分隔重叠区域且形态美观的几何拼接线,提出了一种基于图像解析的沿重叠区域骨架的几何拼接线自动生成算法。首先,顾及复杂背景值区域标记待镶嵌影像的不同类型区域;然后每一处待处理重叠区域被认为是一个图像对象,迭代搜索图像对象的轮廓线,并按指定规则对轮廓线逐像素标记,直至重叠区域中不存在这样的待处理区域;最后基于4邻域搜索标记沿骨架的几何拼接线。实验表明,即使顾及背景值影像重叠区域为复杂多边形时,该算法也能快速自动生成美观的几何拼接线。算法鲁棒性优于基于图形解析的方法。更多还原

【Abstract】 In response to geometry seamline detection with average separation of overlapped region and beautiful shape for Remote Sensing images mosaicking considering complex background, a fast automated geometry seamline generation approach along skeleton is proposed on image analysis. Firstly, different kinds of region for image mosaicking are marked in view of background value. Secondly, each block overlapped region to be processed is considered as one image object, the footprints of these image objects are detected and re-marked following specific rules pixel by pixel until the overlapped regions to be processed don’t exist. Finally, the geometry seamline along skeleton is marked on four neighborhood search. The experimental result shows that even if the overlapped region is complex polygon considering background value, the fine seamline can be quickly automatically selected. The algorithm robustness is better than those algorithms on graphic analysis.更多还原