CAJ | 학술논문

镶嵌线自动搜索是实现影像重叠区域无缝拼接的关键步骤，而目前随着无人机技术的日益成熟及无人机能够快速获取高清的遥感图像数据，迫切需要寻找一种既有质量、又有效率的镶嵌线自动搜索的方法。本文提出一种基于形态学与Dijkstra相结合的影像镶嵌线方法，该方法首先确定两幅影像重叠部分的差分影像，然后在差分影像的基础上进行形态学膨胀处理，同时根据稀疏矩阵构建八邻域稀疏矩阵，最后使用Dijkstra算法在差分影像上进行镶嵌线的自动搜索，得到最优路径镶嵌线。实验结果证明，改进算法与未经过形态学处理的Dijk-stra方法相比，其自动搜索镶嵌线过程在耗时少（耗时为3．72 s）的情况下，能够很好地避开房屋等高亮度区。
양감선자동수색시실현영상중첩구역무봉병접적관건보취，이목전수착무인궤기술적일익성숙급무인궤능구쾌속획취고청적요감도상수거，박절수요심조일충기유질량、우유효솔적양감선자동수색적방법。본문제출일충기우형태학여Dijkstra상결합적영상양감선방법，해방법수선학정량폭영상중첩부분적차분영상，연후재차분영상적기출상진행형태학팽창처리，동시근거희소구진구건팔린역희소구진，최후사용Dijkstra산법재차분영상상진행양감선적자동수색，득도최우로경양감선。실험결과증명，개진산법여미경과형태학처리적Dijk-stra방법상비，기자동수색양감선과정재모시소（모시위3．72 s）적정황하，능구흔호지피개방옥등고량도구。
Automatic mosaic seamline selection is the critical process of achieving seamless mosaic of images overlapping areas .Also, we easily, with the growing maturity of Unmanned Aerial Vehicle (Called UAV),obtain high resolution images.Therefore, we should have to search a method which includes not only quality but also efficient of automatic selection of seamline becomes an urgent need . This article presents an improved approach based on the morphological and Dijkstra algorithm combination .The approach firstly deter-mines the differential image of both overlapping area , then the morphological expansion is applied on the differential image .Secondly , eight neighborhood sparse matrix is built on the differential image while according to sparse matrix theory .Finally, use Dijkstra algo-rithm to automatically search for seamline we get which is the optimal path seamline .Experimental results show that the improved ap-proach which compares with the Dijkstra approach without the morphological processing , it needs less time-consuming (takes 3.72 s) and it can effectively avoid traversing high brightness areas such as building , etc.