基于拉普拉斯金字塔融合的岩心图像拼接算法①徐圣滨1,李立华2,何小海1,卿粼波1,陈洪刚11(四川大学电子信息学院,成都610065)2(河北省地质矿产勘查开发局第六地质大队,石家庄050086)通信作者:陈洪刚,E-mail:honggang_chen@scu.edu.cn摘要:针对岩心图像拼接效率低以及易出现鬼影现象的问题,提出了一种基于最佳缝合线的拉普拉斯金字塔融合的岩心图像拼接方法.首先将待拼接的两幅岩心图像进行灰度变换,根据ORB算法计算并描述特征点;其次使用改进的randomsampleconsensus(RANSAC)算法对特征点进行提纯,完成特征点匹配;根据匹配的特征点计算图像间的配准关系,最后根据最佳缝合线实现岩心图像的拉普拉斯金字塔融合,完成拼接.实验结果表明,改进的RANSAC算法能在保证正确率的同时提升速度,而且本文提出的图像融合方法避免了鬼影的产生,在融合区域的PSNR、SSIM和DoEM客观评价指标上与另外两种图像融合算法相比都有所提升.关键词:岩心图像;RANSAC;最佳缝合线;图像融合;图像拼接引用格式:徐圣滨,李立华,何小海,卿粼波,陈洪刚.基于拉普拉斯金字塔融合的岩心图像拼接算法.计算机系统应用,2023,32(2):316–321.http://www.c-s-a.org.cn/1003-3254/8979.htmlCoreImageStitchingAlgorithmBasedonLaplacianPyramidFusionXUSheng-Bin1,LILi-Hua2,HEXiao-Hai1,QINGLin-Bo1,CHENHong-Gang11(CollegeofElectronicsandInformationEngineering,SichuanUniversity,Chengdu610065,China)2(TheSixthGeologicalBrigadeofHebeiBureauofGeologyandMineralExplorationandDevelopment,Shijiazhuang050086,China)Abstract:AcoreimagestitchingmethodbasedonLaplacianpyramidfusionwiththebestseam-lineisproposedtoaddresstheproblemsoflowcoreimagestitchingefficiencyandthetendencyofghosting.Firstly,twocoreimagestobestitchedareprocessedthroughgrey-leveltransformation,andthenfeaturepointsarecalculatedanddescribedaccordingtotheORBalgorithm.Secondly,theimprovedrandomsampleconsensus(RANSAC)algorithmisusedtopurifythefeaturepointsandcompletefeaturepointmatching.Accordingtothematchedfeaturepoints,thealignmentrelationshipbetweentheimagesiscalculated.Finally,theLaplacianpyramidfusionofthecoreimagesisrealizedbasedonthebestseam-line,andthestitchingiscompleted.TheexperimentalresultsshowthattheimprovedRANSACalgorithmcanimprovethespeedwhileen...
