计算机与现代化JISUANJIYUXIANDAIHUA2023年第1期总第329期摘要：针对沙漠、戈壁等特征不显著场景在配准过程中存在快速性、精确性等问题，提出一种基于局部重叠区域的特征不显著图像配准方法。首先利用图像标记对待配准图像进行预处理增强其特征，接着通过多相机三维投影对多幅待配准图像重叠区域进行预算，并采用图像掩膜和图像分割技术将重叠区域分割出来，最后对重叠区域使用ORB+GMS（Ori⁃entedBrief-Grid-basedMotionStatisticsforFast）融合算法进行配准，完成多幅图像的配准工作。基于图像重叠区域的配准避免了无显著特征图像在进行整体配准时精确性低的缺点，并且由于是局部配准，相较于全局配准拥有更快的配准速度。对比传统配准方法和本文提出的改进配准方法，实验结果显示，本文提出的改进方法配准精度在传统配准方法的基础上提升了28%，同时，算法具有更高的鲁棒性和实时性。关键词：特征匹配；无显著特征；图像标记；ORB+GMS融合算法中图分类号：TP391文献标志码：ADOI：10.3969/j.issn.1006-2475.2023.01.005Non-saliencyFeatureImageRegistrationAlgorithmBasedonLocalOverlappingRegionYANGXu-zhao，LEIZhi-yong，WANGJiao-jiao（SchoolofElectronicInformationEngineering,Xi’anTechnologyUniversity,Xi’an710021,China）Abstract：Aimingattheproblemsofrapidityandaccuracyintheregistrationprocessofsceneswithnon-saliencyfeaturessuchasdesertandGobi，thispaperproposesanon-saliencyfeatureimageregistrationmethodbasedonlocaloverlappingregion.Firstly，theimagetoberegisteredispreprocessedbyusingtheimagemarktoenhanceitsfeatures，thentheoverlappingareaofmultipleimagestoberegisteredisbudgetedthroughmulticamerathree-dimensionalprojection，andtheoverlappingareaissegmentedbyusingtheimagemaskandtheimagesegmentationtechnology.Finally，theoverlappingareaisregisteredbyORB+GMS（OrientedBrief-Grid-basedMotionStatisticsforFast）fusionalgorithmtocompletetheregistrationofmultipleimages.Theregistrationbasedontheimageoverlappingregionavoidsthedisadvantageoflowaccuracyintheoverallregistrationofimageswithnon-saliencyfea⁃tures.Duetothelocalregistration，ithasafasterregistrationspeedthantheglobalregistration.Comparedthetraditionalregistra⁃tionmethodwiththeimprovedregis...
