基于改进SIFT算法的视频图像序列自动拼接Automatic video mosaic imaging based on improved SIFT algorithm
卢斌,宋夫华
摘要(Abstract)：
本文提出了一种自适应帧采样和限定特征提取区域的拼接方法,根据帧间重叠率和帧间隔建立线性模型,并把各帧图像对准到其前后的关键帧上。在特征点提取方面,提出了一种改进的SIFT算法进行特征点提取,并采用随机采用一致性(RANSAC)方法来更新匹配点,在图像融合中采用线性加权渐入渐出的自然融合算法。实验结果表明:该方法对一般场景能稳定的抽取到关键帧,并进行拼接,取得了较好的拼接效果。
关键词(KeyWords)： 视频图像;重叠率;SIFT;图像融合
基金项目(Foundation):
作者(Author): 卢斌,宋夫华
DOI: 10.16251/j.cnki.1009-2307.2013.01.013
参考文献(References)：
- [1]周天宏,张建清.基于图像映射法的快速无缝拼接技术[J].武汉理工大学学报,2008,30(7):131-133.
- [2]Matthew B,David L.Automatic panoramic image stitch-ingusing invariant features[J].International Journal of-Computer Vision,2007,74(1):59-73.
- [3]Mikolajczyk K,Schmid C.A performance evaluation of lo-cal descriptors[J].IEEE Trans Pattern Ana and MachineIntelligence,2005.
- [4]Steedly D,Pal C,Szeliski R.Efficiently register-ing vid-eo into panoramic mosaics[C]//Tenth International Con-ference on Computer Vision(ICCV).Washington,DC,USA:IEEE Computer Society,2005:1300-1307
- [5]Maitra M,Chatterjee A.Hybrid multiresolution Slantlettransform and fuzzy cmeans clustering approach for normal-pathological brain MR image segregation[J].MedicalEngineering&Physics,2008,30(5):615-623.
- [6]C’AMARA-CH’AVEZ G,De ALBU QU ERQUE ARA’UJO A.Harris-SIFT descript or for video event detectionbased on a machine learning approach[C]//IEEE In-ternational Sympo sium on Multimedia.USA:IEEE Com-puter Society,2009:153-158.
- [7]冯宇平.图像快速配准与自动拼接技术研究[D].长春:长春光学精密机械与物理研究所,2010.
- [8]严大勤,孙鑫.一种基于区域匹配的图像拼接算法[J].仪器仪表学报,2006,27(S1):749-750.
- [9]Forssen P,Lowe D.Shape descriptors for maximally stableextremal regions[C]//IEEE International Conference onComputer Vision,2007.
- [10]Brown M,Lowe D.Automatic panoramic image stitching u-sing invariant features[J].International Journal of Com-puter Vision,2007,74(1):59-73.
- [11]PING GUAN,XINRONG HU,YI ZHUANG.Research onDigital Image Mosaic[C]∥International Conference onCom-puter and Communication Technologies in Agricul-ture Engineering.[S.l.]:CCTAE,2010:67-69.
- [12]于仕琪,刘瑞祯.学习OpenCV.[M].北京:清华大学出版社,2009.