正交多幕特种电影的自适应拼接技术
|
摘要
本文分析了前人在图像拼接领域的研究成果和研究现状,介绍了图像拼接的流程、图像的采集以及常见的图像变换模型。重点研究了图像拼接技术中的图像配准及图像融合等关键技术。分析和总结了现有的图像拼接的主要方法、性能以及存在的问题。通过对各种拼接算法优缺点的比较,本文提出了一种基于特征的改进算法。该算法综合利用了兴趣点的亮度信息和空间信息,使用Harris角检测器提取出兴趣点,采用兴趣点主方向和互相关系数相结合的方法提取出初始匹配对,根据特征点的整体匹配度剔除伪匹配对,得到两帧图像的对应兴趣点特征对;根据匹配对计算出图像的变换参数;在原始图像上得到初始匹配点后采用RANSAC算法稳健估计变换矩阵H;为了提高配准算法的精度,采用变换矩阵H,引导两帧原始图像上的所有特征点重新进行匹配,对得到的匹配点集重新用RANSAC算法估计变换矩阵,并采用LM非线性优化算法进一步优化,最后通过基于最佳缝合线的方法进行融合图像。
把本文提出的图像配准和融合的改进算法结合在一起,通过实验分析,该算法准确率高,鲁棒性强,具有较高的使用价值。可以使多个投影机合成的图像构成一个整体,无缝隙,亮度均匀的图像。从而使观众体验到沉浸感超强的完美体验,是正交多幕特种电影生产制作过程中必不可少的重要环节。最后对本文还存在的问题进行总结并对图像拼接的发展前景进行展望。
The paper analyzes the effort of image mosaic that the researchers have achieved, and then the thesis discusses the basic steps of image mosaic, image acquisition methods and models of image mosaic. Moreover, it researches with emphasis in the two core technology of image mosaic, which are the image registration technology and the image fusion technology. The common image registration methods and image fusion methods are introduced, and their performance as well as the existing problems is analyzed. Through a variety of advantages and disadvantages of the splicing method of comparison, this paper, a feature-based algorithm. The intensity information and position information of interest point is utilized in this algorithm. Interest points are firstly extracted by the Harris angle detector in this method, and initial matches are extracted by the method of dominant orientation and cross correlation, then the false matches are rejected according to the matching degrees of feature points. Transform parameters are determined by the matches, and weighted average method is used to blend images. After getting initial matching points on the original image, use an RANSAC algorithm to estimate transformation matrix between two images; the transformation matrix is used to guide local point matching again, then combine RANSAC algorithm and Levenberg-Marquardt optimization to decrease registration error. Finally, based on the best methods of suture fusion image.
In this paper, the image registration and image integration to combine image mosaic for the experiment show that the algorithm accuracy, robustness and high value. So that the projector can be more than a synthetic image as a whole, seamless, uniform brightness of the image. So that viewers experience a sense of immersion into the super-perfect experience, is the special orthogonal multi-screen film production process essential to the production of an important link. Finally, this paper, there are still problems and sum up the image mosaic for the future development prospects.
把本文提出的图像配准和融合的改进算法结合在一起,通过实验分析,该算法准确率高,鲁棒性强,具有较高的使用价值。可以使多个投影机合成的图像构成一个整体,无缝隙,亮度均匀的图像。从而使观众体验到沉浸感超强的完美体验,是正交多幕特种电影生产制作过程中必不可少的重要环节。最后对本文还存在的问题进行总结并对图像拼接的发展前景进行展望。
The paper analyzes the effort of image mosaic that the researchers have achieved, and then the thesis discusses the basic steps of image mosaic, image acquisition methods and models of image mosaic. Moreover, it researches with emphasis in the two core technology of image mosaic, which are the image registration technology and the image fusion technology. The common image registration methods and image fusion methods are introduced, and their performance as well as the existing problems is analyzed. Through a variety of advantages and disadvantages of the splicing method of comparison, this paper, a feature-based algorithm. The intensity information and position information of interest point is utilized in this algorithm. Interest points are firstly extracted by the Harris angle detector in this method, and initial matches are extracted by the method of dominant orientation and cross correlation, then the false matches are rejected according to the matching degrees of feature points. Transform parameters are determined by the matches, and weighted average method is used to blend images. After getting initial matching points on the original image, use an RANSAC algorithm to estimate transformation matrix between two images; the transformation matrix is used to guide local point matching again, then combine RANSAC algorithm and Levenberg-Marquardt optimization to decrease registration error. Finally, based on the best methods of suture fusion image.
In this paper, the image registration and image integration to combine image mosaic for the experiment show that the algorithm accuracy, robustness and high value. So that the projector can be more than a synthetic image as a whole, seamless, uniform brightness of the image. So that viewers experience a sense of immersion into the super-perfect experience, is the special orthogonal multi-screen film production process essential to the production of an important link. Finally, this paper, there are still problems and sum up the image mosaic for the future development prospects.
引文
[1]Anandan P.IEEEE Workshop on Representations of Visual Scenes.Cambridge,Massacbusetts[J].IEEE Computer society Press,2002
[2]Szeliski R.Video Mosaics for Virtual Environments[J].IEEE Computer Graphics and Applications,1996,16(2):22-30
[3]Shmuel Peleg,Benny Rousso,Alex Rav-Acha,et al.Mosaicing on Adaptive Manifolds.IEEE Transactions on pattern and machine intelligence,2004,22(10):1144-1154
[4]C.Kuglin and D.Hines.The phase correlation image alignment method.In Proc IEEE Conference on Cybernetics and Society,New York,September1975:163-165.
[5]E.De Castro and C,Morandi.Registration of translated and rotated images using finite Fourier transforms,IEEE Trans.Pattern Analysis and Machine Inte Ⅱ igence,1987,vol.9(5):700-703
[6]B.Strinivasa Reddy and B.N.Chatterji An FFT-based technique for translation,rotation,and scale-invariant image registration[J].IEEE Transactions on Image Processing,1996,8(5):1226-1271
[7]陶波,于志伟.图像的自动拼接[J].中国生物医药工程学报,2006,16(4):316-322
[8]丁迎,洪继光.图像拼接中伪匹配的判断和消解[J].中国图像国形学报,2003,4A(10):886-890
[9]许雷,张恒义等.一种基于相位相关法及数学形态学方法的眼底血管图像自动拼接方法[J].生物医学工程杂志,2003,15(3):286-290
[10]Bao P,Xu D.Panoramic image mosaics via complex wavelet pyramid[C],Proceedings of ISSPR' 98,Hong Kong September1998:175-180
[11]王俊杰,刘家茂,胡运发等.图像拼接技术.计算机科学学报.2006,11(6):141-150
[12]潘立公,汤晓安.一种360度柱面全景图快速生成算法.研究与设计,2005,17(4):19-35
[13]杨枝灵,王开.Visual C++数字图像获取处理及实践应用.北京:人民邮电出版社,2003:100-360
[14]刘翔,曾芬芳.基于IBR披术的全景图生成.华东船舶工业学院学报(自然科学版).2005,15(1):32-37
[15]沈建国,戴永刚.基于局部熵差的栅格地图拼接方法.华东师范大学学报(自然科学版).2002,(4):48-53
[16]唐琏,谷士文,杨安平等.立方体表面上的图象拼接.小型微型计算机系统.2004,23(1):93-97
[17]兰培真,马越,邱志雄等.不同视点重叠图象拼接方法研究.中国图象图形学报,2002,7(10):1032-1037
[18]潘华伟,邹北骥.一种圆柱形全景图生成新算法及其实现.计算机工程与科学.2003,25(6):13-16
[19]漆驰,刘强,孙家广等.摄像机图像序列的全景图拼接.计算机辅助设计与图形学报,2003,13(7):605-609
[20]侯舒维,郭宝龙.一种图像自动拼接的快速算法[J].计算机工程,2005,31(24):70-72
[21]王俊杰.刘家茂.图像拼接技术.计算机科学,2004,(30):141-144
[22]朱云芳,叶秀清,顾伟康.视频序列的全景图拼接技术.中国图形图像学报,2006,11(8):198-203
[23]蒋苏蓉.王松,冯刚.基于图像的全景视图的实现披术.计算机应用,2005,22(6):86-87
[24]毛晓冬.图像配准与拼接方法研究:[硕士学位论文].西安:西安电子科技大学,2006
[25]于明言.一种基于时域和频域特征的图像拼接方法:[硕士学位论文].大连:大连理工大学,2006
[26]马龙.超声宽景成像技术研究:[硕士学位论文].西安:西安电子科技大学,2006
[27]高春霞.基于脉冲耦合神经网络和进化算法的图像分割方法研究:[硕士学位论文].西安:西安电子科技大学,2007
[28]陈爱萍.基于图像配准的高分辨率重构技术研究:[硕士学位论文].西安:西安电子科技大学,2007
[29]谭康.图像拼接技术研究与实现:[硕士学位论文].南京:南京理工大学计算机应用技术,2006
[30]叶朝阳.大视场图像拼接:原理与应用:[硕士学位论文].浙江:浙江大学计算机应用,2004
[31]张静,胡志萍,刘志泰等.基于轮廓相位相关的图像自动拼接.大连理工大学学报,2005,45(1):68-74
[32]张世阳.一种快速全景图像拼接技术.计算机应用与软件,2004,21(3),77-79
[33]张静.基于图像绘制技术的虚拟场景环视和纵向漫游:[博士学位论文].大连:大连理工大学,2004
[34]钟立,胡晓峰.重叠图像拼接算法[J].中国图象图形学报,2005,(5):22-37
[35]庄志国,孙惠军,董继扬等.基于角点检测的图像匹配算法及其在图像拼接中的应用.厦门大学学报(自然科学版),2007,7(4):501-508
[36]全红艳,张田文,蔺想红.全景图拼接中的一种快速图像点对应策略[J].机器人,2005,16(3):45-49
[37]Chalechale A,Naghdy G,Mertins A.Sketch-based image matching using angular partitioning systems [J].IEEE Trans on Man and Cybernetics,Part A,2005,35(1):28-41
[38]赵向阳,杜利民.一种全自动稳健的图像拼接融合算法[J].中国图形图像学报,2004,Vol(9):417-422
[39]Reddy B S,Chatterji B N.An FFT-based technique for translation,rotation,and scale invariant image registration[J].IEEE Trans on Image Processing,2006,5(8):1266-1271
[40]容观澳,计算机图象处理.北京:清华大学出版社,2000:80-280
[41]倪明田,吴良芝.计算机图形学.北京:北京大学出版社,1999:153-325
[42]S.Gumustekin,R.W.Hall.Mosaic image generation on a flattened Gaussian sphere[J].In Proc.Of IEEE Workshop on APPlications of Computer Vision,2006,50-55
[43]祝乃庆.全景图自动拼接算法研究及系统实现:[硕士学位论文].南京:南京理工大学,2004
[44]Han B,Lin X.A novel hybrid color regist ration algorithm for image stitching[J].IEEE Trans on Consumer Electronics,2006,52(3):1129-1134
[45]曹俊杰,封靖波,苏志勋.全景图像拼接算法[J].大连理工大学学报,2006,S1:187-189
[46]Richard Szeliski,Heung-Yeung Shum.Creating full view panornamic image mosaics and texture-mapped models[J].SIGGRAPH 97 Conference Proceedings,2005,3(1):251-258
[47]陈永强,王启付.虚拟环境中变形图像拼接技术研究.华中科技大学学报,2005,29(1):114-166
[48]Harris C,Stephens M.A combined corner and edge detector[C]//Proceedings of The Fourth Alvey Vision Conference.Manchester:[s.n],2005:147-151
[49]Moravec H.P.Towards Automatic Visual Obstacle Avoidnace[C].In 5th International Joint Conference on Artificial Intelligence,2004:584-589
[50]唐琏,谷士文.与全方位全景图像有关的几个问题的研究[J].中国铁道科学学报,2004,23(4):89-96
[2]Szeliski R.Video Mosaics for Virtual Environments[J].IEEE Computer Graphics and Applications,1996,16(2):22-30
[3]Shmuel Peleg,Benny Rousso,Alex Rav-Acha,et al.Mosaicing on Adaptive Manifolds.IEEE Transactions on pattern and machine intelligence,2004,22(10):1144-1154
[4]C.Kuglin and D.Hines.The phase correlation image alignment method.In Proc IEEE Conference on Cybernetics and Society,New York,September1975:163-165.
[5]E.De Castro and C,Morandi.Registration of translated and rotated images using finite Fourier transforms,IEEE Trans.Pattern Analysis and Machine Inte Ⅱ igence,1987,vol.9(5):700-703
[6]B.Strinivasa Reddy and B.N.Chatterji An FFT-based technique for translation,rotation,and scale-invariant image registration[J].IEEE Transactions on Image Processing,1996,8(5):1226-1271
[7]陶波,于志伟.图像的自动拼接[J].中国生物医药工程学报,2006,16(4):316-322
[8]丁迎,洪继光.图像拼接中伪匹配的判断和消解[J].中国图像国形学报,2003,4A(10):886-890
[9]许雷,张恒义等.一种基于相位相关法及数学形态学方法的眼底血管图像自动拼接方法[J].生物医学工程杂志,2003,15(3):286-290
[10]Bao P,Xu D.Panoramic image mosaics via complex wavelet pyramid[C],Proceedings of ISSPR' 98,Hong Kong September1998:175-180
[11]王俊杰,刘家茂,胡运发等.图像拼接技术.计算机科学学报.2006,11(6):141-150
[12]潘立公,汤晓安.一种360度柱面全景图快速生成算法.研究与设计,2005,17(4):19-35
[13]杨枝灵,王开.Visual C++数字图像获取处理及实践应用.北京:人民邮电出版社,2003:100-360
[14]刘翔,曾芬芳.基于IBR披术的全景图生成.华东船舶工业学院学报(自然科学版).2005,15(1):32-37
[15]沈建国,戴永刚.基于局部熵差的栅格地图拼接方法.华东师范大学学报(自然科学版).2002,(4):48-53
[16]唐琏,谷士文,杨安平等.立方体表面上的图象拼接.小型微型计算机系统.2004,23(1):93-97
[17]兰培真,马越,邱志雄等.不同视点重叠图象拼接方法研究.中国图象图形学报,2002,7(10):1032-1037
[18]潘华伟,邹北骥.一种圆柱形全景图生成新算法及其实现.计算机工程与科学.2003,25(6):13-16
[19]漆驰,刘强,孙家广等.摄像机图像序列的全景图拼接.计算机辅助设计与图形学报,2003,13(7):605-609
[20]侯舒维,郭宝龙.一种图像自动拼接的快速算法[J].计算机工程,2005,31(24):70-72
[21]王俊杰.刘家茂.图像拼接技术.计算机科学,2004,(30):141-144
[22]朱云芳,叶秀清,顾伟康.视频序列的全景图拼接技术.中国图形图像学报,2006,11(8):198-203
[23]蒋苏蓉.王松,冯刚.基于图像的全景视图的实现披术.计算机应用,2005,22(6):86-87
[24]毛晓冬.图像配准与拼接方法研究:[硕士学位论文].西安:西安电子科技大学,2006
[25]于明言.一种基于时域和频域特征的图像拼接方法:[硕士学位论文].大连:大连理工大学,2006
[26]马龙.超声宽景成像技术研究:[硕士学位论文].西安:西安电子科技大学,2006
[27]高春霞.基于脉冲耦合神经网络和进化算法的图像分割方法研究:[硕士学位论文].西安:西安电子科技大学,2007
[28]陈爱萍.基于图像配准的高分辨率重构技术研究:[硕士学位论文].西安:西安电子科技大学,2007
[29]谭康.图像拼接技术研究与实现:[硕士学位论文].南京:南京理工大学计算机应用技术,2006
[30]叶朝阳.大视场图像拼接:原理与应用:[硕士学位论文].浙江:浙江大学计算机应用,2004
[31]张静,胡志萍,刘志泰等.基于轮廓相位相关的图像自动拼接.大连理工大学学报,2005,45(1):68-74
[32]张世阳.一种快速全景图像拼接技术.计算机应用与软件,2004,21(3),77-79
[33]张静.基于图像绘制技术的虚拟场景环视和纵向漫游:[博士学位论文].大连:大连理工大学,2004
[34]钟立,胡晓峰.重叠图像拼接算法[J].中国图象图形学报,2005,(5):22-37
[35]庄志国,孙惠军,董继扬等.基于角点检测的图像匹配算法及其在图像拼接中的应用.厦门大学学报(自然科学版),2007,7(4):501-508
[36]全红艳,张田文,蔺想红.全景图拼接中的一种快速图像点对应策略[J].机器人,2005,16(3):45-49
[37]Chalechale A,Naghdy G,Mertins A.Sketch-based image matching using angular partitioning systems [J].IEEE Trans on Man and Cybernetics,Part A,2005,35(1):28-41
[38]赵向阳,杜利民.一种全自动稳健的图像拼接融合算法[J].中国图形图像学报,2004,Vol(9):417-422
[39]Reddy B S,Chatterji B N.An FFT-based technique for translation,rotation,and scale invariant image registration[J].IEEE Trans on Image Processing,2006,5(8):1266-1271
[40]容观澳,计算机图象处理.北京:清华大学出版社,2000:80-280
[41]倪明田,吴良芝.计算机图形学.北京:北京大学出版社,1999:153-325
[42]S.Gumustekin,R.W.Hall.Mosaic image generation on a flattened Gaussian sphere[J].In Proc.Of IEEE Workshop on APPlications of Computer Vision,2006,50-55
[43]祝乃庆.全景图自动拼接算法研究及系统实现:[硕士学位论文].南京:南京理工大学,2004
[44]Han B,Lin X.A novel hybrid color regist ration algorithm for image stitching[J].IEEE Trans on Consumer Electronics,2006,52(3):1129-1134
[45]曹俊杰,封靖波,苏志勋.全景图像拼接算法[J].大连理工大学学报,2006,S1:187-189
[46]Richard Szeliski,Heung-Yeung Shum.Creating full view panornamic image mosaics and texture-mapped models[J].SIGGRAPH 97 Conference Proceedings,2005,3(1):251-258
[47]陈永强,王启付.虚拟环境中变形图像拼接技术研究.华中科技大学学报,2005,29(1):114-166
[48]Harris C,Stephens M.A combined corner and edge detector[C]//Proceedings of The Fourth Alvey Vision Conference.Manchester:[s.n],2005:147-151
[49]Moravec H.P.Towards Automatic Visual Obstacle Avoidnace[C].In 5th International Joint Conference on Artificial Intelligence,2004:584-589
[50]唐琏,谷士文.与全方位全景图像有关的几个问题的研究[J].中国铁道科学学报,2004,23(4):89-96