基于CT数据的颅面信息提取及三维重建研究
|
摘要
颅面形态学通过研究找出人的颅骨和面貌之间的形态关系,为颅骨面貌复原提供科学依据,近年来这方面的研究受到了广泛的关注和重视。本文针对颅面信息提取与三维重建展开研究,以人体脑部CT图象作为样本数据,实现了CT图象的预处理、分割及轮廓线提取,为快速重建三维模型提供了良好基础。本文研究工作主要包括:
(1)提出了一种剔除CT图象中冗余影象的方法,该方法结合图像分割与轮廓跟踪,在获得面皮轮廓线的同时能够自动进行冗余影像的剔除,实验表明该算法速度快,剔除效果好,能够实现大量CT数据的批处理。
(2)利用遗传算法得到了吉布斯分布的参数估计,结合最大后验概率准则,提出了GA-MRF-MAP分割算法。与现有的几种CT图象分割算法相比,应用GA-MRF-MAP算法得到的分割结果区域一致性好,正确率高,并实现了无监督分割。
(3)在梯度算子与四邻域法的基础上提出了射线扫射法与坐标法,只提取骨骼的最外层轮廓线,避免了多余轮廓线造成的数据存储及重建压力。
(4)使用基于轮廓线的表面重建方法对颅骨和人脸进行三维模型重建,得到了无厚度的颅骨模型。
以Visual C++ 6.0为主要开发工具,结合VTK开发包,设计实现了对CT图象处理及三维重建的系统“CTManageSys”,该系统为颅面信息后处理提供了基础平台。
Skull-face morphology research is to find the morphologic relationship between the skull and face, it provides a scientific foundation for the skull-face reversion research, the two researches have been paid wide attention these years. This paper does researches on skull-face information extraction and 3D reconstruction, it has realized CT preprocessing, segmentation and contour track based on CT images, providing a better base for fleetly 3D reconstruction. The work in this paper mainly include:
(1)Proposes a new method which combines contour track and segmentation together to clear the redundancy in CT images, the method could extract the face's contour, at the same time, it could automaticly remove the redundancy in CT images. The method is proved fast and good by experiments and it has application value.
(2)Gets the estimation of Gibbs parameter through the genetic algorithm, combines the maximum a posteriori probability rule, proposes a segmentation algorithm called GA-MRF-MAP. Comparing with some existed segmentation algorithm, GA-MRF-MAP algorithm is consistent, accurate, and it is an unsupervised method .
(3)Based on the method of Tidusuzi and Silinyu, this paper proposes the other two methods: line scanning method and coordinate judgement method , the two methods only extract the outer contours of skull, they avoid the pressure from data store and 3D construction.
(4)Gets single-layer 3D models of face and skull through the surface reconstruction method.
Designed a system called "CTManageSys" to dull with CT images and to make 3D reconstruction, this system is a basic flat for later researches.
(1)提出了一种剔除CT图象中冗余影象的方法,该方法结合图像分割与轮廓跟踪,在获得面皮轮廓线的同时能够自动进行冗余影像的剔除,实验表明该算法速度快,剔除效果好,能够实现大量CT数据的批处理。
(2)利用遗传算法得到了吉布斯分布的参数估计,结合最大后验概率准则,提出了GA-MRF-MAP分割算法。与现有的几种CT图象分割算法相比,应用GA-MRF-MAP算法得到的分割结果区域一致性好,正确率高,并实现了无监督分割。
(3)在梯度算子与四邻域法的基础上提出了射线扫射法与坐标法,只提取骨骼的最外层轮廓线,避免了多余轮廓线造成的数据存储及重建压力。
(4)使用基于轮廓线的表面重建方法对颅骨和人脸进行三维模型重建,得到了无厚度的颅骨模型。
以Visual C++ 6.0为主要开发工具,结合VTK开发包,设计实现了对CT图象处理及三维重建的系统“CTManageSys”,该系统为颅面信息后处理提供了基础平台。
Skull-face morphology research is to find the morphologic relationship between the skull and face, it provides a scientific foundation for the skull-face reversion research, the two researches have been paid wide attention these years. This paper does researches on skull-face information extraction and 3D reconstruction, it has realized CT preprocessing, segmentation and contour track based on CT images, providing a better base for fleetly 3D reconstruction. The work in this paper mainly include:
(1)Proposes a new method which combines contour track and segmentation together to clear the redundancy in CT images, the method could extract the face's contour, at the same time, it could automaticly remove the redundancy in CT images. The method is proved fast and good by experiments and it has application value.
(2)Gets the estimation of Gibbs parameter through the genetic algorithm, combines the maximum a posteriori probability rule, proposes a segmentation algorithm called GA-MRF-MAP. Comparing with some existed segmentation algorithm, GA-MRF-MAP algorithm is consistent, accurate, and it is an unsupervised method .
(3)Based on the method of Tidusuzi and Silinyu, this paper proposes the other two methods: line scanning method and coordinate judgement method , the two methods only extract the outer contours of skull, they avoid the pressure from data store and 3D construction.
(4)Gets single-layer 3D models of face and skull through the surface reconstruction method.
Designed a system called "CTManageSys" to dull with CT images and to make 3D reconstruction, this system is a basic flat for later researches.
引文
[1]刘泗岩.DICOM数据的三维重建与协同可视化系统的开发[J].生物医学工程学杂志,2007,24(5):1152-1157
[2]黄志聪.DICOM的发展及其最新的变化[J].中国医疗器械杂志,2004,28(3):203-207
[3]李磊.基于CT断层的颅骨破洞修复技术研究[D].西安:西北大学,2008
[4]孙亦博.基于CT数据的三维重构相关技术的研究与应用[D].西安:西北大学,2007
[5]Gonzalez R.C,Woods R.E.数字图象处理[M].第二版,阮秋琦,阮宇智,等.北京:电子工业出版社,2003:51-54,91-97,183-194,463-474
[6]李燕.交互式空间物体三维测量方法的研究与实现[D].西安:西北大学,2003
[7]李康.基于颅骨的人脸建模技术研究及在法医面貌复原中的应用[D].西安:西北大学,2006
[8]边肇祺,张学工.模式识别[M].第二版.北京:清华大学出版社,2000:235-237,280
[9]周明全,耿国华,韦娜.基于内容图象检索技术[M].北京:清华大学出版社,2007:36-38
[10]何东建,耿楠,张义宽.数字图象处理[M].西安:西安电子科技大学出版社,2003:64-73,90-99
[11]唐振军,张显全.一种二值图象边界提取算法[J].微计算机信息,2006,22(10-3):281-283
[12]王亮,黄飞,等.一种改进的二值图象象素标记算法[J].南京师范大学学报,2006,6(2):23-29
[13]赵于前,桂卫华,等.基于自适应数学形态学的医学图象边缘连接[J].计算机工程,2006,32(22):17-19
[14]李刚,黎燕,等.基于动态数学形态学运算的二值图象边界提取[J].计算机工程与应用,2005,24:68-70
[15]尚斌,徐良贤.一种中风病人脑出血CT图象序列的自动分割方法[J].计算机工程,2004,30(z1):356-514
[16]朱峰.医学图象自动分割若干关键技术研究[D].苏州:江苏大学硕士论文,2006
[17]程广斌,郝立巍,陈武凡.利用Gibbs距离图Snake模型分割医学图象[J].南方医科大学学报,2008,28(1):48-51
[18]Zhang Y.F,Wu Shuang,Yu Ge,et al.A Hybrid Image Segmentaion Approach Using Watershed Transform and FCM[C].Fourth International Conference on Fuzzy Systems and Knowledge Discovery,2007
[19]林亚忠.基于Gibbs随机场模型的医学图象分割新算法研究[D].广州:第一军医大学,2004
[20]Srinivas C,Srinath M.D.Compound Gauss Markov random field model for image segmentation and restoration[J].Acoustics,Speech,and Signal Processing,1989,3(23):1586-1589
[21]Rama Chellappa,Shankar Chatterjee.Classification of Textures Using Gaussian Markov Random Fields[J].IEEE TRANSACTIONS ON ACOUSTICS,SPEECH AND SIGNAL PROCESSING,1985,Assp-33(4):959-963
[22]Charles A.Bouman.Markov Random Fields and Stochastic Image Models[C].IEEE International Conference on Image Processing,1995,23-26:1-103
[23]Philippe Andrey,Philippe Tarroux.Unsupervised Segmentation of Markov Random Field Modeled Textured Images Using Selectionist Relaxation[J].IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHING INTELLIGENCE,1998,20(3):252-262
[24]张翠,郦苏丹,等.基于MRF场的SAR图象分割方法[J].遥感技术与应用,2001,16(1):66-68
[25]陆俊明,王润生.基于MRF模型可靠的图象分割[J].电子学报,1999,27(2):87-89
[26]候一民,郭雷.一种基于马尔可夫随机场的SAR图象分割新方法[J].电子与信息学报 2007,29(5):1069-1072
[27]林亚忠,杨丰,陈武凡.一种基于随机场模型的图象分割算法[DB/OL].http://www.enki.net.2002,54-55
[28]侯一民,郭雷.一种鲁棒的MRF-MAP图象分割框架研究[J].计算机工程与应用,2006,27:62-68
[29]彭祥龙,张扬.马尔可夫随机场在SAR图象处理中的应用[J].电讯技术,2003,(1):63-80
[30]刘伟强,陈鸿,夏德琛.基于马尔可夫随机场的快速图象分割[J].中国图象图形学报,2001,6A(3):228-233
[31]Faguo Yang,Tianzi Jiang.Pixon-based Image Segmentation with Markov random fields[J].ACCV:The 5th Asian Conference on Computer Vision.2002,23(25):1-6
[32]Xia Yong,(David)Feng Dagan,Zhao Rong chun.Adaptive Segmentation of Textured Images by Using the Coupled Markov Random Field Model[J].IEEE TRANSACTIONS ON IMAGE PROCESSING,2006,15(11):3559-3566
[33]William Robson Sehwastz,Helio Pedrini.Textured Image Segmentation Based on Spatial Dependence using a Markov Random Field Model[J].IEEE ICIP.2006,2449-2452
[34]李旭超,朱善安.图象分割中的马尔可夫随机场方法综述[J].中国图象图形学报,2007,12(5):789-798
[35]汪西莉,刘芳,焦礼成.融合上下文信息的多尺度贝叶斯图象分割[J].计算机学报,2005,28(3):386-391
[36]汪西莉,刘芳,焦礼成.一种分层马尔可夫图象模型及其推导算法[J].软件学报,2003,14(9):1558-1563
[37]张红蕾,宋建社,张宪伟.一种基于多尺度马尔可夫随机场的SAR图象分割算法[J].计算机应用,2005,25(12):2830-2831
[38]Charles Bouman,Bede Liu.Multiple Resolution Segmentation of Textured Images[J].IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE,1991,13(2):99-113
[39]Ciro D'Elia,Giovanni Poggi,Giuseppe Searpa.A Tree-Structured Markov Random Field Model for Bayesian Image Segmentation[J].IEEE TRANSACTIONS ON IMAGE PROCESSING 2003,12(10):1259-1273
[40]Zhang Hong-mei,Yuan Ze-jian,Cai Zhong-min,etal Segmentation of MRI using hierarchical markov random field[J].Journal of Software,2002,13(9):1779-1786
[41]金炎芳,张定华,等.基于混合高斯模型MRF场的CT图象分割[J].计算机应用研究,2007,24(6):176-180
[42]余鹏,张震龙,侯至群.基于高斯马尔可夫随机场混合模型的纹理图象分割[J].测绘学报,2006,35(8):224-228
[43]冯衍秋,陈武凡.基于Gibbs随机场与模糊C均值聚类的图象分割新算法[J].电子学报,2004,32(4):645-647
[44]Feng Yanqiu,Chen Wufan.Brain MR Image Segmentation Using Fuzzy Clustering with Spatial Constraints Based on Markov Random Field Theory[J].In:Proc.Of MIAR 2004(LNCS 3150,Beijing).Belin:Springer-Verlag,188-195
[45]吕庆文.基于互信息量的马尔可夫随机场的图象分割研究[D].广州:第一军医大学,2007[46]http://www.ChinaAL.org
[47]http://baike.baldu.com
[48]张玲,刘勇,何伟.自适应遗传算法在车牌定位中的应用[J].计算机应用,2008,28(1):184-186
[49]王力,王成良.基于免疫遗传算法的关系型数据库查询优化技术[J].计算机系统应用,2008,(1):72-75
[50]曾筝,陈晓,等.基于凸包算法的三维表面重建中边缘轮廓提取[J].微机发展,2004,14(12):39-41
[51]W.Yao,P.Abolmaesumi,M.Greenspan.et,al.An Estimation/Correction Algorithm for Detecting Bone Edges in CT images[J].IEEE,2005,24(8),997-999
[52]申静波,李井辉,王惠影.CT图象区域轮廓线提取方法[J].大庆石油学院学报,2007,31(2):75-77
[53]程东旭,秦新强,张太发.基于区域增长的轮廓线提取算法[J].西安理工大学学报,2005,21(4):413-416
[54]程琼.基于二维CT图象的三维重建及其应用[D].上海:同济大学,2006
[55]刘力强.三维可视化技术在颅骨面貌复原系统中的应用研究[D].西安:西北大学,2002
[56]樊宏斌.计算机辅助颅骨面貌复原技术的研究与实现[D].西安:西北大学,2002
[57]刘家康,辛静.人体医学图象三维表面重构的实现[J].计算机工程与应用,2002,38(14):245-247
[58]沈艳,郭兵.粒子群优化算法及其与遗传算法的比较[J].电子科技大学学报,2005,34(5):696-699
[2]黄志聪.DICOM的发展及其最新的变化[J].中国医疗器械杂志,2004,28(3):203-207
[3]李磊.基于CT断层的颅骨破洞修复技术研究[D].西安:西北大学,2008
[4]孙亦博.基于CT数据的三维重构相关技术的研究与应用[D].西安:西北大学,2007
[5]Gonzalez R.C,Woods R.E.数字图象处理[M].第二版,阮秋琦,阮宇智,等.北京:电子工业出版社,2003:51-54,91-97,183-194,463-474
[6]李燕.交互式空间物体三维测量方法的研究与实现[D].西安:西北大学,2003
[7]李康.基于颅骨的人脸建模技术研究及在法医面貌复原中的应用[D].西安:西北大学,2006
[8]边肇祺,张学工.模式识别[M].第二版.北京:清华大学出版社,2000:235-237,280
[9]周明全,耿国华,韦娜.基于内容图象检索技术[M].北京:清华大学出版社,2007:36-38
[10]何东建,耿楠,张义宽.数字图象处理[M].西安:西安电子科技大学出版社,2003:64-73,90-99
[11]唐振军,张显全.一种二值图象边界提取算法[J].微计算机信息,2006,22(10-3):281-283
[12]王亮,黄飞,等.一种改进的二值图象象素标记算法[J].南京师范大学学报,2006,6(2):23-29
[13]赵于前,桂卫华,等.基于自适应数学形态学的医学图象边缘连接[J].计算机工程,2006,32(22):17-19
[14]李刚,黎燕,等.基于动态数学形态学运算的二值图象边界提取[J].计算机工程与应用,2005,24:68-70
[15]尚斌,徐良贤.一种中风病人脑出血CT图象序列的自动分割方法[J].计算机工程,2004,30(z1):356-514
[16]朱峰.医学图象自动分割若干关键技术研究[D].苏州:江苏大学硕士论文,2006
[17]程广斌,郝立巍,陈武凡.利用Gibbs距离图Snake模型分割医学图象[J].南方医科大学学报,2008,28(1):48-51
[18]Zhang Y.F,Wu Shuang,Yu Ge,et al.A Hybrid Image Segmentaion Approach Using Watershed Transform and FCM[C].Fourth International Conference on Fuzzy Systems and Knowledge Discovery,2007
[19]林亚忠.基于Gibbs随机场模型的医学图象分割新算法研究[D].广州:第一军医大学,2004
[20]Srinivas C,Srinath M.D.Compound Gauss Markov random field model for image segmentation and restoration[J].Acoustics,Speech,and Signal Processing,1989,3(23):1586-1589
[21]Rama Chellappa,Shankar Chatterjee.Classification of Textures Using Gaussian Markov Random Fields[J].IEEE TRANSACTIONS ON ACOUSTICS,SPEECH AND SIGNAL PROCESSING,1985,Assp-33(4):959-963
[22]Charles A.Bouman.Markov Random Fields and Stochastic Image Models[C].IEEE International Conference on Image Processing,1995,23-26:1-103
[23]Philippe Andrey,Philippe Tarroux.Unsupervised Segmentation of Markov Random Field Modeled Textured Images Using Selectionist Relaxation[J].IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHING INTELLIGENCE,1998,20(3):252-262
[24]张翠,郦苏丹,等.基于MRF场的SAR图象分割方法[J].遥感技术与应用,2001,16(1):66-68
[25]陆俊明,王润生.基于MRF模型可靠的图象分割[J].电子学报,1999,27(2):87-89
[26]候一民,郭雷.一种基于马尔可夫随机场的SAR图象分割新方法[J].电子与信息学报 2007,29(5):1069-1072
[27]林亚忠,杨丰,陈武凡.一种基于随机场模型的图象分割算法[DB/OL].http://www.enki.net.2002,54-55
[28]侯一民,郭雷.一种鲁棒的MRF-MAP图象分割框架研究[J].计算机工程与应用,2006,27:62-68
[29]彭祥龙,张扬.马尔可夫随机场在SAR图象处理中的应用[J].电讯技术,2003,(1):63-80
[30]刘伟强,陈鸿,夏德琛.基于马尔可夫随机场的快速图象分割[J].中国图象图形学报,2001,6A(3):228-233
[31]Faguo Yang,Tianzi Jiang.Pixon-based Image Segmentation with Markov random fields[J].ACCV:The 5th Asian Conference on Computer Vision.2002,23(25):1-6
[32]Xia Yong,(David)Feng Dagan,Zhao Rong chun.Adaptive Segmentation of Textured Images by Using the Coupled Markov Random Field Model[J].IEEE TRANSACTIONS ON IMAGE PROCESSING,2006,15(11):3559-3566
[33]William Robson Sehwastz,Helio Pedrini.Textured Image Segmentation Based on Spatial Dependence using a Markov Random Field Model[J].IEEE ICIP.2006,2449-2452
[34]李旭超,朱善安.图象分割中的马尔可夫随机场方法综述[J].中国图象图形学报,2007,12(5):789-798
[35]汪西莉,刘芳,焦礼成.融合上下文信息的多尺度贝叶斯图象分割[J].计算机学报,2005,28(3):386-391
[36]汪西莉,刘芳,焦礼成.一种分层马尔可夫图象模型及其推导算法[J].软件学报,2003,14(9):1558-1563
[37]张红蕾,宋建社,张宪伟.一种基于多尺度马尔可夫随机场的SAR图象分割算法[J].计算机应用,2005,25(12):2830-2831
[38]Charles Bouman,Bede Liu.Multiple Resolution Segmentation of Textured Images[J].IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE,1991,13(2):99-113
[39]Ciro D'Elia,Giovanni Poggi,Giuseppe Searpa.A Tree-Structured Markov Random Field Model for Bayesian Image Segmentation[J].IEEE TRANSACTIONS ON IMAGE PROCESSING 2003,12(10):1259-1273
[40]Zhang Hong-mei,Yuan Ze-jian,Cai Zhong-min,etal Segmentation of MRI using hierarchical markov random field[J].Journal of Software,2002,13(9):1779-1786
[41]金炎芳,张定华,等.基于混合高斯模型MRF场的CT图象分割[J].计算机应用研究,2007,24(6):176-180
[42]余鹏,张震龙,侯至群.基于高斯马尔可夫随机场混合模型的纹理图象分割[J].测绘学报,2006,35(8):224-228
[43]冯衍秋,陈武凡.基于Gibbs随机场与模糊C均值聚类的图象分割新算法[J].电子学报,2004,32(4):645-647
[44]Feng Yanqiu,Chen Wufan.Brain MR Image Segmentation Using Fuzzy Clustering with Spatial Constraints Based on Markov Random Field Theory[J].In:Proc.Of MIAR 2004(LNCS 3150,Beijing).Belin:Springer-Verlag,188-195
[45]吕庆文.基于互信息量的马尔可夫随机场的图象分割研究[D].广州:第一军医大学,2007[46]http://www.ChinaAL.org
[47]http://baike.baldu.com
[48]张玲,刘勇,何伟.自适应遗传算法在车牌定位中的应用[J].计算机应用,2008,28(1):184-186
[49]王力,王成良.基于免疫遗传算法的关系型数据库查询优化技术[J].计算机系统应用,2008,(1):72-75
[50]曾筝,陈晓,等.基于凸包算法的三维表面重建中边缘轮廓提取[J].微机发展,2004,14(12):39-41
[51]W.Yao,P.Abolmaesumi,M.Greenspan.et,al.An Estimation/Correction Algorithm for Detecting Bone Edges in CT images[J].IEEE,2005,24(8),997-999
[52]申静波,李井辉,王惠影.CT图象区域轮廓线提取方法[J].大庆石油学院学报,2007,31(2):75-77
[53]程东旭,秦新强,张太发.基于区域增长的轮廓线提取算法[J].西安理工大学学报,2005,21(4):413-416
[54]程琼.基于二维CT图象的三维重建及其应用[D].上海:同济大学,2006
[55]刘力强.三维可视化技术在颅骨面貌复原系统中的应用研究[D].西安:西北大学,2002
[56]樊宏斌.计算机辅助颅骨面貌复原技术的研究与实现[D].西安:西北大学,2002
[57]刘家康,辛静.人体医学图象三维表面重构的实现[J].计算机工程与应用,2002,38(14):245-247
[58]沈艳,郭兵.粒子群优化算法及其与遗传算法的比较[J].电子科技大学学报,2005,34(5):696-699