基于立体视觉的皮肤表面检测技术的研究
|
摘要
如今，评价皮肤表面的光滑度越来越受到人们的重视，尤其是在化妆品研究领域。科学家们已经建立了几种检测皮肤皱纹、评价皮肤表面光滑度的方法。本文比较详细地介绍了各种皮肤皱纹检测的技术，并且对它们各自的优缺点以及适用范围进行了比较。另外，本文提出了基于立体视觉的皮肤表面检测技术，然后，分别对摄像机定标和立体匹配技术进行了研究。
在摄像机定标部分，本文介绍了摄像机定标的基本原理以及定标的具体步骤。在实验中，我们研究了定标点的密度极其分布对摄像机定标精度的影响。通过对实验结果进行分析，我们提出了用平面方格图对摄像机进行定标的方法。它使得定标点在空间分布更加均匀，定标精度提高。在计算定标点的图象坐标时，我们采用了哈夫变换的方法求出定标方格图上格线的方程，然后再求格线交点的图象坐标，从而使定标的参数更准确。
在立体匹配部分，我们着重介绍了区域匹配技术以及用于全局控制的极线约束方法。并且在实验的基础上，提出了一种基于标准点约束的全局控制策略。与极线约束的控制策略相比，它不受定标精度的影响，而且处理简便。
最后，总结全文。对自摄的图象进行了实验，并且分析了实验结果和系统的性能。指出今后需进一步研究和探索的方向。
Estimation of skin smoothness is of ever-increasing interest especially in the field of cosmetic research nowadays. There are some established methods for assessing skin surface. The paper interprets some measurement methods of skin surface and analyses them in detail. Furthermore, a measurement method of skin surface structure based on the image stereo is introduced. In the following part, the main items of binocular stereo vision are presented systematically. At first, the development of binocular stereo vision and some viewpoints about computer vision are discussed. Then we discuss the techniques of the camera calibration and stereo matching respectively.
In the camera calibration part, the basic principles and specific steps of camera calibration are presented. The effect of the density and distribution of the known calibrating points on the precise of the camera calibration is studied by experiments. By analyzing the results of the experiments, we propose a method of using a plane to calibrate a camera. It makes the distribution of the calibrating points more even in the space and the result of the camera calibration more accurate. In order to find the location of the cross points in the image, we first obtain the equations of the lines in the image plane by using Hough transform and then get the cross points of the lines. The result of the camera calibration using this method is more accurate.
In the stereo matching part, the area-based stereo match and the control strategy based on epipolar line are discussed in detail. On the basis of experiment, we propose a new solution to stereo matching by using the control strategy of the standard points. In contrast to the control strategy based on epipolar line, it is free from the effect of
the camera calibration.
Finally, we review the whole work, process the images, analyze the results and suggest the direction for the future research in this field.
在摄像机定标部分，本文介绍了摄像机定标的基本原理以及定标的具体步骤。在实验中，我们研究了定标点的密度极其分布对摄像机定标精度的影响。通过对实验结果进行分析，我们提出了用平面方格图对摄像机进行定标的方法。它使得定标点在空间分布更加均匀，定标精度提高。在计算定标点的图象坐标时，我们采用了哈夫变换的方法求出定标方格图上格线的方程，然后再求格线交点的图象坐标，从而使定标的参数更准确。
在立体匹配部分，我们着重介绍了区域匹配技术以及用于全局控制的极线约束方法。并且在实验的基础上，提出了一种基于标准点约束的全局控制策略。与极线约束的控制策略相比，它不受定标精度的影响，而且处理简便。
最后，总结全文。对自摄的图象进行了实验，并且分析了实验结果和系统的性能。指出今后需进一步研究和探索的方向。
Estimation of skin smoothness is of ever-increasing interest especially in the field of cosmetic research nowadays. There are some established methods for assessing skin surface. The paper interprets some measurement methods of skin surface and analyses them in detail. Furthermore, a measurement method of skin surface structure based on the image stereo is introduced. In the following part, the main items of binocular stereo vision are presented systematically. At first, the development of binocular stereo vision and some viewpoints about computer vision are discussed. Then we discuss the techniques of the camera calibration and stereo matching respectively.
In the camera calibration part, the basic principles and specific steps of camera calibration are presented. The effect of the density and distribution of the known calibrating points on the precise of the camera calibration is studied by experiments. By analyzing the results of the experiments, we propose a method of using a plane to calibrate a camera. It makes the distribution of the calibrating points more even in the space and the result of the camera calibration more accurate. In order to find the location of the cross points in the image, we first obtain the equations of the lines in the image plane by using Hough transform and then get the cross points of the lines. The result of the camera calibration using this method is more accurate.
In the stereo matching part, the area-based stereo match and the control strategy based on epipolar line are discussed in detail. On the basis of experiment, we propose a new solution to stereo matching by using the control strategy of the standard points. In contrast to the control strategy based on epipolar line, it is free from the effect of
the camera calibration.
Finally, we review the whole work, process the images, analyze the results and suggest the direction for the future research in this field.
引文
1 魏少敏等．皮肤纹理和皱纹深度的图象分析测定研究．日用化学工业．2000．4
2 毛培坤主编．化妆品功能性评价和分析方法．中国轻工业出版社．1998
3 弓娟琴．计算机数字图象分析在皮肤科的应用．临床皮肤科杂志，1997，06
4 Aritcus.K,Khazaka.G and Wilhelm.K.P.The Skin Visiometer—A Photometric Device for the Measurement of Skin Roughness.Bioengineering of the Skin:Skin Surface Imaging and Analysis.1997,CRC Press
5 Tobias W.Fischer,Walter Wigger-Alberti,Peter Elsner.Direct and Non-Direct Measurement Techniques for Analysis of Skin Surface Topography.Skin pharmacology and applied skin physiology.1999,12:1～11
6 Dr.H.P.Nissen.Comparative Studies of Skin Roughness Measurements by Profilometry and a New Image Analysis System. Cosmetic and Toiletries Manufacture World-Wide,Jan.1995
7 姚国正，刘磊，汪云九译．视觉计算理论．科学出版社，1988
8 郑南宁．计算机视觉与模式识别．国防工业出版社．1998
9 游素亚等．立体视觉研究的现状与进展．中国图象图形学报，2(1)：17～24，1997
10 章毓晋编著．图象理解与计算机视觉．清华大学出版社．2000
11 马颂德，张正友．计算机视觉——计算理论与算法基础．科学出版社．1998
12 黄勇等．利用光度法生成真实物体三维模型．中国图象图形学报，1998，3(10)：814～821
13 S.K.Nayar,K.Ikeuchi and T:Kannde.Extracting Shape from Reflectance of Hybrid Surface by Photomegtric Sampling.Image Understandign workshop.1989,563～583
14 B K P Horn,Height and Gradient from Shading.Computer vision,1990,5(1):37～75
15 A.M.Bruckstein.On Shape from Shading.CVGIP.1988(44):139～154
16 Wu ZQ,Li L X.A line-integration based method for depth recovery from surface normal.CVIU.1988,43:53～66
17 孙龙祥等．深度图象分析．电子工业出版社．1996
18 R J.Photometric Method for Determinating Surface Orientation from Multiple Images.Optical Engineering,1980(19):139～141
19 Jarvis R A.A Perspective on Range Finding Technique for Computer Vision.IEEE PAMI,1983,5(2):122～139
20 R.A.Lewis and A.R.Johnstone.A Scanning Laser Rangefinder for a Robotic Vehicle.Proc.of the 5th Int.Joint conference on Artificial Intelligence.1997:762～768
21 张吉华等．激光三角法轮廓测量．天津大学学报，1995，28(2)：265～269
22 唐朝伟等．三维曲面激光精密测量技术．计量学报，1994，15(2)：99～103
23 张广军等．结构光三维视觉系统研究．航空学报。1999，20(4)：365～367
24 扬敬安．关于融合多针图确定表面绝对深度的误差分析．电子学报．1994，22(7)：9～16
25 S.D.Blostein,et al.Error Analysis in Stereo Determination of 3-D Point Positions.IEEE PAMI,1987, 9(6):752～765
26 邱茂林等．计算机视觉中摄像机定标综述．自动化学报，2000，26(1)
27 Tsai.R.Y.An efficient and accurate camera calibration technique for 3D machine vision.in:proc.CUPR'86.364～374
28 Medioni G,Nevatia R.segment-based stereo matching. CVGIP,31(1):2～18
29 W.Eric,L.Grimson.Computational Experiments with a Feature Based Stereo Algorithm.IEEE PAMI.1985,7(1):17～34
30 Guo-Qing.Wei,Wilfried Brauer and Gerd Hirzinger.Intensity-and Gradient-Based Stereo Matching Using Hierarchical Gaussian Basis Functions.IEEE PAMI,1998,20(11):1143～1160
31 William H and Narendra A.Surfaces From Stereo:Integrating Feature Matching,Disparity Estimation,and Contour Detection. IEEE PAMI,1989,11(2):121～136
32 权铁汉等．摄影测量系统的高精度标定与修正．自动化学报，2000，26(6)
33 周新伦等编．数字图象处理．国防工业出版社，1986．6
34 吴建康著．数字图象分析．人民邮电出版社，1991．11
35 朱霞辉等．一种有效的指纹图象的纹线细化处理法．计算机应用研究．1997．6
36 高文，陈熙霖．计算机视觉——算法与系统原理．清华大学出版社．1999
37 李利，马颂德．从二维轮廓线重构三维二次曲面形状．计算机学报，1996，19(6)：401
38 杨新．一种基于梯度算子的图象匹配算法．电子学报，1999，27(10)：30～33
39 傅晓烨．一种眼底图象的三维重建方法．中国图象图形学报，1997，2(7)：509～512
40 S.B.Pollard,J.E.W.Mayhew and J.P.Frisby,PMF:A Stereo Correspondence Algorithm Using a Disparty Gradient Limit, Perception,14:449～470,1985
41 赵荣椿．数字图象处理导论．西北工业大学．1995
42 孙仲康/沈振康著．数字图象处理及其应用．国防工业出版社．1985
43 宣雷，徐建华．摄像机参数的测定方法与参数估计．模式识别与人工能，1993，6(1)：67～70
44 丁震等．一种渐进的立体匹配方法．计算机研究与发展，1997，34(s)：42～47
45 Nasser M.Nasrabadi.A stereo Vision Technique Using Curve-Segments and Relaxation Matching.1992,14(5):566～572
46 Baker.Depth from Edge and Intensity Based Stereo.Proceedings of The 7th Joint Conference on Artificial Intelligence.1981,631～636
47 Tsai.R Y.Multiframe Image Point Matching and 3D Surface Reconstruction.IEEE PAMI.1983,5(2):159～173
48 E.S.Mcvey, et al.Some Accuracy and Resolution Aspects of Computer Vision Distance Measurements.IEEE PAMI,1982,4(6):646～657
2 毛培坤主编．化妆品功能性评价和分析方法．中国轻工业出版社．1998
3 弓娟琴．计算机数字图象分析在皮肤科的应用．临床皮肤科杂志，1997，06
4 Aritcus.K,Khazaka.G and Wilhelm.K.P.The Skin Visiometer—A Photometric Device for the Measurement of Skin Roughness.Bioengineering of the Skin:Skin Surface Imaging and Analysis.1997,CRC Press
5 Tobias W.Fischer,Walter Wigger-Alberti,Peter Elsner.Direct and Non-Direct Measurement Techniques for Analysis of Skin Surface Topography.Skin pharmacology and applied skin physiology.1999,12:1～11
6 Dr.H.P.Nissen.Comparative Studies of Skin Roughness Measurements by Profilometry and a New Image Analysis System. Cosmetic and Toiletries Manufacture World-Wide,Jan.1995
7 姚国正，刘磊，汪云九译．视觉计算理论．科学出版社，1988
8 郑南宁．计算机视觉与模式识别．国防工业出版社．1998
9 游素亚等．立体视觉研究的现状与进展．中国图象图形学报，2(1)：17～24，1997
10 章毓晋编著．图象理解与计算机视觉．清华大学出版社．2000
11 马颂德，张正友．计算机视觉——计算理论与算法基础．科学出版社．1998
12 黄勇等．利用光度法生成真实物体三维模型．中国图象图形学报，1998，3(10)：814～821
13 S.K.Nayar,K.Ikeuchi and T:Kannde.Extracting Shape from Reflectance of Hybrid Surface by Photomegtric Sampling.Image Understandign workshop.1989,563～583
14 B K P Horn,Height and Gradient from Shading.Computer vision,1990,5(1):37～75
15 A.M.Bruckstein.On Shape from Shading.CVGIP.1988(44):139～154
16 Wu ZQ,Li L X.A line-integration based method for depth recovery from surface normal.CVIU.1988,43:53～66
17 孙龙祥等．深度图象分析．电子工业出版社．1996
18 R J.Photometric Method for Determinating Surface Orientation from Multiple Images.Optical Engineering,1980(19):139～141
19 Jarvis R A.A Perspective on Range Finding Technique for Computer Vision.IEEE PAMI,1983,5(2):122～139
20 R.A.Lewis and A.R.Johnstone.A Scanning Laser Rangefinder for a Robotic Vehicle.Proc.of the 5th Int.Joint conference on Artificial Intelligence.1997:762～768
21 张吉华等．激光三角法轮廓测量．天津大学学报，1995，28(2)：265～269
22 唐朝伟等．三维曲面激光精密测量技术．计量学报，1994，15(2)：99～103
23 张广军等．结构光三维视觉系统研究．航空学报。1999，20(4)：365～367
24 扬敬安．关于融合多针图确定表面绝对深度的误差分析．电子学报．1994，22(7)：9～16
25 S.D.Blostein,et al.Error Analysis in Stereo Determination of 3-D Point Positions.IEEE PAMI,1987, 9(6):752～765
26 邱茂林等．计算机视觉中摄像机定标综述．自动化学报，2000，26(1)
27 Tsai.R.Y.An efficient and accurate camera calibration technique for 3D machine vision.in:proc.CUPR'86.364～374
28 Medioni G,Nevatia R.segment-based stereo matching. CVGIP,31(1):2～18
29 W.Eric,L.Grimson.Computational Experiments with a Feature Based Stereo Algorithm.IEEE PAMI.1985,7(1):17～34
30 Guo-Qing.Wei,Wilfried Brauer and Gerd Hirzinger.Intensity-and Gradient-Based Stereo Matching Using Hierarchical Gaussian Basis Functions.IEEE PAMI,1998,20(11):1143～1160
31 William H and Narendra A.Surfaces From Stereo:Integrating Feature Matching,Disparity Estimation,and Contour Detection. IEEE PAMI,1989,11(2):121～136
32 权铁汉等．摄影测量系统的高精度标定与修正．自动化学报，2000，26(6)
33 周新伦等编．数字图象处理．国防工业出版社，1986．6
34 吴建康著．数字图象分析．人民邮电出版社，1991．11
35 朱霞辉等．一种有效的指纹图象的纹线细化处理法．计算机应用研究．1997．6
36 高文，陈熙霖．计算机视觉——算法与系统原理．清华大学出版社．1999
37 李利，马颂德．从二维轮廓线重构三维二次曲面形状．计算机学报，1996，19(6)：401
38 杨新．一种基于梯度算子的图象匹配算法．电子学报，1999，27(10)：30～33
39 傅晓烨．一种眼底图象的三维重建方法．中国图象图形学报，1997，2(7)：509～512
40 S.B.Pollard,J.E.W.Mayhew and J.P.Frisby,PMF:A Stereo Correspondence Algorithm Using a Disparty Gradient Limit, Perception,14:449～470,1985
41 赵荣椿．数字图象处理导论．西北工业大学．1995
42 孙仲康/沈振康著．数字图象处理及其应用．国防工业出版社．1985
43 宣雷，徐建华．摄像机参数的测定方法与参数估计．模式识别与人工能，1993，6(1)：67～70
44 丁震等．一种渐进的立体匹配方法．计算机研究与发展，1997，34(s)：42～47
45 Nasser M.Nasrabadi.A stereo Vision Technique Using Curve-Segments and Relaxation Matching.1992,14(5):566～572
46 Baker.Depth from Edge and Intensity Based Stereo.Proceedings of The 7th Joint Conference on Artificial Intelligence.1981,631～636
47 Tsai.R Y.Multiframe Image Point Matching and 3D Surface Reconstruction.IEEE PAMI.1983,5(2):159～173
48 E.S.Mcvey, et al.Some Accuracy and Resolution Aspects of Computer Vision Distance Measurements.IEEE PAMI,1982,4(6):646～657