利用曲率多特征融合改进人体三维脊椎模型特征点的标注方法

惠宇; 武君胜; 鱼滨; 张琛; 武震

doi:10.3788/OPE.20162411.2872

您当前的位置：

首页 >

文章列表页 >

利用曲率多特征融合改进人体三维脊椎模型特征点的标注方法

信息科学 | 更新时间：2020-07-07

- 利用曲率多特征融合改进人体三维脊椎模型特征点的标注方法
- Improvement of feature point annotation for three-dimensional spine model based on curvature multiple feature fusion
- 光学精密工程 2016年24卷第11期页码：2872-2879
- 作者机构：
  
  1.西北工业大学计算机学院, 陕西西安 710072
  2.西北工业大学软件与微电子学院, 陕西西安 710072
  3.西安电子科技大学计算机学院, 陕西西安 710072
- 作者简介：
  
  惠宇(1984-),男,陕西岐山人,博士研究生, 2009年于西安电子科技大学获得硕士学位,主要从事医学影像处理、软件工程、有限元计算的研究. E-mail:289178013@qq.com. E-mail:289178013@qq.com.
  [ "武君胜(1962-),男,陕西礼泉人,教授,博士生导师。1983年于西北大学获得学士学位;1990年于中国航空计算技术研究所获得硕士学位;1997年于西北工业大学获得博士学位。主要从事计算方法、科学计算可视化、软件工程研究。E-mail:wujunsheng@nwpu.edu.cn" ]
- 基金信息：
  
  国家自然科学基金资助项目(No.61172147)
- DOI：10.3788/OPE.20162411.2872
  中图分类号： TP391.4
- 收稿日期：2016-07-14，
  
  录用日期：2016-9-4，
  
  纸质出版日期：2016-11-25
- 稿件说明：
移动端阅览
惠宇, 武君胜, 鱼滨, 等. 利用曲率多特征融合改进人体三维脊椎模型特征点的标注方法[J]. 光学精密工程, 2016,24(11):2872-2879.

Yu HUI, Jun-sheng WU, Bin YU, et al. Improvement of feature point annotation for three-dimensional spine model based on curvature multiple feature fusion[J]. Editorial office of optics and precision engineeri, 2016, 24(11): 2872-2879.
惠宇, 武君胜, 鱼滨, 等. 利用曲率多特征融合改进人体三维脊椎模型特征点的标注方法[J]. 光学精密工程, 2016,24(11):2872-2879. DOI： 10.3788/OPE.20162411.2872.

Yu HUI, Jun-sheng WU, Bin YU, et al. Improvement of feature point annotation for three-dimensional spine model based on curvature multiple feature fusion[J]. Editorial office of optics and precision engineeri, 2016, 24(11): 2872-2879. DOI： 10.3788/OPE.20162411.2872.

摘要

由于脊椎生理结构的精准坐标描述和准确匹配尚未达到医学精度的要求，本文对如何精确描述脊柱腰骶段特征点的物理坐标进行研究。介绍了人体脊椎采样特征点的定义和传统标记方法。针对手动标注特征点精确度不够，易产生较大误差等问题提出了一种基于曲率多特征融合的自适应标注特征点的方法。该方法首先找出某个特征点的高斯曲率和平均曲率流的定义值，得到该特征点的法曲率相对极大值，并计算在指定极小半径

范围内的所有模型点的法曲率相对极大值。由于极大值曲率越大，三维模型表面在该点处的弯曲程度越大，该点就越能表现三维模型的大致轮廓，故以

范围内极大值最大的点作为特征点的曲率描述来替换手动拾取的点，从而准确反映该点的特征变化情况。最后，对改进结果进行偏差验证分析。结果表明：改进方法的准确度比现有手动标注特征点方法的准确度提高了约37%，验证了本文方法的有效性。

Abstract

Because the precise coordinate description and accurate matching of physiology of the lumber spine have not yet reached the accuracy requirement of the medicine

this paper researches how to describe the physical coordinates of lumber spine feature points precisely. The definition and traditional annotation method of human spine sample features were introduce. Then

an improved method to adaptively annotate the feature points was proposed based on curvature multiple feature fusing to solve the problem of weaker precision for the manual annotation of feature points. In this method

the definiting values of Gaussian curvature and average curvature of a certain feature point were found out firstly

and a relative maximum value of curvature of the feature point was obtained. Then

the relative maximum values of all the model points within a specified minimum radius

were obtained. Since the larger the curvature of maximum value was

the larger the crook degree of the 3-dimensioanl model surface on the point was

and the point can show the outline of the 3-dimensional model. Therefore

the maximum value was taken as the curvature description of the feature point to replace the manual pick up point

thus the feature variation of the point could be described precisely. Finally

the verification results were analyzed

and the results indicate that the improved method increases the accuracy by 37% as compared with that of traditional annotation methods. These results demonstrate that the new method is feasibility and effectiveness.

关键词

Keywords

references

孟勃,韩广良. 基于改进的尺度不变特征变换特征点匹配的电子稳像方法[J]. 计算机应用,2012,32(10):2817-2820.

MENG B,HAN G L. Electronic image stabilization method using improved scale invariant feature transform[J].Journal of Computer Applications, 2012,32(10):2817-2820.(in Chinese)

李根,李文辉. 基于思维进化算法的人脸特征点跟踪[J]. 吉林大学学报(工学版),2015,45(2):606-612.

LI G,LI W H.Facial feature tracking based on mind evolutionary algorithm[J]. Journal of Jilin University(Engineering and Technology Edition), 2015,45(2):606-612.(in Chinese)

胡俊华,田锦会. 医学图像三维重建技术研究新进展[J]. 中国科技信息,2013(11):117-117.

HU J H,TIAN J H. New research progress of medical image reconstruction technology[J]. China Science and Technology Information,2013(11):117-117.(in Chinese)

BALLERINI L,CALISTI M,CORDON O.Auto-matic feature extraction from 3D range images of skulls[J].Compuational Forensics,2008,5158:58-69.

BIAN Z, TONG R.Feature-preserving mesh denoising based on vertices classidication[J].Computer Aided Geometric Design,2008, 28(1):50-64.

许婉露,李彬,田联房. 基于MC算法的高质量脊柱CT图像三维重建[J]. 计算机应用与软件,2013,30(6):79-92.

XU W L,LI B,TIAN L F. High-quality 3D reconstruction of spine CT image based on MC algorithm[J].Computer Applications and Software,2013,30(6):79-92.(in Chinese)

李康,尚鹏,耿国华. 基于深度图像的人脸模型特征点自动标定[J]. 计算机科学,2014,41(4):287-291.

LI K,SHANG P, GENG G H.. Automatic location of feature points on three-dimensional facial model based on depth image[J]. Computer Science,2014,41(4):287-291.(in Chinese)

翟凤文,党建武,王阳萍,等. 形状特征和置信传播在去除SIFT特征点错误匹配中的应用[J]. 计算机辅助设计与图形学学报,2016,28(3).443-449.

ZHAI F W,DANG J W, WANG Y P, et al.. Application of shape context and belief propagation in removing SIFT mismatches[J]. Journal of Computer-Aided Design & Computer Graphics,2016,28(3):443-449.(in Chinese)

谢易辰,陈健,闫镔,等. 三维特征点距离特征集合求交匹配方法[J]. 红外与激光工程,2014,43(8):2728-2732.

XIE Y CH,CHEN J, YAN B, et al..Distance feature set intersection for 3D feature matching[J].Infrared and Laser Engineering,2014,43(8):2728-2732.(in Chinese)

王澜,孙博,隆昌宇,等. 一种基于共线特征点的线阵相机内参标定方法[J]. 红外与激光工程,2015,44(6):1878-1883.

WANG L,SUN B, LONG CH Y, et al..A novel method for calibrating intrinsic parameters of linear array cameras based on collinear feature points[J].Infrared and Laser Engineering, 2015,44(6):1878-1883.(in Chinese)

WANG J, ZHANG X, YU Z.A cascaded approach for feature-preserving surface mesh denoising[J].Computer-Aided Design,2012, 44(7):597-610.

韩霜. 三维脊椎模型特征点标注的研究与实现[D]. 西安:西安电子科技大学,2013.

HAN SH.Research and implementation of tagging three-dimensional model of the spine with feature points[D].Xi'an:Xidian University,2013.(in Chinese)

JING X,SIMON B,IAIN M, et al..Real-time combined 2D+3D active appearance model[J].Proceeding of the 2004 IEEE Computer Society Conference on Computer Vision and Pattern Recognition,2004,27(2):535-542.

TEKALP A M, OSTERMANN J. "Face and 2-D Mesh animation in MPEG-4,"Tutorial Issue On The MPEG-4 Standard[J].Image Communication Journal, Elsevier, 1999.

李瑞祥. 实用人体解剖彩色图谱[M] 北京:人民卫生出版社,2001:6.

LI R X.Color Maps of Practical Human Anatomical[M].Beijing:People's Medical Publishing House,2001:6.(in Chinese)

陈维桓. 微分几何[M]. 北京:北京大学出版社,2006.

CHEN W H. Differential Geometry[M]. Beijing:Peking University Press,2006.(in Chinese)

浏览量

428

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

暂无数据