J4 ›› 2016, Vol. 43 ›› Issue (1): 116-119.doi: 10.3969/j.issn.1001-2400.2016.01.021

• 研究论文 • 上一篇    下一篇

三维变形体的因式分解重建方法

彭亚丽1,2;刘侍刚2;梁新刚2;廖福轩2   

  1. (1. 陕西师范大学 现代教学技术教育部重点实验室,陕西 西安  710062;
    2. 陕西师范大学 计算机科学学院, 陕西 西安  710119)
  • 收稿日期:2014-08-12 出版日期:2016-02-20 发布日期:2016-04-06
  • 通讯作者: 彭亚丽
  • 作者简介:彭亚丽(1979-),女,讲师,E-mail: pengyl@snnu.edu.cn.
  • 基金资助:

    国家自然科学基金资助项目(61402274,41271387);中国博士后科学基金特别资助项目(200902594);中央高校基本科研业务费专项资金资助项目(GK201402040,GK201302029);陕西师范大学学习科学交叉学科培育计划资助项目;陕西师范大学实验技术研究资助项目(SYJS201314);陕西师范大学大学生创新创业训练计划资助项目(cx14023)

Factorization reconstruction method for the 3D deformable structure

PENG Yali1,2;LIU Shigang2;LIANG Xingang2;LIAO Fuxuan2   

  1. (1. Key Lab. of Modern Teaching Technology of Ministry of Edu., Shaanxi Normal Univ., Xi'an  710062, China;
    2. School of Computer Science, Shaanxi Normal Univ., Xi'an  710119, China;)
  • Received:2014-08-12 Online:2016-02-20 Published:2016-04-06
  • Contact: PENG Yali

摘要:

为了从非标定图像序列中重建出三维非刚体的结构,提出了一种三维变形体的因式分解重建方法.该方法首先利用图像矩阵为低秩矩阵的特性,通过奇异值分解求到射影重建,利用投影矩阵的正交约束,实现射影重建到欧氏重建的过渡.该方法的优点是,重建过程全部为线性求解,且将图像及图像点平等对待.模拟实验和真实实验数据结果表明,该重建方法具有鲁棒性好和重投影误差小等优点.

关键词: 变形体, 重建, 因式分解

Abstract:

To recover the 3D deformable structure from an un-calibration image sequence, a factorization reconstruction method for 3D deformable reconstruction is presented. Based on the low rank constraint on the image matrix, the projective reconstruction can be obtained by single value decomposition. By using the orthogonality of the projective matrix, the Euclidean reconstruction is upgraded from projective reconstruction. The innovation of the method is that the solving of the 3D non-rigid reconstruction is linear and that all the images and the image points are treated uniformly. The experiments with both simulated and real data show that the method presented in the paper is efficient.

Key words: deformable structure, reconstruction, factorization