English

应用科学学报

• 论文 • 上一篇    下一篇

边界轮廓的多分辨率弹性配准

荣成城1 周 健2 曹国刚1 罗立民1   

  1. 1. 东南大学影像科学与技术实验室，江苏 南京210096；
    2.法国国家健康研究院 信号与图像处理实验室，法国 雷恩 35000
  • 收稿日期:2007-04-03 修回日期:2007-08-20 出版日期:2007-11-30 发布日期:2007-11-30

Multi-resolution Elastic Registration Using Edge Contour Featrues

RONG Cheng-cheng1, ZHOU Jian2, CAO Guo-gang1, LUO Li-min1   

  1. 1. Laboratory of Image Science and Technology, Southeast University, Nanjing 210096, China;
    2. U642 INSERM－Laboratoire Traitement du Signal et de l’Image (LTSI), Rennes 35000, France
  • Received:2007-04-03 Revised:2007-08-20 Online:2007-11-30 Published:2007-11-30

摘要： 提出一种基于边界轮廓的多分辨率弹性配准方法。首先用C-V（Chan-Vese）模型提取人脑图像的边界轮廓。接着使用模拟退火方法结合B样条插值函数来配准边界轮廓点集，在退火过程中通过使用多分辨率策略，在获得精细边界轮廓的同时加快了配准速度。最后利用B样条插值函数完成图像配准。实验结果表明，该算法在保证与同类算法相同精度的前提下，实现了快速弹性配准。

关键词: C-V模型, B样条, 模拟退火, 多分辨率, 弹性配准

Abstract: A multi-resolution elastic registration method based on edge contour is presented. The C-V（Chan-Vese） model is applied to extract source and target contours from two human brain MR images. A multi-resolution simulated annealing procedure is proposed to match the two contour point sets with low computation cost. This way, a fast elastic registration method using B-spline interpolation function is obtained. Experimental results show that the proposed method provides fast image registration without loss of precision.

Key words: C-V model, B-spline, simulated annealing, multi-resolution, elastic registration