作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程

• 人工智能及识别技术 • 上一篇    下一篇

基于变分和互信息的先验形状配准算法

杨平吕,周则明,石汉青,黄 峰   

  1. (解放军理工大学气象海洋学院,南京 211101)
  • 收稿日期:2012-09-28 出版日期:2013-11-15 发布日期:2013-11-13
  • 作者简介:杨平吕(1989-),男,硕士研究生,主研方向:模式识别,图像配准;周则明,副教授、博士;石汉青,教授、博士;黄 峰,教授
  • 基金资助:
    国家自然科学基金资助项目(41174164);国家部委基金资助项目

Prior Shape Registration Algorithm Based on Variation and Mutual Information

YANG Ping-lv, ZHOU Ze-ming, SHI Han-qing, HUANG Feng   

  1. (Institute of Meteorology and Oceanography, PLA University of Science and Technology, Nanjing 211101, China)
  • Received:2012-09-28 Online:2013-11-15 Published:2013-11-13

摘要: 针对带先验形状约束的几何活动轮廓模型中的形状配准问题,提出一种基于变分方法和最大互信息准则的先验形状配准算法。利用变分配准模型计算仿射变换参数,将其作为互信息配准算法的初值,通过Powell优化算法计算仿射变换参数的最优解。实验结果表明,该算法在保证配准精度的同时,能明显提高计算效率。

关键词: 先验形状, 图像配准, 变分, 互信息, 梯度下降流, 仿射变换

Abstract: Aiming at the registration problem of Geometric Active Contour(GAC) model with prior shape, this paper proposes a prior shape registration algorithm based on variation method and maximum mutual information criterion. After calculating the affine transform parameters with the variation method, the results are used as the initial values of Powell optimization algorithm to maximize the mutual information between the reference and floating images. Experimental results demonstrate that the proposed algorithm can improve the computational efficiency while maintaining a high registration precision.

Key words: prior shape, image registration, variation, mutual information, gradient descent flow, affine transformation

中图分类号: