计算机工程 ›› 2013, Vol. 39 ›› Issue (8): 200-203.doi: 10.3969/j.issn.1000-3428.2013.08.043

• 人工智能及识别技术 • 上一篇    下一篇

动态时间规整技术在电子稳像中的应用

陈晓露1,2,戴 明1,尹传历1,嵇晓强1,2,王子辰1,2   

  1. (1. 中国科学院长春光学精密机械与物理研究所,长春 130033;2. 中国科学院研究生院,北京 100039)
  • 收稿日期:2012-03-05 出版日期:2013-08-15 发布日期:2013-08-13
  • 作者简介:陈晓露(1986-),女,博士研究生,主研方向:电子稳像技术;戴 明,研究员、博士生导师;尹传历,副研究员、博士;嵇晓强、王子辰,博士研究生
  • 基金项目:
    国家“863”计划基金资助项目(2008AA121803);国家“973”计划基金资助项目(2009CB72400603B)

Application of Dynamic Time Warping Technique in Electronic Image Stabilization

CHEN Xiao-lu 1,2, DAI Ming 1, YIN Chuan-li 1, JI Xiao-qiang 1,2, WANG Zi-chen 1,2   

  1. (1. Changchun Institute of Optics, Fine Mechanics and Physics, Chinese Academy of Sciences, Changchun 130033, China; 2. Graduate University of Chinese Academy of Sciences, Beijing 100039, China)
  • Received:2012-03-05 Online:2013-08-15 Published:2013-08-13

摘要: 当视频序列帧内存在局部运动时,灰度投影算法的稳像精度会有所下降。针对该问题,将动态时间规整技术应用到灰度投影算法中,提出一种新的稳像方法。利用传统的灰度投影算法分别获得图像的行、列投影曲线,通过动态时间规整算法对失准的投影曲线进行修正,得到行、列最佳形变路径,分析最佳匹配点的分布,得到全局运动矢量。实验结果表明,该方法的正确匹配率大于95%,可有效减少误匹配的发生率,提高电子稳像的精度。

关键词: 电子稳像, 灰度投影, 动态时间规整, 运动估计, 灰度段

Abstract: To solve the problem of gray projection algorithm that the accuracy of electronic image stabilization will reduce when there is local movement in the video frame, this paper introduces the Dynamic Time Warping(DTW) technique into to the algorithm. Using the traditional gray projection algorithm to get images’ projection curves on row and column, and correcting them with DTW algorithm, this method can get optimal deformation paths both on row and column and analyze the distribution of the optimal matching points so as to get the global moving vector. Experimental results show that the correct matching rate of this method is above 95%, which reduces occurrence rate of wrong match and improves the accuracy of electronic image stabilization.

Key words: electronic image stabilization, gray projection, Dynamic Time Warping(DTW), motion estimation, gray segment

中图分类号: