计算机工程与应用 ›› 2010, Vol. 46 ›› Issue (3): 151-153.DOI: 10.3778/j.issn.1002-8331.2010.03.046

• 图形、图像、模式识别 • 上一篇    下一篇

序列图像自适应增强目标定位算法研究

侯一民,伦向敏,兰建军   

  1. 东北电力大学 自动化工程学院,吉林市 132012
  • 收稿日期:2008-07-24 修回日期:2008-11-14 出版日期:2010-01-21 发布日期:2010-01-21
  • 通讯作者: 侯一民

Research on object Adaboost location algorithm for image sequence

HOU Yi-min,LUN Xiang-min,LAN Jian-jun   

  1. School of Automation Engineering,Northeast Dianli University,Jilin 132012,China
  • Received:2008-07-24 Revised:2008-11-14 Online:2010-01-21 Published:2010-01-21
  • Contact: HOU Yi-min

摘要: 针对包含某个目标的序列图像,提出了一种基于自适应增强(Adaboost)分类的目标定位方法。在先验知识条件下,运用自适应增强方法将初始目标位置附近的像素分类为目标和背景两部分。对目标像素,运用形态学滤波去除误分类及噪声,此后运用图像的矩特征寻找目标中心,得到目标定位结果。算法基本消除了由于目标包含像素值丰富,而对图像分割产生的不利影响。实验证明,该算法效果较好。

关键词: 自适应增强, 目标定位, 形态学,

Abstract: The paper proposes a location method for an object obtained in an image sequence based on Adaboost.The Adaboost is employed to classify the pixel around the object into the object part and the scenery part according to the prior knowledge.The morphology is used to adapt the miss classification and noise for the object pixels.According to the object moments,the center of the object,that is the location result,is obtained.The algorithm solves the effect of complexity pixel value in the procedure image segmentation.The experiment results show that the proposed method is effective.

Key words: Adaboost, object location, morphology, moments

中图分类号: