Computer Engineering and Applications ›› 2007, Vol. 43 ›› Issue (33): 41-42.

• 学术探讨 • Previous Articles     Next Articles

Target recognition method for sequential images

XIU Chun-bo   

  1. Department of Automatic Control,Tianjin Polytechnic University,Tianjin 300160,China
  • Received:1900-01-01 Revised:1900-01-01 Online:2007-11-21 Published:2007-11-21
  • Contact: XIU Chun-bo

序列图像的目标识别方法

修春波   

  1. 天津工业大学 自动化系,天津 300160
  • 通讯作者: 修春波

Abstract: A new target recognition method for sequential images is proposed to improve the rotation invariance and reduce the matching frequency for minimum absolute difference.The feature points in template and tracking window are detected and the areas around the feature points in template are chosen as feature templates.Feature points matching between tracking window and template can be completed by ring-shaped template matching operation.The lost feature points in tracking window could be determined according to the positions of the feature points matched.Simulation results prove the validity of the method.

Key words: target recognition, template matching, sequential images

摘要: 针对灰度差绝对平均值算法匹配次数多,不具有旋转不变性等缺点,提出一种新的序列图像目标识别方法。首先,分别在模板图像和目标图像中进行特征点检测,然后将模板图像中检测到的特征点附近区域作为特征模板。利用特征模板与目标图像中的特征点相应区域进行环形模板匹配操作,从而在目标图像中找到与模板图像相匹配的特征点。对于因遮挡而丢失的特征点,可根据已匹配特征点之间的相对关系来重新确定,从而实现目标识别的功能。仿真实验验证了该算法的有效性。

关键词: 目标识别, 模板匹配, 序列图像