吉林大学学报(信息科学版) ›› 2015, Vol. 33 ›› Issue (5): 544-.

• 论文 • 上一篇    下一篇

基于模板匹配与线性预测的目标跟踪算法

闫增成, 宋路   

  1. 长春理工大学电子信息工程学院, 长春130022
  • 收稿日期:2014-12-05 出版日期:2015-09-30 发布日期:2015-12-30
  • 作者简介:闫增成(1989—), 男, 山东泰安人, 长春理工大学硕士研究生, 主要从事图像处理、光通信与无线通信研究, (Tel)86-15714314592(E-mail)yzch600@163. com; 宋路(1952—), 女, 长春人, 长春理工大学教授, 博士生导师, 主要从事光通信与信息处理、图像处理研究, (Tel)86-13134460587(E-mail)usong@126. com。
  • 基金资助:

    教育部高校博士学科点专项基金资助项目(20132216110006); 国家自然科学基金重大计划培育基金资助项目(91338116)

Target Tracking Algorithm Based on Template Matching and Linear Prediction

YAN Zengcheng, SONG Lu   

  1. College of Electronic Information Engineering, Changchun University of Science and Technology, Changchun 130022, China
  • Received:2014-12-05 Online:2015-09-30 Published:2015-12-30

摘要:

为解决复杂场景中目标遮挡、算法鲁棒性差的问题, 提出基于模板匹配与线性预测的算法。该算法分为模板匹配和线性预测两部分。场景中无遮挡物时, 执行模板匹配算法, 获取目标跟踪信息; 否则, 通过预测算法估计目标的状态, 进而实现目标跟踪。实验结果表明, 在目标有遮挡干扰的情况下, 采用该算法能稳定、准确地实现目标跟踪, 与传统的模板匹配算法相比具有较强的鲁棒性。

关键词: 边缘检测, 模板匹配, 线性预测, 目标跟踪

Abstract:

In order to solve the complex object shelter in the scene and poor robustness, we put forward the algorithm based on template matching with linear prediction. The algorithm includes two parts, template matching and linear prediction. When there no obstructions in the scene, the executive part of template matches algorithm for geting target tracking information; otherwise, through the process of prediction algorithm to estimate the states of the target, and then to realize tracking. The results show that the method can track target accurately and stably for the target to keep out interference, comparing with the traditional template matching algorithm it has strong robustness.

Key words: edge detection, template matching, linear prediction, target tracking

中图分类号: 

  • TP311