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

• 图形图像处理 • 上一篇    下一篇

基于在线回归学习的轮廓跟踪算法

沈宋衍,陈莹   

  1. (江南大学轻工过程先进控制教育部重点实验室,江苏 无锡 214122)
  • 收稿日期:2015-02-27 出版日期:2016-05-15 发布日期:2016-05-13
  • 作者简介:沈宋衍(1989-),男,硕士研究生,主研方向为计算机视觉;陈莹,副教授。
  • 基金资助:
    国家自然科学基金资助项目(61104213);江苏省自然科学基金资助项目(BK2011146)。

Contour Tracking Algorithm Based on Online Regression Learning

SHEN Songyan,CHEN Ying   

  1. (Key Laboratory of Advanced Process Control for Light Industry,Ministry of Education, Jiangnan University,Wuxi,Jiangsu 214122,China)
  • Received:2015-02-27 Online:2016-05-15 Published:2016-05-13

摘要: 针对目标快速运动以及严重形变导致跟踪失败的问题,基于在线回归学习提出一种轮廓跟踪算法。以当前跟踪区域为中心,通过循环矩阵对其进行循环平移形成训练样本,进行基于核的相关性回归训练。检测帧根据上一帧回归模型计算待测区域与目标区域的相关频域。将相关矩阵返回空域,形成目标定位特征图,将其与待测区灰度图进行融合形成轮廓置信图。利用置信图作为辅助信息,通过水平集模型提取目标轮廓。设计轮廓评价方案判断轮廓质量,当发生畸变时进行轮廓修正。将轮廓所在位置反馈至核相关滤波跟踪器并更新跟踪模板,从而准确得到下一帧特征图。实验结果证明,该方法能快速准确地跟踪目标及其轮廓,并且具有较好的鲁棒性。

关键词: 核相关滤波器, 回归学习, 轮廓跟踪, 水平集, 置信图, 目标定位

Abstract: To solve the problem of fault tracking result from rapid movement and severe deformation,a contour tracking algorithm is proposed.The training samples are generated by cyclic shifts of the current tracking area with a cyclic matrix,which is used for kernel correlation based regression training.According to regression model of last frame,correlation map between target and test area in frequency domain is calculated,and returns to spatial space to form a target position maps.The maps are fused with the gray image of the test frame to establish a contour confident map.Target contour is extracted with active contour model by taking contour confident map as auxiliary information.When the contour distortion is detected by a designed distortion evaluation scheme,the contour will be modified in the next frame.The tracking result is then feedback to the kernalized correlation filter,and helps to update the tracking template.Experimental results show that in various tracking cases,the proposed method achieves more accurate objects position and contour tracking results than other state-of-the-arts methods with better robustness.

Key words: Kernel Correlation Filter(KCF), regression learning, contour tracking, level set, confident map, target position

中图分类号: