ZTE

ZTE Communications ›› 2012, Vol. 10 ›› Issue (4): 54-59.

• Research Paper • Previous Articles     Next Articles

Hierarchical Template Matching for Robust Visual Tracking with Severe Occlusions

Lizuo Jin1, Tirui Wu2, Feng Liu3, and Gang Zeng3   

  1. 1. School of Automation, Southeast University, Nanjing 210096, China;
    2. ZTE Corporation, Nanjing 210012, China;
    3. ZTE Corporation, Chongqing 401121, China
  • Received:2012-08-03 Online:2012-12-25 Published:2012-12-25
  • About author:Lizuo Jin (jinlizuo@gmail.com) received his Ph.D. degree in Pattern Recognition and Intelligent System from Southeast University, Nanjing, China, in 2000. From 2002 to 2004, he was a post-doctoral fellow at the Institute of Industrial Technology, Tokyo University, Japan. He is now an associate professor at School of Automation, Southeast University, and also a member of Information Fusion Technical Committee of CSAA, China. His research interests include theory and methods for machine learning, pattern recognition, computer vision and embedded systems.

    Tirui Wu (wu.tirui@zte.com.cn) received his master degree in Computer Science and Engineering from Jiangsu University, Zhenjiang, China. He is now a researcher manager with Nanjing Institute of ZTE corporation. His research interests include theory and application for computer vision, human machine interface, image fusion and GPS signal processing.

    Feng Liu (liu.feng90@zte.com.cn) received his master degree in Computer Science and Engineering from Chongqing University of Posts and Telecommunications, Chongqing, China. He is now a researcher manager with Chongqing Institute of ZTE corporation. His research interests include theory and application for pattern recognition and artificial intelligence.

    Gang Zeng (zeng.gang@zte.com.cn) received his master degree in Computer Science and Engineering from Chongqing University, Chongqing, China. He is now a researcher manager with Chongqing Institute of ZTE corporation. His research interests include theory and application for software engineering and computer vision.
  • Supported by:
    This work is supported by the Aeronautical Science Foundation of China under Grant 20115169016.

Hierarchical Template Matching for Robust Visual Tracking with Severe Occlusions

Lizuo Jin1, Tirui Wu2, Feng Liu3, and Gang Zeng3   

  1. 1. School of Automation, Southeast University, Nanjing 210096, China;
    2. ZTE Corporation, Nanjing 210012, China;
    3. ZTE Corporation, Chongqing 401121, China
  • 作者简介:Lizuo Jin (jinlizuo@gmail.com) received his Ph.D. degree in Pattern Recognition and Intelligent System from Southeast University, Nanjing, China, in 2000. From 2002 to 2004, he was a post-doctoral fellow at the Institute of Industrial Technology, Tokyo University, Japan. He is now an associate professor at School of Automation, Southeast University, and also a member of Information Fusion Technical Committee of CSAA, China. His research interests include theory and methods for machine learning, pattern recognition, computer vision and embedded systems.

    Tirui Wu (wu.tirui@zte.com.cn) received his master degree in Computer Science and Engineering from Jiangsu University, Zhenjiang, China. He is now a researcher manager with Nanjing Institute of ZTE corporation. His research interests include theory and application for computer vision, human machine interface, image fusion and GPS signal processing.

    Feng Liu (liu.feng90@zte.com.cn) received his master degree in Computer Science and Engineering from Chongqing University of Posts and Telecommunications, Chongqing, China. He is now a researcher manager with Chongqing Institute of ZTE corporation. His research interests include theory and application for pattern recognition and artificial intelligence.

    Gang Zeng (zeng.gang@zte.com.cn) received his master degree in Computer Science and Engineering from Chongqing University, Chongqing, China. He is now a researcher manager with Chongqing Institute of ZTE corporation. His research interests include theory and application for software engineering and computer vision.
  • 基金资助:
    This work is supported by the Aeronautical Science Foundation of China under Grant 20115169016.

PDF

Abstract: To tackle the problem of severe occlusions in visual tracking, we propose a hierarchical template-matching method based on a layered appearance model. This model integrates holistic- and part-region matching in order to locate an object in a coarse-to-fine manner. Furthermore, in order to reduce ambiguity in object localization, only the discriminative parts of an object’s appearance template are chosen for similarity computing with respect to their cornerness measurements. The similarity between parts is computed in a layer-wise manner, and from this, occlusions can be evaluated. When the object is partly occluded, it can be located accurately by matching candidate regions with the appearance template. When it is completely occluded, its location can be predicted from its historical motion information using a Kalman filter. The proposed tracker is tested on several practical image sequences, and the experimental results show that it can consistently provide accurate object location for stable tracking, even for severe occlusions.

Key words: visual tracking, hierarchical template matching, layered appearance model, occlusion analysis

摘要： To tackle the problem of severe occlusions in visual tracking, we propose a hierarchical template-matching method based on a layered appearance model. This model integrates holistic- and part-region matching in order to locate an object in a coarse-to-fine manner. Furthermore, in order to reduce ambiguity in object localization, only the discriminative parts of an object’s appearance template are chosen for similarity computing with respect to their cornerness measurements. The similarity between parts is computed in a layer-wise manner, and from this, occlusions can be evaluated. When the object is partly occluded, it can be located accurately by matching candidate regions with the appearance template. When it is completely occluded, its location can be predicted from its historical motion information using a Kalman filter. The proposed tracker is tested on several practical image sequences, and the experimental results show that it can consistently provide accurate object location for stable tracking, even for severe occlusions.

关键词: visual tracking, hierarchical template matching, layered appearance model, occlusion analysis