Regressing Local to Global Shape Properties for Online Segmentation and Tracking

被引:7
作者
Ren, Carl Yuheng [1 ]
Prisacariu, Victor [1 ]
Reid, Ian [1 ]
机构
[1] Univ Oxford, Dept Engn Sci, Oxford OX1 3PJ, England
关键词
Occlusion recovery; Incremental learning; Level-set based tracking; Discrete cosine transform; VISUAL TRACKING; MODELS; FRAMEWORK; OCCLUSION;
D O I
10.1007/s11263-013-0635-y
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
We propose a novel regression based framework that uses online learned shape information to reconstruct occluded object contours. Our key insight is to regress the global, coarse, properties of shape from its local properties, i.e. its details. We do this by representing shapes using their 2D discrete cosine transforms and by regressing low frequency from high frequency harmonics. We learn this regression model using Locally Weighted Projection Regression which expedites online, incremental learning. After sufficient observation of a set of unoccluded shapes, the learned model can detect occlusion and recover the full shapes from the occluded ones. We demonstrate the ideas using a level-set based tracking system that provides shape and pose, however, the framework could be embedded in any segmentation-based tracking system. Our experiments demonstrate the efficacy of the method on a variety of objects using both real data and artificial data.
引用
收藏
页码:269 / 281
页数:13
相关论文
共 39 条
[1]  
Adam A., 2006, IEEE C COMPUTER VISI, V1, P798, DOI [DOI 10.1109/CVPR.2006.256, 10.1109/CVPR.2006.256]
[2]  
Bibby C, 2008, LECT NOTES COMPUT SC, V5303, P831, DOI 10.1007/978-3-540-88688-4_61
[3]  
Blaschko MB, 2008, LECT NOTES COMPUT SC, V5302, P2, DOI 10.1007/978-3-540-88682-2_2
[4]   Adaptive Fragments-Based Tracking of Non-Rigid Objects Using Level Sets [J].
Chockalingam, Prakash ;
Pradeep, Nalin ;
Birchfield, Stan .
2009 IEEE 12TH INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV), 2009, :1530-1537
[5]  
Corduneanu A., 2001, Artificial intelligence and statistics, P27
[6]  
Cremers D, 2004, LECT NOTES COMPUT SC, V3175, P36
[7]   A review of statistical approaches to level set segmentation: Integrating color, texture, motion and shape [J].
Cremers, Daniel ;
Rousson, Mikael ;
Deriche, Rachid .
INTERNATIONAL JOURNAL OF COMPUTER VISION, 2007, 72 (02) :195-215
[8]   A framework for image segmentation using shape models and kernel space shape priors [J].
Dambreville, Samuel ;
Rathi, Yogesh ;
Tannenbaum, Allen .
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2008, 30 (08) :1385-1399
[9]   Model learning for robot control: a survey [J].
Duy Nguyen-Tuong ;
Peters, Jan .
COGNITIVE PROCESSING, 2011, 12 (04) :319-340
[10]  
Fritz M, 2005, IEEE I CONF COMP VIS, P1363