Reprojection error based annealed particle filter for human upper body pose reconstruction

3D human pose reconstruction is a key concern in computer vision area in recent years. Due to the deficiency of depth information, reconstructing human pose from a single image or image sequences is still a difficult and challenging task. In this paper, we present an annealed particle filter algorithm based on reprojection error to recover the 3D configuration of human upper body, with the annotated joints' position in the image. In addition, we make simplifications to the weak perspective projection, and the 7 camera parameters to be estimated are reduced to only 1. Experiments show that our method is simple but exactly suitable for recovering articulated human upper body pose.

[1]  Baozong Yuan,et al.  Using compact distance energy model to recover human pose in markerless motion capture , 2008 .

[2]  R. Cipolla,et al.  A probabilistic framework for space carving , 2001, Proceedings Eighth IEEE International Conference on Computer Vision. ICCV 2001.

[3]  Yong,et al.  Rapid 3D human body modeling based on Kinect technology , 2013 .

[4]  Andrew Blake,et al.  Articulated body motion capture by annealed particle filtering , 2000, Proceedings IEEE Conference on Computer Vision and Pattern Recognition. CVPR 2000 (Cat. No.PR00662).

[5]  T. Kanade,et al.  Reconstructing 3D Human Pose from 2D Image Landmarks , 2012, ECCV.

[6]  Nanning Zheng,et al.  A new implementation of image-processing engine for 3D visualization and stereo video stream display , 2014 .

[7]  Jian Song,et al.  Parallelized Annealed Particle Filter for real-time marker-less motion tracking via heterogeneous computing , 2012, Proceedings of the 21st International Conference on Pattern Recognition (ICPR2012).

[8]  Bernt Schiele,et al.  Pictorial structures revisited: People detection and articulated pose estimation , 2009, 2009 IEEE Conference on Computer Vision and Pattern Recognition.