Multi-view Head Detection and Tracking with Long Range Capability for Social Navigation Planning

Head pose is one of the important human cues in social navigation planning for robots to coexist with humans. Inferring such information from distant targets using a mobile platform is a challenging task. This paper tackles this issue to propose a method for detecting and tracking head pose with the mentioned constraints using RGBD camera (Kinect, Microsoft). Initially possible human regions are segmented out then validated by using depth and Hu moment features. Next, plausible head regions within the segmented areas are estimated by employing Haar-like features with the Adaboost classifier. Finally, the obtained head regions are post-validated by means of their dimension and their probability of containing skin before refining the pose estimation and tracking by a boosted-based particle filter. Experimental results demonstrate the feasibility of the proposed approach for detecting and tracking head pose from far range targets under spot-light and natural illumination conditions.

[1]  King Ngi Ngan,et al.  Face segmentation using skin-color map in videophone applications , 1999, IEEE Trans. Circuits Syst. Video Technol..

[2]  Taner Arsan,et al.  Fast multi-view face tracking with pose estimation , 2008, 2008 16th European Signal Processing Conference.

[3]  Lawrence O. Hall,et al.  Learning to recognize plankton , 2003, SMC'03 Conference Proceedings. 2003 IEEE International Conference on Systems, Man and Cybernetics. Conference Theme - System Security and Assurance (Cat. No.03CH37483).

[4]  G. Buchsbaum.  A spatial processor model for object colour perception , 1980 .

[5]  William D. Smart,et al.  Using depth information to improve face detection , 2011, 2011 6th ACM/IEEE International Conference on Human-Robot Interaction (HRI).

[6]  Paul A. Viola,et al.  Rapid object detection using a boosted cascade of simple features , 2001, Proceedings of the 2001 IEEE Computer Society Conference on Computer Vision and Pattern Recognition. CVPR 2001.

[7]  Thomas Martinetz,et al.  Face Detection Using a Time-of-Flight Camera , 2009, Dyn3D.

[8]  Robert Pless,et al.  Faster and more accurate face detection on mobile robots using geometric constraints , 2007, 2007 IEEE/RSJ International Conference on Intelligent Robots and Systems.

[9]  Ming-Kuei Hu,et al.  Visual pattern recognition by moment invariants , 1962, IRE Trans. Inf. Theory.

[10]  Daijin Kim,et al.  Pose Robust Human Detection in Depth Images Using Multiply-Oriented 2D Elliptical Filters , 2010, Int. J. Pattern Recognit. Artif. Intell..

[11]  Yoichi Sato,et al.  3D Head Tracking using the Particle Filter with Cascaded Classifiers , 2006, BMVC.

[12]  Gengyu Ma,et al.  Multi-view Human Head Detection in Static Images , 2005, MVA.

[13]  B. Schiele,et al.  Fast and Robust Face Finding via Local Context , 2003 .

[14]  Shan Fu,et al.  Stereovision-Based Object Segmentation for Automotive Applications , 2005, EURASIP J. Adv. Signal Process..