Human pose estimation from depth image using visibility estimation and key points

Authors:
Sungjin Huh;Gyeonghwan Kim
Affiliations:
Department of Electronic Engineering, Sogang University, Seoul, Rep. of Korea;Department of Electronic Engineering, Sogang University, Seoul, Rep. of Korea
Venue:
DHM'13 Proceedings of the 4th international conference on Digital Human Modeling and Applications in Health, Safety, Ergonomics, and Risk Management: human body modeling and ergonomics - Volume Part II
Year:
2013

Citing 8
Cited 0

A Method for Registration of 3-D Shapes

IEEE Transactions on Pattern Analysis and Machine Intelligence - Special issue on interpretation of 3-D scenes—part II
Multigrid

Multigrid
Random Forests

Machine Learning
Human pose estimation from a single view point

Human pose estimation from a single view point
Nonlinear body pose estimation from depth images

PR'05 Proceedings of the 27th DAGM conference on Pattern Recognition
3D body pose estimation using an adaptive person model for articulated ICP

ICIRA'11 Proceedings of the 4th international conference on Intelligent Robotics and Applications - Volume Part II
Human skeleton tracking from depth data using geodesic distances and optical flow

Image and Vision Computing
Graph cuts optimization for multi-limb human segmentation in depth maps

CVPR '12 Proceedings of the 2012 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper, we propose the upper body pose estimation algorithm using 3-dimensional model and depth image. The conventional ICP algorithm is modified by adding visibility estimation and key points - extreme points and elbow locations. The visibility estimation keeps occluded points from participating in pose estimation to alleviate the affection of self-occlusion problem. Introduction of extreme points and elbow locations, which are extracted using geodesic distance map and particle filter, improves the accuracy of pose estimation result. The optimal parameters of the model are obtained from nonlinear mathematical optimization solver. The experimental results show that the proposed method accurately estimates the various human poses with self-occlusion.