3DBODY.TECH 2017 - Paper 17.281

B.-K. D. Park and M. P. Reed, "A Model-based Approach to Rapid Estimation of Body Shape and Postures Using Low-Cost Depth Cameras", in Proc. of 3DBODY.TECH 2017 - 8th Int. Conf. and Exh. on 3D Body Scanning and Processing Technologies, Montreal QC, Canada, 11-12 Oct. 2017, pp. 281-287, doi:10.15221/17.281.


A Model-based Approach to Rapid Estimation of Body Shape and Postures Using Low-Cost Depth Cameras


Byoung-Keon D. PARK, Matthew P. REED


Depth cameras have revolutionized anthropometry providing efficient ways to gather 3-dimensional information of a human body in varied settings. However, the accuracy and resolution of the current depth camera systems limits their applicability for many applications. Commercial markerless motion-capture systems based on depth camera technology are similarly limited, particularly by the lack of a model-based tracking mechanism and a high dependence on unobstructed camera views. This paper presents a series of model-based methods for estimating body configuration and postures based on depth and posture data obtained from a single Kinect v2 sensor. The software system records and processes multiple depth images of a person from different point of views to capture the whole-body shape. A statistical body shape model that can represent a wide variety of human body shapes and poses was generated by analyzing template-fit whole-body laser scans and measured anatomical landmark data using a principal component analysis (PCA). PCA reduces the high dimensionality of the original data source by projecting the dataset to a low dimensional principal component (PC) space. In the PC space, only realistic body shapes and landmark data can be generated, and this space allows for efficient body shape search due to the low dimensionality. Using this model, a rapid fitting method for generating a subject-specific manikin from Kinect depth data was developed that can estimate a minimally-clad body shape under normally clothing. Posture data from subsequent movements estimated by the built-in skeleton tracker in the Kinect system were further improved by fitting each body segment of the manikin to a corresponding partial depth dataset while the segment lengths were preserved as defined in the manikin. This study demonstrated how a model-based approach can improve the accuracy and feasibility of a depth camera system so that the system can be readily applied for various applications, including in-vehicle occupant dynamic analysis, occupant classification, and markerless motion analysis.


Full paper: 17.281.pdf
Proceedings: 3DBODY.TECH 2017, 11-12 Oct. 2017, Montreal QC, Canada
Pages: 281-287
DOI: 10.15221/17.281

Copyright notice

© Hometrica Consulting - Dr. Nicola D'Apuzzo, Switzerland, www.hometrica.ch.
Reproduction of the proceedings or any parts thereof (excluding short quotations for the use in the preparation of reviews and technical and scientific papers) may be made only after obtaining the specific approval of the publisher. The papers appearing in the proceedings reflect the author's opinions. Their inclusion in these publications does not necessary constitute endorsement by the editor or by the publisher. Authors retain all rights to individual papers.

Proceedings of 3DBODY.TECH International Conferences on 3D Body Scanning & Processing Technologies, © Hometrica Consulting, Switzerland