A preview of this full-text is provided by Optica Publishing Group.
Content available from Applied Optics
This content is subject to copyright. Terms and conditions apply.
Research Article Vol. 59, No. 4 / 1 February 2020 / Applied Optics 955
Laser-speckle-projection-based handheld
anthropometric measurement system with
synchronous redundancy reduction
Xiao Yang,1,2Xiaobo Chen,1,2Guangkun Zhai,3AND Juntong Xi1,2,4,*
1School of Mechanical Engineering, Shanghai Jiao Tong University, Shanghai 200240, China
2Shanghai Key Laboratory of Advanced Manufacturing Environment, Shanghai 200030, China
3School of Electronic Information and Electrical Engineering, Shanghai Jiao Tong University, Shanghai 200240, China
4State Key Laboratory of Mechanical System and Vibration, Shanghai 200240, China
*Corresponding author: jtxi@sjtu.edu.cn
Received 11 October 2019; revised 15 December 2019; accepted 15 December 2019; posted 16 December 2019 (Doc. ID 380322);
published 27 January 2020
Human body measurement is essential in modern rehabilitation medicine, which can be effectively combined
with the technology of additive manufacturing. Digital image correlation based on laser speckle projection is a
single-shot, accurate, and robust technique for human body measurement. In this paper, we present a handheld
anthropometric measurement system based on laser speckle projection. A flexible retroreflective marker target is
designed for multi-view data registration. Meanwhile, a synchronous redundancy-reduction algorithm based on
a re-projected global disparity map is proposed. Experiment results validate that the proposed system is effective
and accurate for different human body part measurements. Comparative experiments show that the proposed
redundancy-reduction algorithm has high efficiency and can effectively preserve the features of complex shapes.
The comprehensive performance of the algorithm is better than the other two tested methods. © 2020 Optical
Society of America
https://doi.org/10.1364/AO.380322
1. INTRODUCTION
As a non-contact and precise measurement method, optical
3D measurement has been an important auxiliary method for
modern medical diagnosis and rehabilitation. The acquisition
of 3D shapes for specific body parts is essential in processes
such as body shape measurement and analysis [1–4], aug-
mentation mammoplasty [5], prosthesis design [6,7], fracture
immobilization object design [8,9], etc. As the CAD model
is created via optical 3D measurement, 3D printing technol-
ogy can be directly used to fabricate orthoses or prosthetics.
The 3D printed orthoses or prosthetics are more comfort-
able and effective compared to traditional plaster-made ones,
because 3D printing technology has more flexibility in material
selection and thus can provide products with various physical
properties [10,11].
Structured light stereo vision is a popular optical 3D mea-
surement method because it offers the advantage of large
data capacity to capture the whole field with high precision.
According to the number of projected patterns, structured light
stereo vision can be classified into multi-shot and single-shot
methods. The single-shot measurement method is more suitable
for human body measurement, because it is hard for a person to
keep still even for a very short time. Recent years, stereo vision
based on speckle projection has become a popular single-shot
3D measurement method [12–14], as it is robust to ambient
light and has the same precision as the multi-shot structured
light method [15,16]. Moreover, speckles projected by a laser
source onto human skin have such properties as higher bright-
ness and isotropy, and the contrast and speckle clarity is better
than that projected by digital light processor (DLP) [17,18]. In
recent years, the technique of laser speckle projection has been
utilized for single-shot 3D measurement by Dekiff et al. [19–21]
and Babovsky et al. [22], and found to achieve high precision
and temporal resolution.
Despite the fact that single-shot stereo vision based on laser
speckle projection is an effective solution for human body mea-
surement, it usually needs to register the range data obtained
from multiple views to create a 3D model of the injured part
(waist, back, leg, etc.). Commonly used methods include
iteration-based, multisensor-based, and auxiliary-device-based
registration. The most commonly used iteration-based reg-
istration method is iterative closest point (ICP) [23], which
is based on the principle of local optimization and requires
enough overlapping areas. The registration accuracy of ICP
can be high, but it requires strict initial conditions. If the initial
1559-128X/20/040955-09 Journal © 2020 Optical Society of America
