CE 
'AN 
n medical 
important 
y presents 
( walking 
CD video 
recorded 
imple 3-D 
ment with 
tures. 
  
  
  
1. INTRODUCTION 
The information given by  biostereometrics is 
expected to become more important in medical 
rehabilitation, kinesiology | and biomechanics 
which is an application of principles of Newton's 
dynamics to analyze a human movement. The 
developed system "New Bird Man" has a powerful 
biostereometrics method for CCD video camera 
calibration and orientation. The 3-D digital data 
of major joints of walking twenty persons were 
obtained. 
Fig.1 shows a normal person walking (Case A). 
Fig. 2 shows a person who has experimentally 
fixed by orthosis (Case B). Fig. 3 shows stereo 
image of a person who has experienced a broken leg 
(Case C). The raw serial measurement data are 
difficult to understand about their walking 
features directly because of their complex 
fluctuations. In order to analyze the dynamical 
relations in human movement, this study presents a 
new efficient ^ approach to spectral analysis 
utilizing Auto Regressive Modeling — (AR- 
Modeling). 
2. METHODS AND MATERIALS 
mex eye E AH SON — m à 4 OE RET ea 
Henle = ; 
    
  
  
  
  
Y-axis 
Z-axis 
CCD Video Cameras 
X-axis 
: time monitor 
monitor generator 
  
  
  
  
  
  
  
  
  
  
  
  
HEC 
video recorder video recorder 
Fig. 3 Synchronized 
A. Synchronized Recording System 
The synchronized recording system for stereo image 
composed to two CCD Video camera, recorders and a 
time generator, which solve the stereo image 
matching. After recorded simultaneously a real- 
time control symbol, converted into time letter on 
each right and left serial in a ratio of thirty flames 
per 1 second. Five GCP (24 Ground Control Points) 
poles for camera calibration and orientation were 
set on the floor in the person's moving area, which 
have three points respectively. 
B. New Bird Man System for Biostereometrics 
Unknown 3-D data of body joint are able to analyze 
automatically by New Bird Man System. The flow 
of the system protocol is shown as follows. 
[1] Coordinates Transformation for left and right 
Video Camera Orientation 
Pixel Coordinate (512 X 480) of left and right GCPs 
were transformed to video graphic coordinates. 
[2] Optimization by Least Squares Method for 
Estimation of Orientation Parameters 
Approximate values of parameters were given to 
simultaneous equations by many collinearity 
equations made from GCPs's coordinates. That can 
make a estimation of parameters and reduction of a 
calculation time. The collinearity equations are 
defined as follows: 
  
  
- F(X,.Y,Z o, p, k,dx) Q.B.1) 
y gn X) a(Y-Y)«a,(Z-Z) =, 
aix-X)xalr=-Y}50 (2-7) 
= F(X,,Y,Z,0,p,k,dy) Q.B.2) 
(Calibration Model) 
dx= x +x(dr +dr‘+ dr‘) 
dy = Y T y (dr + dr + dre) (2.B.3) 
X =x+xX,y =y+y,r"=(x/F) +(y/F) 
Orientation parameters 
three rotation angles: 
projection center location: (x y. Z) 
focal distance: f 
principal point difference: (xy, ) 
lens distortion: d.d.d 
rotation matrices: a 
567 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B5. Vienna 1996