Ams Bs e P ren Pi A e e n
2. METHODS
2.1 System of 3-D Videogrammetry and Image Processing for two non-metric CCD Video Cameras "New BirdMan
2.1.1 Orientation and Self-Calibration using Bundle Adjustment
After left and right video images were A/D converted, the coordinates of both image pixel value of twelve GCPs (ground
control points) were transformed to photographic coordinates, and then, oriented using the system "New Bird Man" which
is simple 3-D videogrammetric analyzer using the bundle adjustment by collinearity equation and self calibration. The
collinearity equation is defined as:
xzcf au (X - XS) a; (Y - X) x a4(Z — Z;) + dk 5s F(X,. Y,. Z,, c), 9. c, dx) (1)
a (X — X,) - a4 (Y - YX,)) c a4(Z — Z,)
X — X, — -
5 f a, ( o) * as (Y Y)+az(Z Zo +dy = F(X,,Y,,Z,,0,0,k,dy) (2)
ay (X Xo) tau(Y-Yı)+ au(Z- Zu)
dx = x, + x (dr* +d,r* +dzr®) dy — y, y (d^ + dar + dar) 5
x =x+X y =y+y, r? =(x'/F)* +(y /F) (4)
exterior orientation parameters: three rotation angles (0,9, K }, projection center location (x, Y 2, )
interior orientation parameters: focal distance , principal point difference {7} lens distortion dd, ‚d; ;
rotation matrices: a; :
2.2.2 3-D Digital Image Processing
After sifting up the stereo matching image intensity, the pixel value of each measurement point on major joints was set
to zero, then binarized, and circumscribed by a numbered window. The left and right images coordinates of each measure-
ment point were detected automatically, using a video reader controller, and then transformed to photographic coordinates
and then to ground coordinates using the result of orientation and "New BirdMan" system. With this system, time-series
digital data from 3-D videogrammetry are able to provide efficient information for biomechanical analysis of human move-
ment. Fig. 3 shows a stereo image of patient sit-to-standing, Fig. 4 shows a binarized human model image by linking each
kinetic center of joint.
P «
‘ \
4 \
- ’ s >
mul. "3-17 S
7 e T ?- >: N
1! "n UL Vs
1 rz} 32/5 3a
ii 4 \ 1 “ 3a
4: ‘ * 1 * A
1 ‘ \ 1 x ACA
“Roo £ + ^ A»
> SS -^ y
TS se 9 "T. z^
s P *? v « >
» > nn y
' ‘
> : né . - » \ /
1 , ^ 1
4, ^
/ . \
4 \
1 e e 1
? 1 1 e.
1 e.-. 22-9 \
* . - d
* e
Fig. 3. Stereo Image of Patient Sit-to-Standing Fig. 4 Binarizzed Human Model Image
2.3 Biomechanical 3-D Analysis of Human Movement
*
2.3.1 Biomechanical Modeling
All movement, including human movement, is the result of the application of forces and is subject to the laws and
principles that govern force and movement. The body segments studied are the feet, legs, thighs, trunk, head, neck, upper
arms, fore arms and hands. Each segment is assigned a length, a mass weight and a mass gravitational center. A rotatory
movement of a segment of the body occurs in a plane and around an axis. Fig. 5 shows the location of center of gravity of
each body segment. Fig. 6 shows three imaginary planes which are arranged perpendicular to each other through the
body, with their axes intersecting at the center of gravity of the body, are called the cardinal planes of the body for human
movement in a 3-D space, s is the sagital plane, h is the horizontal plane and f is the frontal plane.
IAPRS, Vol. 30, Part 5W1, ISPRS Intercommission Workshop “From Pixels to Sequences”, Zurich, March 22-24 1995