P2-3-3
3.2 Auto-Tracking
For human motion analysis, auto-tracking of the feature point is
performed as follows:
1) For differencing an image sequences, an image for the
background is taken.
2) Window is previously cut out around the feature area in the
first sequential image.
3) Differencing an image sequence are performed between the
background image and the first sequential image.
4) Extraction of feature area by opening.
5) Calculation of area and area gravity for extracted feature area.
6) Area gravity is regarded as the feature point.
7) Position of the feature point at the first image is decided by
operator.
8) Feature point is circled for the purpose of certifying the
position.
9) These procedures are repeated. But the position of the feature
point from the second image is decided by comparing area and
area gravity between the present image and the before image.
When the feature point can not be extracted owing to the feature
area is hidden in steel pipe or due to insufficient image quality,
"LOST" message is expressed on PC monitor. But if the feature
area appear again on the PC monitor, the feature area is extracted
and calculated the position automatically, and massage
correspond to a kind of work is expressed on the PC monitor.
Work distinction and auto-tracking of the feature point are
usually performed simultaneously at 2 frame intervals in this
ergoma system. But during welding and burnishing, those
procedures are performed at 10 frame intervals since a man
doesn't move during these works.
3.3 Work Analysis
Work analysis was performed using video images for horizontal
and vertical steel pipes. These images were taken for 23 minutes
respectively and include welding and burnishing. Figure 6(a)
shows work analysis for horizontal steel pipe and figure 6(b)
shows work analysis for vertical steel pipe. In these figures,
"OPERATOR" means that the feature point was extracted by
operator, "LOST" means that the feature point couldn't be find
out, and "MOVEMENT" means that the feature point changed
for work or some reasons.
It may be seen from these figures that the "MOVEMENT" ratio
for horizontal steel pipe is larger than vertical steel pipe. It is
suspected that this "MOVEMENT" ratio was caused by the pipe
type, i.e. man should move under the pipe or long round about the
pipe in horizontal case. Then, it can be said that lower
"MOVEMENT" means better work efficiency.
Furthermore, it is ascertained that 92 % for all extracting the
feature area were performed automatically, and it is not requested
the intervention of the operator.
Therefore, it is concluded that the extraction method developed
in this paper is effective method for extracting feature area.
3.4 Motion Analysis
Human motion analysis for welding and burnishing about
horizontal and vertical steel pipe were performed using three
dimensional feature points that were obtained by
photogrammetric techniques.
Figure 7(a) and 7(b) shows the locus of the feature point on X-Z
coordinate for horizontal and vertical steel pipes respectively. On
the other hand, figure 7(c) and 7(d) shows the locus of the feature
point on X-Y coordinate for horizontal and vertical steel pipes
respectively. These figures are the results for ten minutes for each
work. It may be seen from these figure that,
+ In the case of horizontal pipe, man is working on the both side
of pipe within a radius of 1 meter (figure 7(a)), and man is
repeating sitting and standing posture (figure 7(c)).
+ In the case of vertical pipe, man is moving around pipes over
again and working within a radius of 2 meters (figure 7(b)), and
man is working almost with sitting posture (figure 7(d)).
Then, it is evaluated that workspaces for vertical pipe needs
larger area than horizontal pipe. Furthermore, from the point of
view of the improvement of working environment, united part of
pipes for the vertical pipe should be located more upper position
to avoid a load of waist
OPERATOR LOST
(a) Horizontal steel pipe
44%
MOVEMENT
39%
WELDING
OPERATOR
LOST
BURNISHING
9%
(b) Vertical steel pipe
Figure 6 Work analysis