feature points without marker was investigated using 
mathematical morphology(Dougherty, 1992). A basic extraction 
flow is shown in Figure 4 and details are described are as 
follows. 
  
Differential Image 
| 
Image Processing 
| 
Image Coordinate 
  
  
  
  
  
  
  
  
End 
  
  
  
Figure 4. Basic Flow of Extraction 
1. Creation of differential image 
+ The background images for whole test course are 
taken by the HVT. 
+ The each background images have H. V. angle 
information by the HVT. 
+ The foreground images while the subject is walking 
are taken with H,V information. 
+ The background image correspond to the 
foreground image is searched using H. V. angle 
information for the foreground. 
+ Differential image is generated using background 
and foreground image. 
+  Binarization of differential image is performed. 
+ Consequently, differential image is acquired 
effectively by using the HVT system. 
2. Skeletonization 
+ Human body is extracted using binary image. 
Edging for extracted human body is performed. 
A circular mask with 1 pixel radius is prepared. 
Enlarging the mask at the every pixels in the body. 
If the parts of mask connect the edge, enlarging is 
finished. 
The connecting number is countered. 
If the mask has more than two connecting points, 
the center of the mask is adopted as a one of 
skeleton. 
+ If the mask has more than three connecting points, 
the center of the mask is adopted as a brunch point 
of skeleton. 
+ + + + 
+ + 
3. Noise reduction 
+ Elimination of isolated point. 
+ Elimination of small area using continuity. 
4. Identification of feature points 
+ The ankle is identified as the lowest brunch point on 
the each skeleton. 
+ The heel and tiptoe are identified as the two 
terminal points. But, tiptoe have larger X coordinate. 
Figure 5 shows the Skeleton image which was acquired by the 
above procedures. 
  
Figure 5. Skeleton image for the lower body 
4.2 Evaluation of Road Structure 
Current regulation is 2cm for difference in level at crossing and 
less than 5% slope for sidewalk. Then, these two regulations 
were investigated as follows. 
Table 2 shows the results of gait analysis for the tiptoe 
displacement. 
Table 2. Displacement of tiptoe(flat course) 
  
  
  
Average Maximum 
Aged 32cm 6.8 cm 
Young 9.5 cm 12.1 cm 
  
  
  
  
  
This table shows that the average and maximum displacement 
for the aged and the young are 3.2, 6.8 cm and 9.5, 12.1 cm 
respectively. It can be seen from the results that the current 
regulation is satisfied. However, regarding the rate for the tiptop 
belongs from 1.5 cm to 2.5 cm while swing phase, the aged 
showed 48.7%, and the young was 21.1%. This indicates that 
the aged have high risk of tumbling or stumbling even current 
regulation is satisfied. 
Table 3. Displacement of tiptoe(5% slop) 
  
  
  
Average Maximum 
Aged 4.2 cm 7.8 cm 
Young 4.5 cm 8.0 cm 
  
  
  
  
  
Table 3 shows the results of gait analysis for the tiptoe 
displacement using 5% slope. It can be said from the table 3 
that 5% slopes is appropriate value, but the same risk as the 
difference in level can be presumed. 
4.2.1 Gait analysis for the aged and the young 
As other gait elements, stride and acceleration for the aged and 
the young were investigated. 
Table 4 shows the average stride. Obviously the aged stride was 
shorter than the young. However, it can be found interesting 
results that the aged average for flat and slope are the same 
values in spite of the young was changed. The reason doesn’t 
sure, but the young have ability to control by themselves. 
—272- 
re: 
an 
th: 
Gi 
fr 
ar 
Fu 
su