impact 
e from 
lassage 
Ire are 
sary to 
ber of 
ng the 
amber. 
ments 
remote 
"ration 
, Dept. 
Îted in 
we are 
-grade 
. This, 
ces the 
system 
ration 
eatcup 
  
2. Hardware of the system. 
When designing the system, the hardware was given priority 
because of the special requirements of the system. 
To make it possible to measure the teatcup liner inside the 
milking machine without disturbing the process, an adapter with 
a window to the teatcup had to be designed. The vacuum 
process should remain intact while measurements are taken. 
Limited space and limited budget were other constraints in the 
system consideration. 
The hardware solution was a mono-camera system with a 
mirror arrangement in front of the camera to create stereoscopic 
split images. In general, one-camera designs have some obvious 
advantages in price, calibration, etc. (El-Hakim 1990), and in 
this set-up factors like size, data quantity and synchronization 
also play decisive roles. 
The aim was also to use a "bottom to top" strategy in the 
design and start with uncomplicated solutions. 
The selected camera has a remote miniature head sized 17x55 
mm with resolution better than 360 (horizontal x 420 
(vertical). Since the simultaneous images are recorded on the 
same frame, no synchronization or pairing problems occur. 
Figure 2 shows two cross-sections of the teatcup and the 
camera adapter. 
  
The glass plate separate the mirrors and camera from the 
teatcup where varying vacuum occurs. Illumination is by 
external light sources through a diffusive window to avoid 
disturbing reflections. The prototype adapter is made of brass, 
but, the next version will be of a lighter composite material to 
keep down the additional weight of the adapter. 
3. Functions of the system. 
To get the required results of the system, solutions from 
different parts of the subject area had to be adapted and 
adjusted to the tasks. The accuracy requirements of the results 
were the most important factors for the choices of methods of 
measurement, data handling, estimation, etc. The system 
overview is shown by data flow diagram (DFD) (Gane and 
Sarson, 1979) in figure 3, and the more detailed decompositions 
are shown in the following subsections. 
In this article the processes of calibration and measurement are 
emphasized, both because they are the two most decisive parts 
of any system and because of the special solutions in our 
system. 
  
| | mirrors 
  
  
  
  
  
  
  
  
  
  
—>w 
| 
B 
er eer [s 
  
  
  
  
  
  
  
  
  
  
  
  
    
  
teatcup vacuum 
liner chamber 
ne 
Section B-B 
  
   
| 
| camera 
| housing 
| E ; 
i BI mirrors 
glass plate 
teatcup liner 
A 1 1 L 1 — CM 
1:22 (39 £4 5 10 
Section A-A 
  
  
  
Figure 2. Teatcup with adapter.