2 
Lateral single pixel sensitivity is investigated 
with monochromatic light irradiation (for example: 
436 nm; 559 nm; 626 nm) by imaging a diffraction- 
limited light spot (diameter: 1.1 um; 1.3 mj 1.5 
um) on to a pixel of the CCD to be investigated. 
During the measuring procedure the light spot 
step-by-step is moved over the chosen pixel and 
its environment in column and row direction. The 
characteristic curve of lateral sensitivity is 
obtained by recording the videosignal at every 
approached position. In case of an extremely high 
Spot positioning accuracy (£520 nm) it is possible 
to determine the real pixel geometry from the 
lateral sensitivity characteristic curve [1]. 
It can be stated that the expected trapezoidal 
shape of the lateral sensitivity characteristic 
curve will often not appear. 
In the following I would like to demonstrate a: few 
examples on the figures 2 ... 9. 
It can be established that no significant diffe- 
rences could be found in the shape of the lateral 
sensitivity characteristics of the measured pixels 
within one CCD component under constant conditions 
of measurement. 
    
  
     
   
   
    
   
  
    
    
  
    
    
   
     
  
   
    
  
3. MEASUREMENT OF LATERAL PIXEL (SUBPIXEL) SENSITIVITY - PIXEL GEOMETRY 
But obvious differences in the lateral sensitivity 
curve shape appear under constant measuring condi- 
tions on different CCD components 
- components of one and the same type but of 
different producers (WF, F79/F89) 
- components of one and the same producer but 
different batches (F79, F89) 
- different component types (WF, F79, F89, 
matrix) 
and also in the scope of one CCD when sampling 
direction or wavelength were changed (WF, F79, 
F99, matrix). 
Analogous measurement was carried out on sensors 
of different operating modes (for example: CID 
arrays), too. 
The presentation of the special results caused by 
the operating principle would go beyond the scope 
of this discussion. 
4. SPECIAL INVESTIGATIONS ON FRAME GRABBERS 
For tasks in measuring image processing it is of 
essential importance how imaging of light-sensiti- 
ve pixels is realized on frame grabbers Such a 
measurement is realized by moving the spot image 
mentioned in item 3 over the whole matrix along 
the rows or columns and by interpreting the CCD 
video signal and the occupation of the frame 
grabber memory simultaneously (see figures 19; 11) 
On the figures (12; 11) you can see that not all 
light-sensitive pixels of the matrix were imaged 
on to the frame grabber memory. 
Additionally a displacement of the geometric 
centre of the matrix relatively to the memory 
centre occurs on such an image. 
In imaging pixels on to the frame grabber memory 
the operating mode (pixel-synchronous, non-pixel- 
synchronous) is of great importance. 
From the non-pixel-synchronous operating mode it 
follows that a single pixel irradiated in accor- 
dance with the method described in item 3 stati- 
stically can occupy several adjacent memory loca- 
tions of a row in the frame grabber memory. 
Furtnermore in the non-pixel-synchronous mode the 
Space between two matrix pixels can be imaged. 
differently within a row of the frame grabber 
memory. 
That means, in the extreme case a circle imaged 
optically on to the matrix can appear on the frame 
grabber as an ellipse. Pixel-synchronous mode 
excludes effects of such kind and thus ensures an 
image pick-up true to the geometry. This is an 
unavoidable condition for measuring image proces- 
sing [2]. 
During the investigations of the pixel synchronism 
the following essential feature of frame grabbers 
becomes obviously. 
The analog channel located in the input unit of 
the frame grabber causes a type-specific smearing 
of spot or edge images within a row of a frame 
grabber memory over several pixels (figure 12, 
figure 13).