National Instruments IMAQTM 用户手册

Two equally important factors define the success of a pattern matching 
algorithm: accuracy and speed. You can define a search area to reduce 
ambiguity in the search process. For example, if the image has multiple 
instances of a pattern and only one of them is required for the inspection 
task, the presence of additional instances of the pattern can produce 
incorrect results. To avoid this, reduce the search area so that only the 
appropriate pattern lies within the search area.

The time required to locate a pattern in an image depends on both the 
template size and the search area. By reducing the search area or increasing 
the template size, you can reduce the required search time.

In many inspection applications, you have general information about the 
location of the fiducial. Use this information to define a search area. For 
example, in a typical component placement application, each printed 
circuit board (PCB) being tested may not be placed in the same location 
with the same orientation. The location of the PCB in various images can 
move and rotate within a known range of values, as illustrated in 
Figure 5-11. Figure 5-11a shows the template used to locate the PCB in the 
image. Figure 5-11b shows an image containing a PCB with a fiducial you 
want to locate. Notice the search area around the fiducial. If you know, 
before the matching process begins, that the PCB can shift or rotate in the 
image within a fixed range, as shown in Figure 5-11c and Figure 5-11d, 
respectively, you can limit the search for the fiducial to a small region of 
the image.