Sigma LBA-708 사용자 설명서

 

Convolution algorithms in the LBA-PC may take on a number of forms, some of which might not fit the 

exact description that is to follow.  In the broadest sense, convolution refers to a general-purpose 
algorithm that can be used in performing a variety of area process transformations.  One such general-

purpose algorithm will be described here. 
For the purpose of this description, the best way to understand a convolution is to think of it is a 

weighted summation process.  Each pixel in an image becomes the center element in a neighborhood of 

pixels.  A similarly dimensioned convolution kernel multiplies each pixel in the neighborhood.  The 

sum of these products is then used to replace the center pixel. 
Each element of the convolution kernel is a weighting factor called a convolution coefficient.  The 

size and arrangement of the convolution coefficients in a convolution kernel determine the type of area 

transform that will be applied to the image data.  
The figure below shows a 3x3 neighborhood and convolution kernel. 

 

 

Figure 58 

 

The tables below give the convolution coefficients (K values) for some of the included low-pass spatial 
filters. 
  

Operator’s Manual 

 

LBA-PC 

 

146