CHAPTER 4 IMAGE PROCESSING SYSTEM
4-16
COPYRIGHT
©
1999 CANON INC. CANON GP605/605V REV.0 JAN. 1999 PRINTED IN JAPAN (IMPRIME AU JAPON)
2. Random Error Diffusion (R-ED) Method
In the existing error diffusion (ED) method, a fixed pattern (error diffusion matrix) is used to
diffuse (distribute) errors occurring as a result of binary processing over surrounding pixels, possi-
bly generating specific patterns on the output images. To reduce such patterns, the copier adds
random values to the image signals representing pixels (hence, random error diffusion).
Figure 4-305B Conceptual Diagram of the Error Diffusion Method
Pixel in question (e.g., density at 135)
Error Diffusion Matrix (fixed pattern)
×0.2
×0.1
×0.05
×0.03
×0.05
×0.05
×0.01
×0.01
×0.2
×0.1
×0.1
×0.1
Black(255)
Slice level
(128)
White(0)
Error
Pixel in question
(density at 135)
Step 2
Step 1
+
α
-
α
+ β
- β
At this time, a value that brings the sum to
0 is added to the adjacent pixel to prevent
an error caused by the random number.
Each error is diffused (distributed)
to the surrounding pixels according to the
ratio of the error diffusion matrix.
A random number is added to the image signal
representing each pixel.
When the slice level used for binary
processing is 128, this pixel will be a black
pixel (255).
In this case, the diffusion will be 125 (255 -
135).
To Next Page
Loading...
Need help?
General