Facsimile and static presentation processing – Static presentation processing – Attribute control
Reexamination Certificate
1999-11-17
2004-02-17
Rogers, Scott (Department: 2626)
Facsimile and static presentation processing
Static presentation processing
Attribute control
C358S003040, C358S003050, C382S252000, C382S304000
Reexamination Certificate
active
06693727
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention relates to an image processing apparatus for processing an image input by error diffusion or the like.
In general, a text image processing apparatus capable of processing both bit information and image information simply binarizes, in accordance with a fixed threshold, contrast image information such as a character and line drawing on an original read by a reader unit such a scanner. This apparatus binarizes continuous tone image information such as a photograph by a pseudo-halftoning section according to dithering or the like, and outputs the binary information to, e.g., a binary printer having a small number of gray levels.
When read image information is simply binarized in accordance with a fixed threshold, the resolution of a character/line image region is preserved and the image quality does not suffer. However, the resolution of a photographic image region is not preserved, resulting in a low-quality image. When read image information is gradated by ordered dithering or the like, the resolution of a photographic image region is preserved and the image quality does not suffer. However, the resolution of a character/line image region decreases, resulting in a low-quality image. That is, only one binarization processing for read image information cannot simultaneously satisfy the image qualities of regions having different features.
As a binarization/multivalued scheme which satisfies the tone of a photographic image region and attains a higher resolution even for a character/line image region than ordered dithering, an “error diffusion method” is proposed. According to the “error diffusion method” (reference: An Adaptive Algorithm for Spatial Grayscale, by R. W. Floyd and L. Steinberg, Proceedings of the S.I.D. Vol. 17-2, pp. 75-77, Second Quarter 1976), a value obtained by multiplying a binarizing error of a binarized surrounding pixel by a given weighting coefficient is added to the density of a target pixel, and the target pixel is binarized in accordance with a fixed threshold.
FIG. 3
is a block diagram showing general binarization processing using the “error diffusion method”. In
FIG. 3
, an input image signal
2
-
1
is input to a section
2
-
2
for correcting an error from a preceding line (preceding line error) that corrects image information of a target pixel in accordance with a correction amount
2
-
17
diffused from the preceding line. An image signal
2
-
3
corrected in accordance with the preceding line error is input to a section
2
-
4
for correcting a pixel diffused from a preceding pixel. A corrected image signal
2
-
5
is input to a binarizing section (thresholding process)
2
-
6
for binarizing the corrected image information of the target pixel while comparing it with a threshold
2
-
7
. A binarization image signal
2
-
8
is input to a binarizing error calculating section
2
-
9
for calculating a binarizing error of the target pixel on the basis of an output level
2
-
10
and the corrected image signal
2
-
5
depending on the binarization value. The binarizing error calculating section
2
-
9
outputs a binarizing error signal
2
-
11
to an error filtering section
2
-
12
for distributing the binarizing error calculated by the binarizing error calculating section
2
-
9
to surrounding pixels. The error filtering section
2
-
12
is made up of an error calculating section
2
-
13
for calculating an error to an adjacent pixel (adjacent pixel error), and error calculating section
2
-
14
for calculating an error to a next line (next line error).
An adjacent pixel error correction amount
2
-
15
is output to the section
2
-
4
, and an error correction amount
2
-
16
to a next line is stored in an error buffer
2
-
18
.
FIG. 4
is a circuit diagram showing an arrangement of a conventional binarizing circuit using the “error diffusion method”. The input image signal
2
-
1
read by an input device such as a scanner is corrected by the section
2
-
2
in accordance with an image correcting signal
2
-
17
, and output as the error-corrected image signal
2
-
3
. The error-corrected image signal
2
-
3
corrected in accordance with the preceding line error correction amount is corrected by the section
2
-
4
in accordance with the preceding pixel error correcting signal
2
-
15
, and is output as the corrected image signal
2
-
5
.
Upon receiving the corrected image signal
2
-
5
, the binarizing section (comparator)
2
-
6
uses the corrected image signal
2
-
5
and a binarizing threshold Th (for example, “80h”; the suffix “h” means “hex” representing hexadecimal). The binarizing section
2
-
6
outputs “1” (black pixel) as the binarization image signal
2
-
8
if the corrected image signal
2
-
5
is larger than the binarizing threshold Th; otherwise it outputs “0” (white pixel).
The binarizing error calculating section
2
-
9
selects one output level
2
-
10
based on the binarization image signal
2
-
8
(i.e., selects an output level for 0 when the binarization image signal is “0”, and an output level for 1 when the signal is “1”). The binarizing error calculating section
2
-
9
calculates the difference between this output level and the corrected image signal
2
-
5
, and outputs the difference as the binarizing error signal
2
-
11
.
The error filtering section
2
-
12
has an error filter arrangement shown in FIG.
5
. In
FIG. 5
, “&Circlesolid;” indicates a target pixel position. The error filtering section
2
-
12
calculates weighting errors obtained by multiplying the binarizing error signal
2
-
11
by weighting coefficients kA, kB, kC, and kD (where kA={fraction (7/16)}, kB={fraction (1/16)}, kC={fraction (5/16)}, and kD={fraction (3/16)}). In other words, the error filtering section
2
-
12
multiplies the binarizing error of the target pixel by the weighting coefficients kA, kB, kC, and kD to calculate weighting errors to four pixels (pixels corresponding to the positions of the weighting coefficients kA, kB, kC, and kD) surrounding the target pixel.
The error correction amount
2
-
15
to the adjacent pixel in the main scanning direction is calculated by the error calculating section
2
-
13
in accordance with the coefficient kA and binarizing error signal
2
-
11
. The error correction amount
2
-
15
is supplied from a delay element
2
-
19
to the correcting section
2
-
4
in response to a next processing clock. Similarly, the next line error correction amount
2
-
16
is calculated by the error calculating sections
2
-
14
, and delay elements
2
-
20
,
2
-
21
, and
2
-
22
in accordance with the binarizing error signal
2
-
11
and filter coefficients kB, kC, and kD. The error buffer
2
-
18
temporarily stores the error correction amount
2
-
16
in addition to the sum of weighting errors eB, eC, and eD for three pixels from the preceding line. In this scheme, the following condition must be satisfied to diffuse all errors:
kA+kB+kC+kD=1
Recently even in error diffusion processing for an output device having a large number of gray levels (level number), the above-described binarizing section is replaced with a multivalued section using a number of thresholds corresponding to the number of gray levels.
According to the “error diffusion processing”, an error generated by binarization/multivalued processing for a target pixel is diffused to surrounding pixels to correct the error, thereby minimizing the binarizing/multivalued error. One problem of this method is that texture (regular pattern) appears on an output image (particularly upon binarization processing).
Another method using a mask and AND instead of coefficients is also proposed. This is a high-speed method of reducing texture by a synergistic effect with variation noise of a scanner.
FIG. 6
shows a bit mask error diffusion method in detail. The bit mask error diffusion arrangement is the same as the conventional error diffusion arrangement shown in
FIG. 4
except for the error filtering scheme.
Masks mA, mB, mC, and mD shown in
FIG. 7
replace the c
Kanno Hiroki
Rao Gururaj
LandOfFree
Image processing apparatus for performing high-speed error... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Image processing apparatus for performing high-speed error..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Image processing apparatus for performing high-speed error... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3279041