 |
Claims  |
|
|
I claim:
1. An error diffusing method in an image reproducing process wherein an
image is reproduced by approximating a tone of each of a plurality of
pixels constituting the image by either a maximum tone or a minimum tone,
said method comprising:
1) sequentially processing the pixels in a predetermined order to allocate
to each of the pixels one of a first code indicative of a tone value of
the maximum tone and a second code indicative of a tone value of the
minimum tone which is determined depending on an actual tone of the one
pixel;
2) calculating an error as a difference between the actual tone of the one
pixel and the tone value indicated by the code allocated to the one pixel
and setting the error as a negative error when the first code is allocated
to the one pixel and as a positive error when the second code is allocated
to the one pixel;
3) diffusing the calculated error to each of a plurality of the following
adjacent pixels, which are located in predetermined positional
relationships with the one pixel and are to be processed after the one
pixel has been processed, thereby correcting the tone value of each of the
following adjacent pixels, so that when the calculated error is the
positive error, the tone value of each of the following adjacent pixels is
corrected by adding thereto a predetermined proportion of the positive
error, and when the calculated error is the negative error, an absolute
value of the predetermined proportion of the negative error is compared
with an absolute value of the tone value of each of the following adjacent
pixels and when the former is larger than the latter, the tone value of
each of the following adjacent pixels is replaced by the predetermined
proportion of the negative error and when the former is no larger than the
latter, the tone value of each of the following adjacent pixels is
maintained; and
4) correcting the tone value of the one pixel prior to applying the
processing step to the one pixel, when there exists at least one preceding
adjacent pixel which should be processed before the one pixel is processed
and whose error should be diffused to the one pixel, by the error diffused
from the one preceding adjacent pixel.
2. A method according to claim 1, wherein said plurality of pixels are
arranged along a plurality of lines, which are arranged sequentially from
an uppermost one to a lowermost one and each of which extends from left to
right, and wherein said pixels are processed in an order from the pixels
of the uppermost line to the pixels of the lowermost line and from a
leftmost pixel to a rightmost pixel in each line and the error of the one
pixel is diffused to the pixels located at a right side, a lower side and
a right-lower side of said one pixel by proportions of 40%, 40% and 20%,
respectively.
3. A method according to claim 1, wherein said plurality of pixels are
arranged along a plurality of lines, which are arranged sequentially from
an uppermost one to a lowermost one and each of which extends from left to
right, and wherein said pixels are processed in an order from the pixels
of the uppermost line to the pixels of the lowermost line and from a
leftmost pixel to a rightmost pixel in each line and the error of each of
the pixels is diffused to the pixels located at a right side, a lower
side, a right-lower side and a left-lower side of said one pixel by
proportions of 30%, 30%, 20% and 20%, respectively.
4. A method according to claim 1, wherein said processing step includes
storing the tone values of said plurality of pixels in a plurality of
memory areas of a memory allotted to said pixels, respectively, and
reading out the tone values of the respective pixels from the memory areas
sequentially in the same order as that in which the pixels are processed
in the processing step and allotting to a pixel corresponding to each of
the read-out tone values one of the first and second codes depending on a
magnitude of the read-out tone value; and
wherein said correcting step includes replacing the tone value stored in
the memory area allotted to each of said following adjacent pixels by the
corrected tone value of the following adjacent pixel corrected by the
error diffusion.
5. A method according to claim 1, wherein said processing step includes
storing of the tone values of said plurality of pixels in a plurality of
memory areas of a first memory allotted to said pixels, respectively, and
reading out the tone values of the respective pixels from the memory areas
sequentially in the same order as that in which the pixels are processed
in the processing step, storing the tone value of each of the read-out
pixels into a second memory and allotting to each of the read-out pixels
one of the first and second codes depending on a magnitude of the tone
value of that pixel;
said diffusing step includes temporarily storing the calculated error of
each of the read-out pixels in one of the memory areas of a third memory
allotted to that pixel and correcting, when the tone value of one of the
following adjacent pixels, to which the corrected error of the read-out
pixel is to be diffused, is read out from the first memory and stored in
the second memory, the tone value of said one of the following adjacent
pixels based on said calculated error and replacing the tone value of said
one of the following adjacent pixels by the corrected tone value thereof;
and wherein the allotting of one of the first and second codes to said one
of the following adjacent pixels by said reading out step is carried out
based on the corrected tone value thereof.
6. An image processing apparatus comprising:
1) means for processing a plurality of pixels constituting an image in a
predetermined order to allocate to each of the plurality of pixels one of
a first code indicative of a tone value of a maximum tone and a second
code indicative of a tone value of a minimum tone which is determined
depending on a tone value of the one pixel;
2) means for calculating an error as a difference between the tone value of
the one pixel and the tone value indicated by the code allocated to the
one pixel and setting the error as a negative error when the first code is
allocated to the one pixel and as a positive error when the second code is
allocated to the one pixel;
3) means for diffusing the calculated error to each of a plurality of the
following adjacent pixels, which are located in predetermined positional
relationships with the one pixel and are to be processed after the one
pixel has been processed, thereby correcting the tone value of each of the
following adjacent pixels, such that when the calculated error is the
positive error, the tone value of each of the following adjacent pixels is
corrected by adding thereto a predetermined proportion of the positive
error, and when the calculated error is the negative error, an absolute
value of the predetermined proportion of the negative error is compared
with an absolute value of the tone value of each of the following adjacent
pixels and when the former is larger than the latter, the tone value of
each of the following adjacent pixels is replaced by the predetermined
proportion of the negative error and when the former is no larger than the
latter, the tone value of each of the following adjacent pixels is
maintained; and
4) means for correcting the tone value of the one pixel prior to processing
the one pixel by the processing means, when there exists at least one
preceding adjacent pixel which is processed before the one pixel is
processed and whose error is to be diffused to the one pixel, by the error
diffused from the one preceding adjacent pixel by the diffusing means.
7. An apparatus according to claim 6, wherein said plurality of pixels are
arranged along a plurality of lines, which are arranged sequentially from
an uppermost one to a lowermost one and each of which extends from left to
rights, and wherein said plurality of pixels are processed by said
processing means in an order from the pixels of the uppermost line to the
pixels of the lowermost line and from a leftmost pixel to a rightmost
pixel in each line and the error of the one pixel is diffused by said
diffusing means to the pixels located at a right side, a lower side and a
right-lower side of said one pixel by proportions of 40%, 40% and 20%,
respectively.
8. An apparatus according to claim 6, wherein said plurality of pixels are
arranged along a plurality of lines, which are arranged sequentially from
an uppermost one to a lowermost one and each of which extends from left to
right, and wherein said plurality of pixels are processed by said
processing means in an order from the pixels of the uppermost line to the
pixels of the lowermost line and from a leftmost pixel to a rightmost
pixel in each line and the error of the one pixel is diffused by said
diffusing means to the pixels located at a right side, a lower side, a
right-lower side and a left-lower side of said one pixel by proportions of
30%, 30%, 20% and 20%, respectively.
9. An apparatus according to claim 6, wherein said processing means
includes a memory having a plurality of memory areas allotted to said
plurality of pixels for storing the tone values of said plurality of
pixels, respectively; and means for reading out the tone values of the
respective pixels from the memory areas sequentially in the same order as
that in which said plurality of pixels are processed by said processing
means and allotting to a pixel corresponding to each of the read-out tone
values one of the first and second codes depending on a magnitude of the
read-out tone value; and
said correcting means includes means for replacing the tone value stored in
the memory are allotted to each of said following adjacent pixels by the
corrected tone value of the following adjacent pixel corrected by said
diffusing means.
10. An apparatus according to claim 6, wherein said processing means
includes a first memory having a plurality of memory areas allotted to
said plurality of pixels for storing the tone values of said plurality of
pixels, respectively; a second memory; means for reading out the tone
values of the respective pixels from the memory areas sequentially in the
same order as that in which said plurality of pixels are processed in said
processing means and storing the tone value of each of the read-out pixels
in said second memory and means for allotting to each of the read-out
pixels one of the first and second codes depending on a magnitude of the
tone value of that pixel;
said diffusing means includes a third memory for temporarily storing the
calculated error of each of the read-out pixels and means for correcting,
when the tone value of one of the following adjacent pixels, to which the
corrected error of the read-out pixel is to be diffused, is read out from
the first memory and stored in the second memory, the tone value of said
one of the following adjacent pixels based on said calculated error and
replacing the tone value of the one of the following adjacent pixels by
the corrected tone value thereof; and
wherein the allotting of tone of the first and second codes to said one of
the following adjacent pixel by said allotting means is carried out based
on the corrected tone value thereof. |
|
 |
|
 |
Claims |
|
|
|
Description |
|
|
BACKGROUND OF THE INVENTION
1. Field of the Invention
The invention relates to an error diffusing method in an image reproducing
process and an image processing apparatus using such a method. More
particularly, it relates to error diffusing in an image reproducing
process with continuous-tone image data and an image processing apparatus
using such a method.
2. Description of the Related Art
An error distributing algorithm of a conventional error diffusing method
will be first described. For instance, as disclosed in J. C. Stoffel, "A
Survey of Electronic Techniques for Pictorial Image Reproduction", IEEE
Transactions on Communications, Vol. COM-29, No. 12, December, 1981,
according to the error diffusing method, for example, in case of a pixel
arrangement as shown in FIG. 6, the pixels are sequentially processed in
an order of the pixels
A.fwdarw.B.fwdarw.C.fwdarw.D.fwdarw.E.fwdarw.F.fwdarw.. In the process of
each pixel, when the tone of the pixel is binary-coded, a threshold value
S which is set to a value between the displayable maximum grade M and
minimum grade L is compared with tone data DATA of each pixel. When the
tone data DATA is not smaller than the threshold value S, a dot is
generated. When the tone data DATA is smaller than the threshold value S,
no dot is generated.
By executing the above processes, when the dot is generated, the pixel
presents the maximum grade M. When no dot is generated, on the contrary,
the pixel presents the minimum grade L. Therefore, when the tone data DATA
is in a range of L<DATA<S, the pixel which actually has an intermediate
tone is printed at the minimum grade L (ordinarily, zero grade: therefore,
the pixel is printed in the background color of a recording paper).
In this case, a tone of pixel as printed has an error of (DATA-L) as
compared with the actual pixel tone. Such an error is distributed to the
adjacent pixels which will be processed later than the processed pixel so
that the image is displayed as a whole at a tone grade near the actual
image. In the case where the tone of pixel as printed is lower than the
actual pixel tone, it is assumed that the error is a positive error, and
hence the above error (DATA-L) is the positive error. The positive error
of a pixel, for example, pixel D is distributed to adjacent pixels E, G,
and H. In the case where one of the adjacent pixels, for instance, the
pixel E has tone data of DATA', the distributed error is added to the tone
data DATA'. There is a possibility that the errors are distributed to the
pixel E from not only the pixel D but also the other adjacent pixels A and
B. When the tone data corrected by the accumulation value of all of the
distributed errors is larger than the threshold value S, a dot is
generated when the pixel E is printed. Similar processes are also executed
with respect to the other pixels.
On the other hand, when the tone data DATA of the pixel D is in a range of
S<DATA<M, a dot is generated when H is printed. Therefore, the pixel D is
printed at a tone which is higher than the actual tone by (M-DATA). Since
(M-DATA) is a negative error, when the error is distributed to the
adjacent pixels, the distributed negative error is subtracted from the
actual tone data of each of the adjacent pixels. As mentioned above, each
pixel is printed based on the sum of the actual tone data of the pixel
itself and the positive or negative errors distributed from the adjacent
pixels. The errors are thus cancelled with each other and hence the total
error is minimized when the whole image is viewed.
The threshold value S of the error diffusing method is generally set to a
value near 50% (128 in case of displaying an image with 256 grades) in
order to improve the accuracy in position of dots corresponding to the
pixels having tones of intermediate to higher grades. Therefore, the
negative error occurs as mentioned above. However, the characteristic of
the positive error is different from that of the negative error, that is,
they are not symmetrical. Therefore, when the positive and negative error
are processed in the same manner, the displayed picture presents visually
and actually undesired appearance although an average tone of the original
picture is maintained. This is because the tone which is desired to be
reproduced is not an average tone of the whole image, but a tone of a
narrow area including the pixel of interest.
For instance, in case of printing two long vertical lines a and b as shown
in FIG. 5A, the negative errors occurring due to the generation of the
left line a are progressively accumulated in the direction of the
progressing pixels, namely, in the right downward direction. Due to the
accumulated errors, the diffused negative errors are added to the tone
data of the pixel at a position where the right line b should be generated
so that the corrected tone value of the pixel may become smaller than the
threshold value S. In such a case, as shown in FIG. 5B, no dot is
generated at the position where the dot should be generated and the line
is broken.
On the other hand, for example, a case of generating vertical lines at the
columns n=1 and n=4, as shown in FIG. 6 will now be considered. It is now
assumed that the tone of each vertical line is 200, the maximum grade M is
255 and the minimum grade L is 0. As for the error diffusion amount due to
the generation of the line at the column n=1 including the pixels A, D, G,
and J, the negative errors increase progressively in the right downward
direction as shown in FIG. 7. That is, in FIG. 7, a numerical value in [ ]
of the first line of each pixel indicates a tone of that pixel. The value
of the second line indicates an amount of error which is distributed to
that pixel from the upper pixel. The value of the third line indicates an
amount of error which is distributed from the left-upper pixel. The value
of the fourth line indicates an amount of error which is distributed from
the left side pixel. The value in ( ) of the lowest line indicates a total
value of the errors. In the example, for simplicity of explanation, it is
assumed that the error is distributed to each of the right and lower
pixels by 40% and to the right-lower pixel by 20%.
As will be obviously understood from FIG. 7, since the negative errors
produced by the generation of the line a at the column n=1 are
progressively accumulated in the right direction and in the right-lower
direction, the accumulated negative error increases according to the pixel
position which approaches the right lower position. Therefore, with
respect to the line b at the column n=4, the negative errors distributed
from the pixels of the line a to each of the pixels of the line b are
larger than the tone value of each pixel of the line b and the total
negative errors of lower pixels of the line b become larger than the total
negative errors of upper pixels of the line b, so that the possibility of
generation of the dot becomes smaller for the lower pixels than for the
upper pixels of the line b. Moreover, since the negative error cannot be
cancelled by generation of a dot unlike the positive error, the negative
error becomes progressively larger, when dots are vertically generated
along a line, in the right-down direction from the line. Thus, no dot is
generated and a line is broken at a position where a dot should be
generated as mentioned-above.
SUMMARY OF THE INVENTION
An object of the present invention is to provide an error diffusing method,
which is used for an image reproducing process in which an image is
reproduced by approximating a tone of each of pixels constituting the
image by one of the maximum tone and the minimum tone and wherein an error
between the actual tone of each pixel and the maximum or minimum tone by
which the actual tone of the pixel is approximated is distributed to a
plurality of adjacent pixels located adjacent to the pixel so as to
correct the tone data of each of the adjacent pixels. The method
eliminates an adverse effect on the reproduced image caused by
accumulation of the negative errors, which may be produced when the actual
tone of each pixel is approximated by the maximum tone higher than the
actual tone and the error is distributed to the plurality of the adjacent
pixels.
Another object of the invention is to provide an image processing apparatus
using the above error diffusing method.
In order to achieve the above object, according to the present invention,
an error diffusing method, which is used in an image reproducing process
wherein an image is reproduced by approximating a tone of each of the
pixels constituting the image by either the maximum tone or the minimum
tone, comprises sequentially processing the pixels in a predetermined
order to allocate to each pixel a first code indicative of a tone value of
the maximum tone or a second code indicative of a tone value of the
minimum tone which is determined depending on an actual tone of the one
pixel; calculating an error as a difference between the actual tone of the
one pixel and the tone value indicated by the code allocated to the one
pixel and setting the error as a negative error when the first code is
allocated to the one pixel and as a positive error when the second code is
allocated to the one pixel; diffusing the calculated error to each of a
plurality of following adjacent pixels, which are located in predetermined
positional relationships with the one pixel and are to be processed after
the one pixel has been processed, thereby correcting the tone value of
each of the following adjacent pixels, such that when the calculated error
is the positive error, the tone value of each of the following adjacent
pixels is corrected by adding thereto a predetermined proportion of the
positive error, and when the calculated error is the negative error, an
absolute value of the predetermined proportion of the negative error is
compared with an absolute value of the tone value of each of the following
adjacent pixels and when the former is larger than the latter, the tone
value of each following adjacent pixel is replaced by the predetermined
proportion of the negative error and when the former is no larger than the
latter, the tone value of each of the following adjacent pixels is
maintained; and correcting the tone value of the one pixel prior to
applying the first step to the one pixel, when there exists at least one
preceding adjacent pixel which should be processed before the one pixel is
processed and whose error should be diffused to the one pixel, by the
error diffused from the one preceding adjacent pixel.
An image processing apparatus according to the invention comprises: means
for sequentially processing pixels constituting an image in a
predetermined order to allocate to each of the pixels either a first code
indicative of a maximum tone or a second code indicative of a minimum tone
which is determined depending on an actual tone of the one pixel; means
for calculating an error as a difference between the actual tone of the
one pixel and the tone value indicated by the code allocated to the one
pixel and setting the error as a negative error when the first code is
allocated to the one pixel and as a positive error when the second code is
allocated to the one pixel. Furthermore, the apparatus includes means for
diffusing the calculated error to a plurality of following adjacent
pixels, which are located in predetermined positional relatonships with
the one pixel and are to be processed after the one pixel has been
processed, thereby correcting the tone value of each of the following
adjacent pixels. It is done so that when the calculated error is the
positive error, the tone value of each of the following adjacent pixel is
corrected by a predetermined proportion of the positive error, and when
the calculated error is the negative error, an absolute value of the
predetermined proportion of the negative error is compared with an
absolute value of tone value of the each of the following adjacent pixels
and when the former is larger than the latter, the tone value of each of
the following adjacent pixels is replaced by the predetermined proportion
of the negative error and when the former is no larger than the latter,
the tone value of each of the following adjacent pixels is maintained.
Also means is provided for correcting the tone value of the one pixel
prior to allocating one of the first and second codes to the one pixel by
the processing means, when there exists at least one preceding adjacent
pixel which should be processed before the one pixel is processed and
whose error should be diffused to the one pixel, by the error diffused
from the preceding adjacent pixel by the diffusing means.
The error diffusing method in the image reproducing process of the
invention includes the above steps. When the pixel to which the first code
indicative of the maximum tone is allocated is printed, a dot is
generated. When the pixel to which the second code indicative of the
minimum tone is allocated is printed, no dot is generated. The positive
error which occurs when no dot is generated is diffused to a plurality of
the following adjacent pixels and cancelled by generating a dot at each of
the following adjacent pixels. However, when the negative error, which is
produced by generation of a dot, is diffused to a plurality of the
following adjacent pixels, there is a possibility that another negative
error is again produced in each of the following adjacent pixels to which
the original negative error is diffused. Therefore, the negative error is
further diffused to the further following adjacent pixels and the error is
gradually accumulated. According to the error diffusing method of the
invention, in case of the negative error, the absolute value of a
predetermined proportion of the negative error is compared with the
absolute value of the tone value of each of the following adjacent pixels.
On the basis of the result of the comparison, the tone value of each of
the following adjacent pixels is held unchanged or replaced by the
predetermined proportion of the negative error, so that the negative
errors are not infinitely accumulated.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a diagram showing a circuit construction of a main section of an
image processing apparatus of the first embodiment of the invention;
FIG. 2 is a flowchart showing an error diffusing process by the image
processing apparatus in FIG. 1;
FIG. 3 is a diagram showing a circuit construction of a main section of an
image processing apparatus of the second embodiment of the invention;
FIG. 4 is a flowchart showing an error diffusing process by the image
processing apparatus in FIG. 3;
FIGS. 5A and 5B are diagrams respectively showing an actual image of two
lines and an image of the lines as printed after processed by the
conventional error diffusing method;
FIG. 6 is a diagram showing diffusing directions of an error of each pixel;
and
FIG. 7 is a diagram showing amounts of errors which are diffused when
processed by the conventional error diffusing method.
DESCRIPTION OF THE PREFERRED EMBODIMENTS
Embodiments of the invention will be described with reference to the
attached drawings.
FIG. 1 is a circuit diagram of an image processing apparatus is illustrated
showing an embodiment of the invention, whereas FIG. 2 is a flowchart for
explaining the process carried out by the apparatus of FIG. 1.
In FIG. 1, reference numeral 1 denotes an image memory; 2 is a binary
coding circuit; 3 is an output buffer; 4 is an operating circuit; 5 is an
error buffer memory; 6 a weighting circuit; and 7 is an adding circuit.
The image memory 1 stores data and the binary coding circuit 2 determines
whether a dot is to be generated or not for each pixel on the basis of the
tone data DATA of the pixel which is supplied from the image memory 1.
That is, in the binary coding circuit 2, the threshold value S is provided
between the maximum tone M and the minimum tone L which can be reproduced
by the image processing apparatus. The threshold value S is compared with
the tone data DATA of each pixel as supplied. When the tone data DATA is
no smaller than the threshold value S, a dot is generated. When DATA is
smaller than S, no dot is generated.
The output buffer 3 temporarily stores an output of the binary coding
circuit 2. The operating circuit 4 calculates an error produced as a
result of the binary coding of the tone data. Further, the error buffer
memory 5 temporarily stores the generated error. The weighting circuit 6
is provided to weight the tone data in accordance with the position of the
pixel to which the error is distributed. The adding circuit 7 adds the
weighted error to the data as will be explained hereinlater.
The operation of the embodiment will now be described on the basis of the
above construction.
The respective tone data of the pixels are sequentially read out from the
image memory 1 and processed in an order of the pixels
A.fwdarw.B.fwdarw.C.fwdarw.D.fwdarw.E.fwdarw.F.fwdarw. in the pixel
arrangement shown in FIG. 6. The tone data DATA of each pixel supplied
from the image memory 1 is inputted to the binary coding circuit 2. In the
binary coding circuit 2, the threshold value S (127) is set at a value
between the maximum tone M (255) and the minimum tone L (0). The threshold
value S is compared with the input data DATA and when the input data DATA
is no smaller than the threshold value S, the binary coding circuit 2
generates a code "1". When the input data DATA is smaller than the
threshold value S, a code "0 is generated. The above process is executed
in step 502 in the flowchart of FIG. 2.
The binary code obtained by the binary coding circuit 2 as mentioned above
is sent to the output buffer 3 in order to reproduce an image and is also
sent to the operating circuit 4.
When the output of the binary coding circuit 2 is the code "1", a value
which is obtained by subtracting 255 from the tone data DATA is generated
as a negative error from the operating circuit 4. When the output of the
binary coding circuit 2 is the code "0", a value of the tone data DATA is
generated as a positive error (step 503). The above error is temporarily
stored into the error buffer memory 5 and weighted by the weighting
circuit 6 in accordance with the positional relationship of each of the
pixels to which the error is distributed. For example, in this embodiment,
the error is distributed to the pixels located at the right side, lower
side and right-lower side of the processed pixel by 40%, 40% and 20%,
respectively, of the error. Therefore, the error is weighted by 40% for
the right-side pixel, 40% for the lower-side pixel and 20% for the
right-lower side pixel (step 504).
The adding circuit 7 reads out the tone data stored in an address of the
image memory 1 allotted to each of the error receiving pixels to which the
error is to be distributed, corrects the read out data by adding to it the
weighted error for that error receiving pixel and again storing the
corrected tone data into the original address. Thus, the tone data of each
pixel is replaced by a tone data corrected by an error distributed from
each of the pixels which have been processed before the one pixel is
processed. A method of adding the weighted error to the read-out data is
executed in accordance with a predetermined rule as will be explained
hereinafter.
The adding circuit 7 also judges the sign of the error (step 505). When a
positive error occurs in the operating circuit 4, the data in the address
of each of the pixels to which the error is to be diffused is simply read
out and the error is added thereto for correcting the data. Then, the
corrected data is again written into the same address (step 506).
In the case where a negative error occurs in the operating circuit 4,
however, the data in the address of each of the error receiving pixels is
read out and the sign of the read-out data is discriminated (step 507).
The read-out data has a value obtained by adding the errors distributed
from the adjacent pixels which had already been processed to the original
tone value of the pixel itself.
When the discrimination in step 507 indicates that the sign of the read-out
data is positive, the error multiplied by the weight coefficient is added
to the read-out data and the resultant new data is again written into the
same address (step 506).
On the other hand, when the sign of the read-out data is negative, an
absolute value of the read-out data is compared with an absolute value of
the weighted error to be distributed (step 508). When the absolute value
of the read-out data is not smaller than the absolute value of the
weighted error to be distributed, the processing routine is finished while
maintaining the read-out data unchanged, that is without adding the error.
When the absolute value of the weighted error to be distributed is larger
than the absolute value of the read-out data, however, the read-out data
is cancelled and the weighted error to be distributed is written into the
address from which the data was read out (step 509). The read-out tone
data of the pixel may be negative when the tone data of the pixel has been
corrected by the negative error distributed from an adjacent pixel which
has been processed before the read-out pixel.
By executing the above processes, when the corrected tone data of the error
receiving pixel in the image memory 1 is negative, the tone data is
replaced by the distributed error. When a large negative error newly
occurs, therefore, the errors accumulated so far are cancelled and the
preceding errors are not diffused any more. In other words, the expansion
of a distribution range of the negative error is stopped where a large
negative error occurs. Therefore, it is possible to eliminate the
conventional drawback such that when two vertical lines are reproduced,
the lower portion of the right line disappears because the negative error
is infinitely diffused.
The second embodiment of the invention will now be described with reference
to FIGS. 3 and 4.
FIG. 3 is a constructional diagram of an image processing apparatus of the
second embodiment of the invention, whereas FIG. 4 is a flowchart for
explaining an error diffusing portion of the apparatus of FIG. 3.
The image processing apparatus comprises: the image memory 1 which stores
the image data; the binary coding circuit 2 which converts the tone data
DATA of each pixel which is supplied from the image memory 1 into one of
two binary codes; the output buffer 3 which temporarily stores the output
of the binary coding circuit 2; an error calculating circuit 10 which
calculates the error between the tone data and the tone value represented
by the converted binary code; an error diffusion control circuit 11 which
determines whether the error is diffused or not; an error buffer memory
12; a weighting circuit 13; and an adding circuit 14. In the embodiment of
FIG. 3 as well, with respect to the image having a pixel arrangement as
shown in FIG. 6, the pixels are sequentially processed in an order of the
pixels A.fwdarw.B.fwdarw.C.fwdarw.D.fwdarw.E.fwdarw.F.fwdarw.. Unlike in
the first embodiment, however, the tone data of each pixel stored in the
image memory 1 is not corrected during the execution of the process.
First, the tone data DATA of each pixel supplied from the image memory 1 is
corrected by adding thereto the diffusion errors from adjacent pixels by
the adding circuit 14. After that, the corrected tone data is sent to the
binary coding circuit 2. In the binary coding circuit 2, the threshold
value S (127) is set at a value between the maximum tone M (255) and the
minimum tone L (0). The threshold value S is compared with the corrected
tone data DATA. When the corrected tone data DATA is no smaller than the
threshold value S, the binary coding circuit 2 generates a "1" signal.
When the corrected tone data DATA is smaller than the threshold value S, a
"0" signal is generated. The binary code generated from the binary coding
circuit 2 is sent to the output buffer 3 and also to the error calculating
circuit 10.
When the output of the binary coding circuit 2 is set to "1", the error
calculating circuit 10 outputs a value obtained by subtracting 255 from
the tone data DATA as a negative error. When the output of the binary
coding circuit 2 is set to "0", the tone data DATA is outputted as a
positive error.
The error G(X) of the pixel generally represented by X obtained as
mentioned above is temporarily stored in the error buffer memory 12. When
the pixel of the error receiving side is processed later, the error G(X)
is weighted by the weighting circuit 13 with a coefficient determined
depending on the position of the error receiving pixel relative to the
pixel X. For instance, in the embodiment, a weight of 30% is allotted to
the pixel located at the right side of the processed pixel. Similarly, a
weight of 30% is allotted to the lower pixel, a weight of 20% to the
right-lower pixel and a weight of 20% to the left-lower pixel. The error
G(X) calculated by the error calculating circuit 10 is stored into the
error buffer memory 12 and held at least until completion of the process
of the error receiving pixels to which the error G(X) is distributed. The
operation of each of the above circuits is controlled by the error
diffusion control circuit 11. In the diagram, a solid line indicates a
flow of data and a broken line indicates a flow of control signals.
An error diffusing method of the second embodiment will now be described
with reference to FIG. 4.
In the case where the tone data DATA of, for example, the pixel E in FIG. 6
is read out from the image memory 1, the errors {G(A), G(B), G(C), G(D)}
generated by the pixels A, B, C and D, for which the pixel E is designated
as an error receiving side, are read out from the error buffer memory 12
under the control of the error diffusion control circuit 11. A weighting
process is executed to the read-out errors by the weighting circuit 13.
The weighted errors are added to or replaced for the tone data of the
pixel E by the adding circuit 14 in accordance with a predetermined rule,
which will be explained hereinlater. An output signal of the adding
circuit 14 is sent to the binary coding circuit 2.
The adding circuit 14 discriminates the sign of each of the errors G(A),
G(B), G(C) and G(D), generally represented by error G(i) and the processes
are executed in an order of (1) to (4) as follows:
(1) The error diffusion control circuit 11 first fetches the tone data DATA
of the pixel E in FIG. 6 as read out from the image memory 1 into a buffer
memory 14B in the adding circuit 14.
(2) Subsequently, the sign of the error G(A) read out from the address of
the pixel A in the error buffer memory 12 is discriminated (step 703).
(3) When the sign of the error G(A) is positive, the error read out from
the address of the pixel A in the error buffer memory 12 is added, after
being multiplied by a weight coefficient, to the data in the memory 14B in
the adding circuit 14. The original data in the memory 14B is replaced by
the added value. (step 704)
(4) When the sign of the error G(A) is negative, the data stored in the
memory 14B (at this timing, it is the original tone data DATA of the pixel
E in FIG. 6) is supplied to the binary coding circuit 2. The binary coding
circuit 2 converts the tone data into one of two binary codes, applied to
the error calculating circuit 10. The binary code is not stored into the
output buffer 3. Upon completion of the process regarding the pixel D, the
binary code as generated is stored into the output buffer 3. The error
calculating circuit 10 calculates the error G (M14) of the tone data read
out from the memory 14B, that is the difference between the tone data and
a value represented by the binary code.
The error diffusion control circuit 11 discriminates the sign of the error
G (M14) (step 707). When the sign of the error G (M14) is positive, the
error G(A) of the pixel A in the error buffer memory 12 is multiplied by a
weight coefficient by the weighting circuit 13 and, after that, the
weighted error is added to the data stored in the memory 14B in the adding
circuit 14 and the original data in the memory 14B is replaced by the
added value.
When the sign of the error G (M14) is negative, the absolute value of the
error G (M14) is compared with the absolute value of the error G(A)
multiplied by the weight coefficient by the weighting circuit 13 (step
708).
When the absolute value of the error G (M14) is larger than the absolute
value of the weighted error G(A), the error diffusing process is returned
to the step 703 for processing the error of the next pixel B, without
adding the weighted error G(A) from the weighting circuit 13 and hence,
without changing the data in the memory 14B (step 705).
When the absolute value of the weighted error G(A) supplied from the
weighting circuit 13 is larger than the absolute value of the error G
(M14), the data in the memory 14B is replaced by the weighted error G(A)
supplied from the weighting circuit 13.
The above steps are applied to the errors of the pixels A, B, C, and D,
sequentially (i=A, B, C, D), which errors are distributed to the pixel E.
Upon completion of the error diffusing process for the pixel D (step 711),
the data in the memory 14B is outputted to the binary coding circuit 2.
Subsequently, the binary output is supplied to the output buffer 3. The
output of the error calculating circuit 10 is written into the address of
the pixel E in the error buffer memory 12. The processing routine of the
pixel E is completed as mentioned above.
As described above, in the first embodiment shown in FIG. 1, the error of
each pixel is diffused to the adjacent pixels at a time when the binary
coding of the tone data of that pixel has been completed. In the second
embodiment, however, the error of each pixel is stored when the binary
coding of the tone data of the pixel has been completed and each pixel
receives the errors diffused from the adjacent pixels, when the tone data
of that pixel is to be converted to one of the binary codes.
Even in the second embodiment, therefore, when a large negative error newly
occurs, the errors which have been accumulated so far are cancelled and
the preceding errors are not diffused any more. In other words, the
expansion of a range in which the negative errors are diffused is stopped
where a large negative error occurs. Even | | |
