 |
Claims  |
|
|
What is claimed is:
1. A method of generating a binary halftone image based on image data
representing an original image, comprising the steps of:
measuring spatial frequencies in said original image by:
(a) obtaining density data indicative of a density of each pixel in said
original image from said image data of said original image, and
(b) applying a second differential filter to said density data to thereby
generate filtered density data indicative of a spatial frequency in the
proximity of each pixel in said original image; and
generating said binary halftone image by producing high resolution dots in
a first image region and low resolution dots in a second image region,
said first image region having spatial frequencies higher than a threshold
frequency, said second image region having spatial frequencies lower than
said threshold frequency, said generating step being performed by:
(a) comparing said filtered density data with said threshold frequency to
thereby generate a selection signal for each pixel, and
(b) selecting one of said high resolution dots and said low resolution dots
to each pixel in response to said selection signal.
2. A method in accordance with claim 1, wherein said step (f) comprises the
step of:
(g) dividing said original image into said first image region and said
second image region based on a distribution of said selection signal.
3. A method in accordance with claim 2, wherein said step (g) comprises the
steps of:
dividing said original image into a plurality of unit areas of an identical
size; and
obtaining a logical sum of said selection signal for a plurality of pixels
in each of said plurality of unit areas, to thereby identifying each of
said plurality of unit area as one of said first and second image regions.
4. A method in accordance with claim 2, wherein said step (g) comprises the
steps of:
displaying a binary image representing said distribution of said selection
signal in said original image on a display device; and
dividing said original image into said first image region and said second
image region through interactive operation while observing said binary
image displayed on said display device.
5. A method in accordance with claim 2, wherein said step (g) comprises the
steps of:
displaying a multi-valued image representing a distribution of said
filtered density data on a display device; and
dividing said original image into said first image region and said second
image region through interactive operation while observing said
multi-valued image displayed on said display device.
6. A method of generating a binary halftone image based on image data
representing an original image, comprising the steps of:
measuring spatial frequencies in said original image by:
(a) providing a first threshold pattern for said high resolution dots and a
second threshold pattern for said low resolution dots,
(b) obtaining density data indicative of a density of each pixel in said
original image from said image data of said original image,
(c) applying a second differential filter to said density data to thereby
generate filtered density data indicative of a spatial frequency in the
proximity of each pixel in said original image, and
(d) determining a first coefficient for said high resolution dots and a
second coefficient for said low resolution dots from said filtered density
data of each pixel; and
generating said binary halftone image by producing high resolution dots in
a first image region and low resolution dots in a second image region,
said first image region having spatial frequencies higher than a threshold
frequency, said second image region having spatial frequencies lower than
said threshold frequency, said generating step being performed by:
(a) reading out a first threshold value from said first threshold pattern
and a second threshold value from said second threshold pattern with
respect to each pixel in said original image,
(b) multiplying said first threshold value by said first coefficient and
said second threshold value by said second coefficient, and generating a
third threshold value by adding the results of multiplication,
(c) comparing said image data with said third threshold with respect to
each pixel in said original image, to thereby generate a dot signal
representing said binary halftone image, and
(d) generating said binary halftone image from said dot signal.
7. An apparatus for generating a binary halftone image based on image data
representing an original image, comprising:
frequency measuring means for measuring spatial frequencies in said
original image, said frequency measuring means including:
(a) means for obtaining density data indicative of a density of each pixel
in said original image from said image data of said original image, and
(b) means for applying a second differential filter to said density data to
thereby generate filtered density data indicative of a spatial frequency
in the proximity of each pixel in said original image, and
dot generation means for generating said binary halftone image by producing
high resolution dots in a first image region and low resolution dots in a
second image region, said first image region having spatial frequencies
higher than a threshold frequency, said second image region having spatial
frequencies lower than said threshold frequency said dot generation means
including:
(a) means for comparing said filtered density data with said threshold
frequency to thereby generate a selection signal for each pixel, and
(b) selection means for selecting one of said high resolution dots and said
low resolution dots to each pixel in response to said selection signal.
8. An apparatus in accordance with claim 7, wherein said selection means
comprises:
division means for dividing said original image into said first image
region and said second image region based on a distribution of said
selection signal.
9. An apparatus in accordance with claim 8, wherein said division means
comprises:
means for dividing said original image into a plurality of unit areas of an
identical size; and
means for obtaining a logical sum of said selection signal for a plurality
of pixels in each of said plurality of unit areas, to thereby identifying
each of said plurality of unit area as one of said first and second image
regions.
10. An apparatus in accordance with claim 8, further comprising:
a display device for displaying a binary image representing said
distribution of said selection signal in said original image; and wherein
said division means comprises:
means for specifying to divide said original image into said first image
region and said second image region through interactive operation while
observing said binary image displayed on said display device.
11. An apparatus in accordance with claim 8, further comprising:
a display device for displaying a binary image representing said
distribution of said selection signal in said original image; and wherein
said division means comprises:
means for specifying to divide said original image into said first image
region and said second image region through interactive operation while
observing said multi-valued image displayed on said display device.
12. An apparatus for generating a binary halftone image based on image data
representing an original image, comprising:
a first threshold memory for storing a first threshold pattern for said
high resolution dots; and
a second threshold memory for storing a second threshold pattern for said
low resolution dots;
frequency measuring means for measuring spatial frequencies in said
original image said frequency measuring means including:
(a) means for obtaining density data indicative of a density of each pixel
in said original image from said image data of said original image,
(b) means for applying a second differential filter to said density data to
thereby generate filtered density data indicative of a spatial frequency
in the proximity of each pixel in said original image, and
(c) means for determining a first coefficient for said high resolution dots
and a second coefficient for said low resolution dots from said filtered
density data of each pixel; and
dot generation means for generating said binary halftone image by producing
high resolution dots in a first image region and low resolution dots in a
second image region, said first image region having spatial frequencies
higher than a threshold frequency, said second image region having spatial
frequencies lower than said threshold frequency, said dot generation means
including:
(a) means for reading out a first threshold value from said first threshold
memory and a second threshold value from said second threshold memory with
respect to each pixel in said original image,
(b) means for multiplying said first threshold value by said first
coefficient and said second threshold value by said second coefficient,
and generating a third threshold value by adding the results of
multiplication,
(c) a comparator for comparing said image data with said third threshold
with respect to each pixel in said original image, to thereby generate a
dot signal representing said binary halftone image, and
(d) means for generating said binary halftone image from said dot signal.
13. A method of generating a binary halftone image based on image data
representing an original image, said method comprising the steps of:
(a) measuring spatial frequencies in said original image by applying a
differential filter to said image data; and
(b) generating said binary halftone image by producing high resolution dots
in a first image region and low resolution dots in a second image region,
said first image region having spatial frequencies higher than a threshold
frequency, said second image region having spatial frequencies lower than
said threshold frequency, said low resolution dots being conventional
halftone dots, said high resolution dots being frequency modulation dots
whose spacing varies with image density.
14. An apparatus for generating a binary halftone image based on image data
representing an original image, said apparatus comprising:
frequency measuring means for measuring spatial frequencies in said
original image by applying a differential filter to said image data; and
dot generation means for generating said binary halftone image by producing
high resolution dots in a first image region and low resolution dots in a
second image region, said first image region having spatial frequencies
higher than a threshold frequency, said second image region having spatial
frequencies lower than said threshold frequency, said low resolution dots
being conventional halftone dots, said high resolution dots being
frequency modulation dots whose spacing varies with image density. |
|
 |
|
 |
Claims |
|
|
|
Description |
|
|
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention pertains to a method of and an apparatus for
generating a binary halftone image based on image data representing an
original image.
2. Description of the Related Art
In color offset printing, an original image, such as a photograph image, is
divided into four color separation images corresponding to four color
inks, for example, yellow, magenta, cyan, and black. A printed image is
then produced by overprinting the color separation images of the
respective color inks one upon another.
An image on a printing plate for each color ink (hereinafter referred to as
"printing plate image" or "halftone image") is expressed by small dots of
the color ink, which are called "halftone dots". Halftone dots are
arranged in lattice at a regular interval, and a ratio of the dot area per
unit area represents the density of an image. A percentage of the dot area
per unit area is generally called a halftone-dot percent or a dot percent.
The interval between halftone dots is defined by a screen ruling, and the
orientation of a halftone-dot array is defined by a screen angle. The
screen ruling represents the number of halftone dots formed per inch. The
greater screen ruling produces a reproduced image with higher resolution.
Conventionally, the screen ruling is set at about 175 lines/inch.
Recent advancement in printing technology allows to utilize halftone dots
of a higher resolution as much as about 300 screen lines/inch or mere,
which are called high definition halftone dots. The greater screen ruling
makes each halftone dot smaller and allows an original image to be
reproduced with higher resolution.
Halftone dots have a fixed structure of an array, and their size in an
image area is varied according to the density of the image area. In other
words, halftone dots represent the density of an image area by means of
Amplitude Modulation. Recently, another method of representing the density
of an image area by Frequency Modulation has been put to practical use,
which is called FM screening or FM dot or stochastic screening. In the FM
screening, ink dots have a fixed size, but dot spacing varies according to
the density of the image area. Dots used in FM screening are much smaller
than conventional halftone dots, and reproduce an original image with high
resolution accordingly.
Unlike the conventional halftone dots, FM dots are not arranged at a
regular interval. In this specification, a term `printing dots` or simply
`dots` will be used to refer to both of the halftone dots and FM dots. The
above-mentioned high definition halftone dots and the FM dots will be
referred to as "high resolution dots" while halftone dots having a
relatively low screen ruling, e.g. 175 lines/inch, will be referred to as
"low resolution dots".
The high resolution dots reproduce details of an original image more
clearly and precisely than the low resolution halftone dots. The small
size of printing dots makes rosette moires sufficiently inconspicuous. The
small high-resolution dots, however, have a drawback; it is difficult to
reproduce the dot size precisely in printing of a halftone image. This
means that the low resolution dot is preferable to the high resolution dot
for the better reproducibility of the dot size. It is accordingly
preferred to apply the high resolution dots to some areas of an original
image and low resolution dots to the other areas.
SUMMARY OF THE INVENTION
An object of the present invention is accordingly to generate a binary
halftone image by taking full advantage of high resolution dots and low
resolution dots.
The present invention is directed to a method of generating a binary
halftone image based on image data representing an original image. The
method comprises the steps of: (a) measuring spatial frequencies in the
original image; and (b) generating the binary halftone image by producing
high resolution dots in a first image region and low resolution dots in a
second image region, the first image region having spatial frequencies
higher than a threshold frequency, the second image region having spatial
frequencies lower than the threshold frequency.
The assignment of the high resolution dots to the first image region of
high spatial frequencies will reproduce image details precisely and
clearly, while the assignment of the low resolution dots to the second
image region of low spatial frequencies will improve the reproducibility
of the dot size.
In a preferred embodiment of the present invention, the step (a) comprises
the steps of: (c) obtaining density data indicative of a density of each
pixel in the original image from the image data of the original image; and
(d) applying a second differential filter to the density data to thereby
generate filtered density data indicative of a spatial frequency in the
proximity of each pixel in the original image; and the step (b) comprises
the steps of: (e) comparing the filtered density data with the threshold
frequency to thereby generate a selection signal for each pixel; and (f)
selecting one of the high resolution dots and the low resolution dots to
each pixel in response to the selection signal.
Preferably, the step (f) comprises the step of: (g) dividing the original
image into the first image region and the second image region based on a
distribution of the selection signal.
In one embodiment of the step (g) comprises the steps of: dividing the
original image into a plurality of unit areas of an identical size; and
obtaining a logical sum of the selection signal for a plurality of pixels
in each of the plurality of unit areas, to thereby identifying each of the
plurality of unit area as one of the first and second image regions.
In another embodiment of the present invention, the step (g) comprises the
steps of: displaying a binary image representing the distribution of the
selection signal in the original image on a display device; and dividing
the original image into the first image region and the second image region
through interactive operation while observing the binary image displayed
on the display device.
In still another embodiment of the present invention, the step (g)
comprises the steps of: displaying a multi-valued image representing a
distribution of the filtered density data on a display device; and
dividing the original image into the first image region and the second
image region through interactive operation while observing the
multi-valued image displayed on the display device.
Preferably, the step (a) comprises the steps of: providing a first
threshold pattern for the high resolution dots and a second threshold
pattern for the low resolution dots; obtaining density data indicative of
a density of each pixel in the original image from the image data of the
original image; applying a second differential filter to the density data
to thereby generate filtered density data indicative of a spatial
frequency in the proximity of each pixel in the original image; and
determining a first coefficient for the high resolution dots and a second
coefficient for the low resolution dots from the filtered density data of
each pixel. The step (b) comprises the steps of: reading out a first
threshold value from the first threshold pattern and a second threshold
value from the second threshold pattern with respect to each pixel in the
original image; multiplying the first threshold value by the first
coefficient and the second threshold value by the second coefficient, and
generating a third threshold value by adding the results of
multiplication; comparing the image data with the third threshold with
respect to each pixel in the original image, to thereby generate a dot
signal representing the binary halftone image; and generating the binary
halftone image from the dot signal.
The high resolution dots are frequency modulation dots whose spacing varies
with image density.
The present invention is also directed to an apparatus for generating a
binary halftone image based on image data representing an original image.
The apparatus comprises: frequency measuring means for measuring spatial
frequencies in the original image; and dot generation means for generating
the binary halftone image by producing high resolution dots in a first
image region and low resolution dots in a second image region, the first
image region having spatial frequencies higher than a threshold frequency,
the second image region having spatial frequencies lower than the
threshold frequency.
These and other objects, features, aspects, and advantages of the present
invention will become more apparent from the following detailed
description of the preferred embodiments with the accompanying drawings.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a block diagram illustrating the structure of a system for
generating a binary halftone image according to an embodiment of the
present invention;
FIG. 2 is a block diagram showing an internal structure of a dot generator
32;
FIG. 3 is a flowchart showing a process routine of generating dot selection
signals;
FIGS. 4(A) through 4(C) show a process of expanding an image area;
FIG. 5 shows density data DS stored in a line buffer 21;
FIGS. 6(A) and 6(B) are plan views illustrating Laplacian filters;
FIGS. 7(A) through 7(G) show a density-based variation in the shape of a
square dot or a low resolution dot;
FIGS. 8(A) through 8(G) show a density-based variation in the shape of FM
dots or high resolution dots;
FIGS. 9(A) through 9(E) show a process of separating image areas based on
the distribution of a dot selection signal SEL;
FIG. 10 shows a concept on which the user specifies an image area;
FIG. 11 is a block diagram showing structure of a threshold combining
circuit;
FIGS. 12(A) and 12(B) are graphs showing input-to-output characteristics of
two look-up tables 72 and 74; and
FIG. 13 is a block diagram showing another structure of the dot generator
32.
DESCRIPTION OF THE PREFERRED EMBODIMENT
A. System Structure
FIG. 1 is a block diagram illustrating the structure of a prepress process
system for generating a binary halftone image according to an embodiment
of the present invention. The prepress process system includes: a CPU 10
functioning as an operation means for analyzing spatial frequencies in an
original image; an image controller 12 functioning as a control means; an
input/output interface 14; a disk controller 16; an Screen Pattern Memory
(SPM) controller 18; a memory 20 as a memory means; a line buffer 21 as a
two-dimensional expanding means; a network interface 22; a magnetic disk
30 controlled by the disk controller 16, working as a storage means; a dot
generator 32 controlled by the SPM controller 18, functioning as a dot
generating means; a scanner input interface 34; and a scanner output
interface 36. The elements 10, 12, 14, 16, 18, 20, 21, and 22 are
connected to one another via a 10 bus 11.
The prepress process system further includes a reading scanner 100 as an
image input means connected with the scanner input interface 34; and a
recording scanner 200 as an image output means connected with the scanner
output interface 36.
Image data D.sub.IM, representing a photograph image or other original
image, are captured by the reading scanner 100 and transmitted to the
image controller 12 via the scanner input interface 34. The image
controller 12 stores the image data D.sub.IM into the magnetic disk 30 by
means of Direct Memory Access (DMA) transfer. The network interface 22 is
used to receive image data D.sub.IM from various devices on the network,
such as an image data base and an image processing apparatus.
The image controller 12 synchronizes the elements of the system while a
binary halftone image is recorded by the recording scanner 200. In
recording a binary halftone image, the image controller 12 reads out the
image data D.sub.IM from the magnetic disk 30 and transfers the image data
D.sub.IM to the dot generator 32. The dot generator 32 subsequently
converts the image data D.sub.IM to a dot signal Sd. The internal
structure and operations of the dot generator 32 will be described later
in detail. The dot signal Sd is transferred via the scanner output
interface 36 to the recording scanner 200, which records a binary halftone
image responsive to the dot signal Sd.
FIG. 2 is a block diagram showing the internal structure of the dot
generator 32. The dot generator 32 includes: a primary-scanning clock
generator (Y clock generator) 40; a secondary-scanning clock generator (X
clock generator) 42; a primary-scanning address counter (Y address
counter) 44; a secondary-scanning address counter (X address counter) 46;
first and second screen pattern memories (SPM) 48 and 50; a selector 52
working as a selection means; and a comparator 54 working as a comparator
means. The two SPMs 48 and 50 store threshold data used for converting the
image data D.sub.IM to the dot signal Sd. The first SPM 48 stores
threshold data DTH1 used in generating low resolution dots while the
second SPM 50 stores threshold data DTH2 used in generating high
resolution dots. Differences between the low resolution dots and the high
resolution dots will be described later.
The Y clock generator 40 and the X clock generator 42 generate a primary
scanning clock YCLK and a secondary scanning clock XCLK respectively from
a reference clock CLK given from a rotary encoder 210 (FIG. 1) of the
recording scanner 200. The Y address counter 44 is a ring-counter for
counting the number of pulses of the primary scanning clock YCLK to
generates a primary scanning address YADD for the SPMs 48 and 50. The X
address counter 46 is a ring-counter for counting the number of pulses of
the secondary scanning clock XCLK to generate a secondary scanning address
XADD for the SPMs 48 and 50.
The primary scanning address YADD and the secondary scanning address XADD
are commonly given to the SPMs 48 and 50. The SPMs 48 and 50 respectively
output the threshold data DTH1 and DTH2 in response to the addresses. The
selector 52 selects one of the two threshold data DTH1 and DTH2 in
response to a dot selection signal SEL given from the image controller 12.
The image controller 12 generates the dot selection signal SEL according
to a process described later.
The comparator 54 compares the image data D.sub.IM transferred from the
image controller 12 with the threshold data DTH (DTH1 or DTH2) output from
the selector 52, and determines the level of the dot signal Sd based on
the result of comparison. The recording scanner 200 receives the dot
signal Sd and records a binary halftone image responsive to the dot signal
Sd onto a recording medium such as a photosensitive film.
B. Process of Generating Dot Selection Signal
In the process of recording a binary halftone image, the dot selection
signal SEL is generated to determine whether the high resolution dots or
the low resolution dots are to be produced. FIG. 3 is a flowchart showing
a process routine of generating the dot selection signal. This process
routine is executed by a filtering processor 10a and a selection signal
generator 10b shown in FIG. 1. The CPU 10 executes software programs
stored in the memory 20 to implements the functions of the units 10a and
10b.
At step S1 in the flowchart of FIG. 3, the filtering processor 10a expands
an image area of an original image by a width of one pixel. FIGS. 4(A)
through 4(C) shows the process of expanding an image area. The filtering
process is executed by applying a Laplacian filter LF of a predetermined
size, for example, a 3.times.3 pixel matrix, to each pixel PX in the
original image as shown in FIG. 4(C). In order to execute the filtering
process on the outer-most pixels in the original image, which are filled
with slant lines in FIG. 4(C), extra pixels are necessary outside the
outer-most pixels in the original image area. Therefore, at step S1, image
data for the outer-most pixels are copied to an area of a pixel width
outside the outer-most pixels, to thereby expand the image area. For
example, when the image area of the original image has a size of mxn
pixels as shown in FIG. 4(A), the expanded image area consists of
(m+2).times.(n+2) pixels as shown in FIG. 4(B). As illustrated in FIG.
4(C), pixel coordinates (X,Y) are (0,0) at an origin (upper-left point) of
the original image area, and (-1,-1) at the upper-left point of the
expanded image area. Image data of the expanded image area are stored in
the magnetic disk 30.
At step S2, the filtering processor 10a reads out image data for one
scanning line from the magnetic disk 30 and generates density data DS from
the image data. For example, when the image data has a yellow component Y,
a magenta component M, a cyan component C, and a block component K, the
density data DS is given by:
DS=a1*Y+a2*M+a3*C+a4*K (1)
where a1 through a4 are coefficients.
The density data DS is a total of the components Y, M, C, and K of the
image data weighed by the coefficients a1-a4, respectively. The weighting
coefficients a1-a4 can be set in various ways. In one preferable example,
the coefficient a2 for the magenta component is set equal to one whereas
the other coefficients are equal to zero. This is because the magenta
component, in many cases, distinguishably represents the high-frequency
component of the density variation in the image.
In general, at least one of the four weighting coefficients a1-a4 are to be
set at a non-zero value. The density data DS may not precisely represent
the total density of the original image but it may express some kind of
density of each pixel in the original image.
At step S3 in the flowchart of FIG. 3, the density data DS for one scanning
line generated at step S2 are written into the line buffer 21. FIG. 5
shows the content stored in the line buffer 21. The line buffer 21 is a
memory for storing the density data DS for a predetermined number of
scanning lines, for example, for three scanning lines, which is equal to
the side width of the Laplacian filter LF used in the filtering process.
At a first execution cycle of steps S2 and S3 in the process routine of
FIG. 3, the density data DS for three scanning lines are generated and
written in the line buffer 21.
In FIG. 5, the density data DS for each pixel is referred to as DS(X,Y),
where X and Y respectively denote a secondary-scanning coordinate and a
primary-scanning coordinate of the pixel. In the state of FIG. 5, the line
buffer 21 stores the density data DS for the three scanning lines whose
secondary-scanning coordinates X are -1, 0, and 1, respectively. The
scanning line at X=-1 is supplemented on the left side of the image area
of the original image by the expansion process shown in FIG. 4(C). The
pixels at Y=-1 and Y=(n+1) are supplemented above and below the image area
of the original image by the expansion process.
At step S4, the filtering processor 10a executes the filtering process on
the density data DS stored in the line buffer 21. In the filtering
process, the Laplacian filter LF of 3.times.3 matrix is applied to each
3.times.3 density data to obtain a filtered value for a target pixel at
the center, which exists on the central line of the three scanning lines.
FIGS. 6(A) and 6(B) illustrate examples of Laplacian filters LF1 and LF2
adoptable in the filtering process.
For example, the Laplacian filter LF1 shown in FIG. 6(A) is applied to the
density data to execute a product-sum operation, and an absolute value of
the sum is determined as filtered density data FDS(X,Y):
FDS(X,Y)=.vertline.DS(X,Y-1)+DS(X-1,Y)-4*DS(X,Y)+DS(X+1,Y)+DS(X,Y+1).vertli
ne. (2)
The filtered density data FDS indicates the degree of density variation in
the vicinity of each target pixel. Various differential operators of a
second order other than the Laplacian filters shown in FIGS. 6(A) and 6(B)
can be used as spatial filters. The size of the filter can be greater than
the 3.times.3 matrix.
At step S5, the selection signal generator 10b compares the filtered
density data FDS with threshold data TH to generate the dot selection
signal SEL as follows:
If TH<FDS: SEL=1 (selecting high resolution dots); and
If TH.gtoreq.FDS: SEL=0 (selecting low resolution dots).
When the filtered density data FDS is greater than the threshold data TH,
the spatial frequency is high in the vicinity of the target pixel. The dot
selection signal SEL is accordingly set equal to one, which means
selection of the high resolution dots. When the filtered density data FDS
is equal to or less than the threshold data TH, on the contrary, the
spatial frequency is low in the vicinity of the target pixel. The dot
selection signal SEL is accordingly set equal to zero, which means
selection of the low resolution dots. Based on the results of comparison
between the filtered density data FDS and the threshold data TH, the high
resolution dots are assigned to the areas having a relatively steep
density change whereas the low resolution dots are allocated to the other
areas having a relatively gentle density change.
The value of the threshold data TH is determined empirically by examining
the results of applying the Laplacian filter to actual original images.
For example, when the original image data D.sub.IM and the density data DS
and FDS are 8-bit digital data, a preferable value for the threshold data
TH is about 140. The threshold data TH also depends on the coefficients of
the Laplacian filter LF.
After the dot selection signal SEL is generated for one scanning line at
step S5, the program goes to step S6 at which it is determined whether the
process is completed for all the scanning lines. When not completed, the
program returns to step S2 to generate the density data DS for a next
scanning line, and the density data DS is written in the line buffer 21 at
step S3. At step S3, the density data DS for the oldest one of the three
scanning lines in the line buffer 21 are replaced by the new density data
DS. For example, the density data DS(-1,0) through DS(-1,n+1) for the
oldest scanning line shown in FIG. 5 are replaced by the new density data
DS(2,0) through DS(2,n+1) for a new scanning line. In this case, the new
density data for a new scanning line are considered to exist on the
right-side of the line buffer 21 in the filtering process at step S4.
Steps S2 through S6 are repeated to update the density data DS by one
scanning line, to store the update density data DS in the line buffer 21,
to execute the filtering process with respect to the density data DS
stored in the line buffer 21, and to generate the dot selection signal SEL
based on the filtered density data FDS. Although the dot selection signal
SEL is stored in the magnetic disk 30 in the embodiment, the dot selection
signal SEL may not be stored in the magnetic disk 30 but generated at real
time when the recording scanner 200 records a binary halftone image.
C. Recording of Halftone Image
In recording a binary halftone image by the recording scanner 200, the
image data D.sub.IM and the dot selection signal SEL stored in the
magnetic disk 30 are read out simultaneously and supplied to the dot
generator 32 while their timings are adjusted by the image controller 12.
As illustrated in FIG. 2, the dot selection signal SEL is given to the
selector 52, and the image data D.sub.IM is transmitted to the comparator
54. The selector 52 selects one of the two threshold data DTH1 and DTH2
output from the two SPMs 48 and 50 for each pixel in response to the dot
selection signal SEL, and transfers the selected data to the comparator
54. When the dot selection signal SEL is equal to zero, the selector 52
selects the first threshold data DTH1 for the low resolution dots. When
the dot selection signal SEL is equal to one, on the contrary, the
selector 52 selects the second threshold data DTH2 for the high resolution
dots. The comparator 54 compares the selected threshold data DTH with the
image data D.sub.IM to generate the binary dot signal Sd.
FIGS. 7(A) through 7(G) show a sequence of shape change of a square dot or
a low resolution dot according to density change, and FIGS. 8(A) through
8(G) show the sequence of FM dots or high resolution dots. Each of FIGS.
7(A) through 7(G) shows one halftone dot area having a size of 23 spots by
23 spots. The spot denotes a record pixel, which is a unit of recording in
the recording scanner 200. Each of FIGS. 8(A) through 8(G) shows high
resolution dots in the area of 23 spots by 23 spots. In the square dot
area shown in FIGS. 7(A) through 7(G), the size of the black dot area
increases with the density. In the FM dots shown in FIGS. 8(A) through
8(G), on the other hand, the positional frequency of black dots increases
with the density. One pixel of image data D.sub.IM has, for example, the
size of 10 spots by 10 spots.
The first SPM 48 (FIG. 2) stores a threshold pattern of the threshold data
DTH1 representing the density-dependent change of the square dot shown in
FIGS. 7(A) through 7(G). The threshold data DTH1 is read out according to
the addresses YADD and XADD. The second SPM 50 stores a threshold pattern
of the threshold data DTH2 representing the density-dependent change of
the FM dots shown in FIGS. 8(A) through 8(G). The threshold data DTH2 is
also read out according to the addresses YADD and XADD. The threshold
pattern of the square dot or that of the FM dots are repeatedly applied
without any gaps on an image plane of a recorded image. In order to
implement the repeated application of the threshold pattern, the address
counters 44 and 46 execute the ring counting at predetermined periods. On
the assumption that the first SPM 48 and the second SPM 50 store threshold
patterns in the area of 23.times.23 spots, the address counters 44 and 46
execute the ring counting to produce the addresses YADD and XADD in the
range of 0 through 22, respectively. In actual application, both the SPM
48 and 50 store threshold patterns greater than 23.times.23 spots in size.
Each FM dot is small in size and arranged at random, thus making rosette
moires inconspicuous and preventing object moires which are generally
caused by interference of a picture pattern with an array of dots.
Further, the small FM dots have an advantage of high reproducibility in
details of the image. In reproduction of a color print by overprinting a
plurality of color separation images, different threshold patterns are
used for the respective color separations to produce their halftone
images. This effectively prevents color shift, or deviation of colors in a
resulting print from proper colors, due to registering mismatch of plates
in printing process.
There is a type of FM dots whose spacing is changed according to the
density while the spot size is fixed, and another type whose spot size
depends on the density as well as its spacing. The present invention can
apply to both types of FM dots. In this specification, "FM dots" are
defined to be the dots whose spacing is at least varied with the density.
The resolution power of the FM dots is greater than about 600 dpi (dot per
inch), and more preferably greater than about 800 dpi. When the FM dots
are recorded by a high-resolution recording scanner whose resolution is as
high as 4,000 dpi, each FM dot may be set equal to an area of 2.times.2
spots or 3.times.3 spots of the recording scanner. This prevents the size
of each dot from being excessively small and improves the dot
reproducibility in printing.
Other types of high resolution dots other than the FM dots can be also used
in the present invention. High resolution dots are halftone dots having a
screen ruling as high as about 300 screen lines per inch or more while a
low resolution dot has a screen ruling of about 200 lines per inch or a
less. The high resolution dot is small in size, which effectively prevents
rosette moires and improves the reproducibility of image details.
The low resolution dot is inferior to the high resolution dot in the
reproducibility of image details but has better reproducibility of the dot
size in printing process than the high resolution dot, especially in the
image areas having relatively uniform density.
As described above, the Laplacian filter is applied to the density data DS
to obtain the filtered density data FDS representing the spatial
frequencies in the image. The high resolution dots are applied to the
pixels having the higher spatial frequencies than a threshold value
whereas the low resolution dots are applied to the pixels having the lower
spatial frequencies than the threshold value. In the image areas having
relatively high spatial frequencies, the image details are clearly
reproduced by the high resolution dots. In image areas having relatively
low spatial frequencies, on the other hand, the low resolution dots
improve the reproducibility of the dot size in printing process.
D. Image Region Separation
In the above embodiment, the dot selection signal SEL is generated for each
pixel in order to select one of the high resolution dot and the low
resolution dot. In another possible embodiment described below, an
original image is divided into first-type image regions to which the low
resolution dots are applied and second-type image regions to which the
high resolution dots are applied, based on the distribution of the dot
selection signals SEL.
FIGS. 9(A) through 9(E) show a process of the region separation based on
the distr | | |
