Note: Descriptions are shown in the official language in which they were submitted.
~22157~
~ETHOD OF REPRODUCING I~AGES
WIT~ GREY VALUES
Field Of_The Invention
This invention relates to a method of reproducing
images with grey values using dot pattern picture ~ ;
elements.
Prior Art
The invention concerns a method of reproducing
images witll grey values using picture elements (PELS)
for different grey values, which are represented by dot
patterns with a different distribution and arrangement
of the individual dots. For representing grey values,
many different methods have become known which have
proved more or less successful in simulating the
continuously changing grey tones of a natural image.
For this purpose, different quality criteria are
desirable, for example, a resolution which is such that
the grey, above all the light grey, tones are
satisfactorily reproduced, taking into account spatial
frequencies, i.e., the lateral dot resolution, and that
the uniformity and fidelity achieved are high.
A high resolution and uniformity are particularly
desirable in areas with a constant or slowly changing
grey tone. For the following reasons, photography is
ideal both for the human eye ~and for optical scanning:
The grain, i.e., the dots, in addition to being
stochastically distributed, have different sizes. The
known grain structure of photographs has the advantage
that the totally non-uniform size and the equally
non-uniform distribution of the grains prevents the
GE9-83-021
122157~
1 occurrence of a m~ire pattern if the image is to be
scanned or printed at a particular spatial resolution.
Such moire patterns result from raster and scanning
frequencies being superimposed upon each other and from
the natural frequencies of printers or image screens.
During the scanning of rastered halftone images, such
as in newspaper printing, stripes are formed.
There are serious disadvantages of all sys-tems
with constant dot addressing, such as those used in
lithography or newspaper printing, as well as of
simulation methods, such as supercircle, monocircule,
and the like.
The most efficient system used for such purposes
is MECCA (Multiple Error Correction Computation
Algorithm). It is based on the fact that a complex
algorithm uses the continuous siqnal Q (x, y), a
weighted mean value of previously computed quantization
errors, and a print-out reflecting the local image
texture, in order to ensure that a dot is printed and
displayed with a constant size. This is in contrast
with photography or newspaper printing and with the
variable dot size supercircle and monocircle methods
simulating same.
As will be explained in detail below, both these
methods have specific disadvantages. Simulation
systems, for example, with a fixed pel size necessitate
elaborate programming and storage means.
GE9-83-021 2
122~71
1 Summary of the Invention
Therefore, it is the object of the invention to
provide a method of the previously described kind which
supplies a continuous or quasi-continuous grey value
scale of almost white to black. In accordance with the
invention, this is accomplished in that for each grey
value there is a stock of quasi-equivalent grey value
pels, and that for representing each grey value of an
original to be reproduced, a pel corresponding to this
grey value is selected from the associated stock of
quasi-equivalent grey value patterns.
Further embodiments of the invention may be seen
from the sub-claims.
One way of carrying out the invention wilL be
described in detail below with reference to drawings
which illustrate only one specific embodiment, in
which:
Figs. la and lb show pels with only one of, for
example, 16 dots;
Figs. 2a, 2b and 2c show different clusters or dot
groups in pels;
Fig. 3 shows the grey stages for a pel with 16
dots;
Fig. 4 shows a pel with four adjacent dots as a
cluster;
Fig. 4A shows the relevant grey scale;
GE9-83-021 3
'~
r- _
t
~2ZiS~71
l Fig. 5 shows another grey scale and
Figs. 5a and 5b sho~ a pel ~lith 1/4 dot;
Figs. 6a - 6e show different pels with 4 dots eachi
Figs. 7a and 7b show pels which are quasi-equivalent with respect
to their grey value;
Fig. 8 is a diagram representing the dependence of the
grey value on the respective pattern, the diameter of
the clusters and the diameter of the dots, and
Fig. 9 is a block diagram explaining the entire method.
The various disadvantages of the methods known so far will he described
below.
According to Figs. la and lb, at a high resolution of, for example, 600
pels per inc~,`~ne such pel consists of a gridlike, preferably square,
structure of dots with a spacing d which for 16 individual dots may be of
the order of 42.8 ~m. Conceivable for this purpose are grids with 9 dots,
16 dots, 49 dots, 64 dots, etc. For simplicity's sake, it is assumed that
there is a gr~d with 16 dots. If all these dots are used to reproduce a
pel, a black image is obtained, and if none of the dots is used, the
resultant image is white. This shows quite clearly that 16 grey levels can
be obtained, ranging from 1 to 15 of 16 possible dots. The smallest--black
area obtainable has a blackness of 1/16 = 6.2 %. If, for obtaining the
lowest grey value, the individual dot is recurrently arranged at the same
point of the grid, then stripe or moiré patterns occur which are highly
undesirable.
If, according to Fig. 2, for example, 4, 6, 8, 10, 12 or 14 dots are used
from directly adjacent positions arranged one below the other, clusters are
formed which have a particular repetition frequency that will detrimentally
affect the reproduced image. The grey levels obtained in this case are
shown in Fig. 3~ '
. .
GE9 83-021 4
.... .
_ ._
lZ21~71
1 There are several possibilities of using these 16 grey levels. Assuming,
for example, according to Figs. 4 and 4a, that a pel comprises 4 dots, then
the smallest black area has 1/4 or about 25 6 blackness, i.e., at the
lowest grey value, the blackness is not less than 25 ~.
If, according to Figs. 5, 5a and ~b, it is assumed that each pel comprises
only 1/4 dot, then the smallest pel area is 1/64 of the blackness or 1.6 ~.
In this case, the maximum blackness is 25 ~. Thus, with 16 grey levels,
the very light grey tones are quite satisfactorily reproduced, ~ut the
blackness is only about 25 ~. This shows that both methods are sub-
stantially unsuitable.
Assuming that there are square dots and a grid spacing d equalling the dot
si~e with the edge length d, then a maximum of 16 grey levels is obtained
in a 16-dot array. Figs. 6a to 6e show that several dot arrays may yield
the same quasi-equivalent grey value. Four dots in an array may be
arranged, for example, adjacent to each other at the edges or in the
center. They may also be arranged diagonally or in rows or in cross form,
etc. All these ~rays yield practically the same grey level.
, ,~ .
~owever, irrespective of this, spurious frequencies of the printer, the
respective paper and the processing steps used will lead to greatly
differing pri,nt results in practice. If, for example, as a result of the
printer used, the fourth line always varies in the Y-direction, then only
the pattern of Fig. 6d will be insusceptible to this variation. Thus,
given the following paramete~s, a fixed size 16-dot array will yield a
maximum number of 16 non-redundant grey levels.
1. If the edge length of the dots is less than or equal to the grid
spacing, there will be 16 grey levels. If, on the other hand, the
edge length of the Z~ts exceeds the grid spacing, then fewer than 16
grey levels will be obtained.
2. The lightest grey level is the lighter, the smaller the edge length of
the dot is. But if the edge length of the dots is less than the grid
spacing, ~he full b~ackness will not be obtained. This means
. .
GE~-83-021 5
.~.
~i
" v
lZ2~7~
l that at a fixed dot size t~.e minimum and the maximum grey value and
the number of non-redundant grey levels will be equally ixed.
Patterns with the same grey level may be selected SUC}I that the
reproduced print is substantially free from defects.
If, for example, the size of the individual dots is varied from one array
to another, while it is fixed for the dots in a 16-dot pel, then the
advantages explained with reference to the example with 1 dot per pel, 4
dots per pel and 1/-1 dot per pel are obtained.
1. The grey value range can be increased in the afore-mentioned e:cample
from 1.6 up to 100 ~ blackness.
2. The number of grey value stages can be increased almost arbitrarily
after the desired dot sizes have been chosen.
3. The number of suitable patterns for the same grey value is increased
considerably.
-) For the purpose of the afore-mentioned examp1e providing for a blackness of25 ~, it is possible, according to Figs. 7a and 7b, to have 4 grid points
with 1 dot each or 1 grid point with a cluster of four dots or 16 grid
points with 1/4 dot each. If, in addition, patterns of the same grey
value, which may be referred to as quasi-equivalent grey value patterns,
are mixed in a random generator, the print result or the reproduced image
will be largely free from random defects attributable to the printer or the
paper. ~efects attributable to the system are largely eliminated by the
choice of pattern. By varying the number of dots in the pel array and by
varying the dot size from one array to another, the range of possible grey
~) values and their scaling can be increased almost arbitrarily and the
susceptibility to defects at higher resolutions be reduced.
By storing the patterns and their associated dot sizes, a grey value
entered into the printer can be processed and printed quasi-analogously.
GE9-83-021 6
~2~71
l This is the basic concept underlying the invention. By varying the number
of the individual dots of a pel and/or their position and/or their cluster
and/or their si~e, a vast number of pels with equivalent grey values may be
represented. Thus, the invention offers the following substantial
advantages:
1. There is an arbitrary number of grey tones for each pel, so that the
resolution is subs~tantially higher than that previously obtainable
with identical pel sizes.
2. As there is a plurality of patterns for each grey tone, the choice of
the most favourable cluster and/or the most favourable dot size is
controlled by microcode.
3. As a result, the patterns may be adapted to the characteristics and
natural frequencies of the reproduction means used in the subsequent
processing steps.
4. The savings in software are enormous.
2~ , ~
5. Detrimental paper and macLine frequencies are taken into account.
Fig. 8 is a diagram showing the dependence of the grey values of the
patterns of Figs. 6a, c, d and e on the diameter of the individual dots
from the smallest to the largest dot diameter and from the smallest to the
largest cluster diameter. The diagram shows how the grey values of the
various patterns depend on the dot and the cluster diameters, respectively.
For selecting the patterns for the individual pels, patterns susceptible to
all sorts of disturbances have to be determined either empirically or by
computation and be sampled out.
Fig. 9 is a block diagram of the new print or reproduction method for image
information. Input 1 receives analogue signals of a dot grid from a video
pick-up tube and the like. It is, of course, also possible for input 1 to
be supplied with digital data. Input 1 is followed by ~l analog-to-
,
r .
GE9-83-021 7
"
{
~;~2157~
1 digital converter ~ whicn is controlled by a clock generator 3. It is
essential with this analo~-to-digital converter that the quantization
effected is relatively fine. Owing to the inertia of the human eye, a
quantization comprising at least 3~ stages is expedient. It is not
difficult, however, to have up to 256 quantization stages which, assuming
there is an equal nu~ber of grey values, would lead to a grey value scale
no longer recognizable by the human eye. This analogue-to-digital
converter is followed by a processor, for exa~ple, a microprocessor 4 which
is also clock-controlled by clock generator 3. This processor controls the
reproduction unit which, in the present case, is an electroerosion printer.
Each grey value thus stored is associated with a grey value in microcode 5.
The microcode may optionally refer to a customer- or user-addressable
control unit 6, by means of which the grey value scale stored in microcode
5 may be changed. In such a case, microcode 5 controls the table look-up
in a storage 8 containing the dot patterns for the different grey values.
A storage 9 is addressed in a similar marner. This storage serves to
accommodate the associated current and voltage values, as the different
diameters or sizes of the dots of a pel are obtained by varying their
current or voltage values. For selecting the dot patterns stored in
storage 8, a-ran~om generator 7 is provided which is also controlled by
clock generator 3. In this case, the two storages 8 and 9 control the
driver stages 10 which effect the print process. It is assumed that the
driver stages apply the associated currents to the electrodes for printing
the pels, so ~hat the step in which printing is effected is designated as
11. The new method ensures grey values ranging, for example, from 1.6 to
100 % blackness and an arbitra;rily finely distributed grey scale. It is
particularly advantageous that when the analogue signals received are
quantized, the maximum amplitude values may be compressed. Similarly,
quantization may be effected to a logarithmic scale. The new method
improves an image which is coarse-grained per se to such an extent that its
grains are no longer recognizable. The new method makes it possible for
the first time to obtain image representations that are comparable to
photographs taken on a fine-grained film by means of a high-precision lens.
Thus, the new method permits the production of images of photographic
3 J quality, using printing methods employed for transparent foils.
., " ,
. ~
GE9-83-021 8 i
"
.:. 5