Note: Descriptions are shown in the official language in which they were submitted.
8~(~
~IEI'HOD OF AUTO~TICALLY DETER~IINING ~ND
CONTROLLING TH~ PRINT QUALITY OF A PRINTER
_
Field of the Invention
The inven-tion concerns a computer-controlled method of automatically
determining the respective print quality value of a printer towards an
optimum value, using both the empirically detennined average image contrast
signal PCS and the standard deviation cPCS, said average image contrast
RB - Ri '
signal being derived from the relation R ~ where RB is the reflection
of the background and Ri is the reflection of the ith picture element or
dot.
The print quality is a subjective quantity and can be empirically de-
termined by a number of originals of equal quality or originals whose
quality increases (or decreases) in steps or continuously being subjected
to a statistically relevant number of individual evaluations by an equally
statistically relevant number of observers. In this manner, as a number of
tests with similar results have shown, relatively reliable print quality
data are obtained, particularly when the results are associated with
graded, say 6, quality classes.
The new method is to be used in particular to continuously monitor and, if
necessary, correct the image of a line printer, raster printer, or the
like. For this purpose, a numher oE print quality parameters have to be
observed which, depending upon the respective print method and the printer
used, are of different significance or importance. If characters are
partly cut off, the control of the hammer flight time, for example, is
incorrect or the type carrier is too fast or too slow, the hammer is
triggered too early or too late. Smudged or heavy printing may result, for
example, from the print hammer or print head impact being too strong or too
weak. Heavy or weak printing may also be caused by a ribbon or inking
cloth which is too new or worn, so that the ribbon or cloth structure
becomes visible. To overcome these difficu~ties, it is possible,
lZD~
for e~ample, to proceed from a predetermined mean print quali-t~f value which
may nevertheless be described as very good and which constitutes a reliable
optimum value for long--term control and adjus-tment. In accordance with the
inventi.on, this ls accomplished by the actual value of the irnage contrast
RB ~ ~
signal being derived from the individual values PCS - ~ for al~ n
picture elements by optoelec-tronically scanning all picture elernents i of a
character along the skeleton line and the background, by deriving from
n _ _ _ __
~ ,/ (PCS-PCSi)2
these scanning values a factor ~ = L ¦ - and a print
PCS i=l ~ n-l
quality factor x2 = PCS and by deriving ln x2 Erom said print quality
factor, that subsequently the actual print quality value DQ = aO ~ al ln
X2 is determined, and that by itexative automatic comparison of the stored
optimum print quality value DC t and the print quality thus detennilled,
control variables are derived and used in a closed control loop for
adjusting the difEerent printer parameters.
Further embodiments of the invention may be seen from the subclaims.
One way of carrying out the invention will be described in detai]. below
with reference to the attached drawings referred to in the specification.
For determining a subjective print quality factor, a representative
quantity of originals of the same and/or of different quality are submitted
to an equally representative group of evaluators for evaluation. The
evaluators are requested to rate the quality, using a factor from 1 to 6 or
interrnediate values 1.5, 2.5, etc. These evaluations are assessed by means
of statistical methods. As a result, a formula,X2, is obtained which ~ermit~
the evaluations of the evaluators to be mathema-tically predicted with
adequate accuracy if the relevant print quality parameters have been
determined by measuremen-t. One of these parameters is the print contrast
signa]. PCS. Using the above-mentioned method, this signal is determine~d by
scanning along the skeleton line according to DIN 66 223 and across the
background of the original. Thus, this pararneter is derived from the
diEferent reflection values oE the skeleton line relative to the back-
ground.
,,, _ _, .. . __.. , . __ ...... _ . . . . _. ... ... . . _ . , ,
~ 3 ~ ~ 2 ~ 4 8 8 ~
This scanning yields different values, depending upon the quality of the
scanned characters. The variation of this parameter in the original may
be described by means of three expressions:
1. PCS (mean value),
2. PCS (standard deviation or ~ PCS)' and
3. PCS (voids).
m us the value (l-FCS voids), as the average value, also denotes the
comp]eteness of a character. By combining these expressions in the
formula x = (1 - PCS (voids¦. PCS sufficiently accurate data are
2 PCS (standard deviation)
obtained as to the evaluation behaviour of this group of human evaluators
with regard to the print quality of the original. m is is shown in the
drawing, whose ordinate represents the evaluation values and whose
abscissa indicates the print quality factor x2. The dots represent
different samples evaluated and the crosses denote the regression line
which is a smoothed line showing the natural logorithm as described in
the following paragraph. For this purpose, it is assum~d that-the
totality of the evaiuated originals is represented with adequate accuracy
by this subquantity. It is assu~ed further than the standard deviation
has a value other than zero. This is normally the case.
'rhe x2 value can be linearized by means of the natural algorithm, thusyielding a-linear scale of 1 - 6. This linearization step permits
extrapolation to values of less than 1 or greater than 6. In most
cases, the characters will be ccmplete, so that the formula x2 may be
reduced to
x = PCS
2 PCS (standard deviation).
This formula always applies along the skeleton line.
If the PCS standard deviation is zero, which is unlikely~ this formula
cannot be used. me reason for this is that the true function of x2 is
exp cted to be logarithmic and that the formula for classifying the
print quality value is only an approximation from an expansion m-to a
series, so that the formula does not become too abstract. For tests
where the printed or typed original is of very good yuality, x2 should
be replaced by ln x2. This leads to a linear relationship.
OE 9-82~031
~Z~48B~
The above-mentioned formula can be used for all originals in which the ,
print contrast, the variations of the print contrast along the skeleton
line, the variations of the print contrast between different characters,
and the completeness of the individual characters, as well as the edge
definition of the individual characters are the essential components of the
print quality factor.
The control process is such that the controllable printer parameters aré
stored in a list storage and that the control variable causes the in-
dividual parameters, ordered according to their efficiency, to be suc-
cessively adjusted until the optimum print quality has been reached or has
been approximately reached by comparison with the stored value.
,
The scanning values may be derived in principle either from the characters
of a print line or from characters to be printed outside the image and
which are particularly suited for accurate adjustment to the optimum print
quality. Such patterns may be, for example, dot patterns, line patterns or
patterns with only one dot, or patterns with concentric circles or the
like. For implementing the method, a photoelectronic scanner, arranged in
a minimum of space directly above the print line, records the scanning
values, forming the pr mt quality factor DQ in a simple computer circuit.
.:
~ ' .
