EP0914945A2 - Process for regulating the inking in a printing machine - Google Patents

Abstract

The offset printing machine has a sheet (3) with a matrix of color printed elements (4) that are scanned by a sensor (2). The color pattern is compared with that of a reference and a sensitivity matrix is established. In particular the effect of variations such as sheet thickness are established and the printing process modified.

Description

The invention relates to a method for regulating the color application in a Printing machine according to the preamble of the independent claim.

Such a is referred to as the color distance controlled control process e.g. known from EP-B2-0 228 347 and from DE 195 15 499 C2. With these The process involves printing a printed sheet in a number using the printing press of test areas with respect to a selected color coordinate system measured colorimetrically. The color coordinates obtained thereby become the Color distance vectors to target color coordinates based on the same color coordinate system calculated. These color distance vectors are created using Sensitivity matrices converted into layer thickness change vectors, and the Regulation of the color guide of the printing press is based on the Color distance vectors converted layer thickness change vectors made. The fields from with the actual print image are used as test areas printed color control strips used.

In the meantime, i.a. scanning devices known as scanners have become known, which allow the entire image content of a printed sheet in large numbers of relatively small picture elements with reasonable effort and in a very short time measured colorimetrically or spectrophotometrically. These scanners offer the basic metrological requirements for the regulation of Ink guide of a printing machine not only to use test strips printed with it, but the color information from all picture elements of the whole actual To use the printed image for this purpose. A difficulty with this as a so-called Measurement in the picture, however, is by the in the Given the four-color printing problem of the black component, to which As is well known, not only the printing ink black itself, but also that superimposed colored colors contribute. A reliable determination of the for the calculation of the input variables required for the color control Color value gradients for all occurring in a printed image, very Different printing situations are not possible using the usual methods. Another difficulty arises from the enormously high required Computational effort and the associated unacceptably long time Computing times.

Based on this prior art, it is an object of the present Invention to improve a method of the generic type in that it also for the so-called measurement in the image with practically justifiable effort can be carried out. The measurement in the image is the colorimetric Measurement of the entire printed image in a very large number (typically several thousand) of small picture elements (typically a few millimeters in diameter) and the evaluation of those obtained from the individual picture elements colorimetric values for the calculation of the control variables for the coloring of the Printing machine understood. Another object of the invention is that Process also to improve the influence of everyone involved Printing inks, in particular also the printing ink black, can be safely separated can.

The solution to this problem underlying the invention results from the im characterizing part of the independent claim 1 described features. Particularly advantageous refinements and developments are the subject of dependent claims.

Fig.1

ein Prinzip-Schema einer Anordnung zur Steuerung bzw. Regelung einer Druckmaschine und

Fig. 2

eine Einrichtung zur bildelementweisen Abtastung von Druckbögen und zur Auswertung der Abtastwerte für die Steuerung bzw. Regelung einer Druckmaschine.

The invention is explained in more detail below with reference to the drawing. Show it:

Fig. 1

a schematic diagram of an arrangement for the control or regulation of a printing press and

Fig. 2

a device for the pixel-by-pixel scanning of printed sheets and for evaluating the samples for the control or regulation of a printing press.

1, a printing press 1, in particular a multi-color offset printing press, Print sheets 3, which the desired print image and possibly additional Have pressure control elements. The printed sheets 3 are up to date Removed printing process and a spectrophotometric scanner 2 fed. This scans the printed sheets 3 essentially over the entire Surface from pixel to pixel. The size of the individual picture elements 4 is typical about 2.5 mm x 2.5 mm corresponding to around 130,000 picture elements in one Sheet 3 common dimensions. The generated by the scanner 2 Samples - typically spectral reflectance values - are stored in one Evaluation device 5 analyzed and input variables for one of the Control device 9 assigned to printing press 1 which in turn processes the Coloring elements of the printing press 1 in accordance with these input parameters controls. The input variables are, at least in the case of an offset printing press, typically around zonal layer thickness changes for the individual inks involved in printing. The determination of the above Input variables or changes in layer thickness are made by comparing the Sampled values or quantities derived therefrom, in particular color measurement values (Color locations or color vectors) of a so-called OK sheet 3 with the corresponding sizes of one taken from the current printing process Printing sheet 3 in the sense that the by the input sizes or Changes in layer thickness caused changes in the settings of the Coloring organs of the printing press 1 the best possible adjustment of the color impression of the continuously generated printed sheets 3 to the OK sheet Have consequence. For comparison, another OK sheet 3 can also be used Reference can be used, for example corresponding default values or corresponding values obtained from prepress.

In this generality, the arrangement outlined essentially corresponds conventional, e.g. in EP-B2 0 228 347 and DE-A 44 15 486 in detail described arrangements and methods for color control of Printing machines and therefore requires no closer for the specialist Explanation.

The basic structure of the scanning device 2 and the evaluation device 5 go from Fig. 2.

The scanning device 2 comprises a substructure in the form of a somewhat inclined one rectangular measuring table T, on which the printed sheet 3 to be measured is positioned can be. A measuring carriage W is arranged on the measuring table T, on or in which is a spectrophotometric measuring unit, not shown here. Of the Measuring carriage W extends over the entire depth of the measuring table T in Coordinate direction y and is motorized across its width in the coordinate direction x linearly movable back and forth, with appropriate drive and control devices A are provided in the measuring carriage W and on or under the measuring table T.

The evaluation device 5 comprises a computer C with a keyboard K and a monitor M. The computer C works together with the drive and control device A on the measuring table T or in the measuring carriage W, controls the movement of the measuring carriage W and processes that of the one in the measuring carriage W located spectrophotometric measuring unit generated scanning signals. The scanning signals or quantities derived therefrom, typically for example the color values of the individual picture elements 4, can be displayed on the monitor M, for example in terms of pictures. Furthermore, monitor M and fact K are used for interactively influencing the evaluation processes, but this is not the subject of the present invention and is therefore not explained in more detail.
The spectrophotometric measuring unit comprises a plurality of reflectance measuring heads arranged linearly along the measuring carriage W and a spectral photometer optically connected to these measuring heads via an optical fiber multiplexer. The measuring unit scans the printed sheet 3 when moving the measuring carriage W back and forth across the entire printed sheet surface in a plurality - typically 320 - of parallel linear tracks spectrophotometrically, with each track having a large number of individual picture elements 4, the dimensions of which in the coordinate direction x are defined by the speed of movement of the measuring carriage W and the temporal resolution of the individual scanning processes. The dimensions of the picture elements 4 in the coordinate direction y are determined by the spacing of the scanning tracks. Typically, the dimensions of the individual scanned image elements 4 are approximately 2.5 mm × 2.5 mm, which results in a total number of approximately 130,000 image elements in the case of a printing sheet 3 of conventional size. After a complete scanning process, the reflectance spectra of the picture elements 4 are available as scanning signals for each individual picture element 4 of the printing sheet 3, which the computer C evaluates and processes further in the manner described below for determining the input variables for the printing press control device 9.

Scanning devices 2, which a printing sheet 3 in two dimensions allow to be measured densitometrically or spectrophotometrically, are widespread in the graphics industry and therefore need for the Expert no further explanation, especially for the concerns of the present Invention the pixel-by-element measurement of the printed sheets 3 also by means of a Handheld colorimeter or handheld spectrophotometer could be done. A special one A suitable scanning device 2 corresponding to the one outlined above is e.g. in the German patent application 196 50 223.3 described in all details.

An essential aspect of the present invention is the inclusion of the printing ink black in the calculation of the input variables for the printing press control or in the calculation of the intermediate variables required for these input variables. For this reason, the printed sheets 3 are not only measured in the visible spectral range (approx. 400-700 nm), but also at at least one point in the near infrared, where only the printing ink black has a significant absorption. The reflectance spectra of the individual picture elements 4 thus consist of reflectance values in the visible spectral range, typically 16 reflectance values at intervals of 20 nm each, and a reflectance value in the near infrared range. Color values (color coordinates, color vectors, color locations) relating to a selected color space are calculated from the reflectance values of the visible spectral range. It is preferable to choose a color space that is equally spaced in terms of perception, typically the so-called L, a, b color space according to CIE (Commission Internationale de l'Eclairage). The calculation of the color values L, a, b from the spectral reflectance values of the visible spectral range is standardized by CIE and therefore requires no explanation. The reflectance value in the near infrared is converted into an infrared value I, which corresponds qualitatively to the brightness value L of the color space. This is done analogously to the calculation formula for L according to the relationship I. = 116 3rd II Iin - 16th wherein I _i measured in the respective picture element 4 infrared reflectance and I measured _in the unprinted at a location of the printing sheet 3 infrared reflectance mean. The infrared value I, like the brightness value L, can therefore only assume values from 0-100. The calculation of the color values L, a, b and the infrared value I from the spectral reflectance values takes place in the computer C. For the sake of completeness, it should be mentioned that the determination of the color values L, a, b (or corresponding values of another color space) also without spectral scanning could take place using suitable color measuring devices.

The color and infrared values L, a, b and I present for each individual picture element 4 after the scanning of a printing sheet 3 form the starting point for the calculation of the input variables for the printing press control device 9. These calculations are also carried out in the computer C. For the The following description is for each picture element 4, the three color values L, a, b (or the corresponding values of another color system) and the value quadruple comprising the infrared value I for simplifying purposes as the (four-dimensional) color vector F of the relevant picture element 4, so: F = (L, a, b, I)

The term "color locus" in the four-dimensional color space is understood to mean a point whose four coordinates in the color space are the four components of the color vector. The color difference of a picture element 4 with respect to a reference picture element 4 or with the corresponding picture element 4 in a reference, typically an OK sheet 3, is referred to as the color distance vector ΔF, which depends on the relationship ΔF = (ΔL, Δa, Δb, ΔI) = F i - F r = (L i - L r , a i - a r , b i - b r , I i - I r ) results in which the values provided with the index i are those of the picture element 4 under consideration and the values provided with the index r are the components of the color vector of the reference picture element 4 or of the corresponding picture element 4 of the OK sheet 3. The color vectors of the picture elements 4 of the OK sheet 3 or another reference are often also referred to as target color vectors. The color distance ΔE between two picture elements 4 or a picture element 4 and the corresponding picture element 4 of the OK sheet 3 is understood to mean the absolute amount of the relevant color distance vector ΔF, that is ΔE = | ΔF | = {(L i - L r ) 2nd + (a i - a r ) 2nd + (b i - b r ) 2nd + (I i - I r ) 2nd } 0.5 where the indices i and r in turn have the meaning given. The computer C calculates the color distance vector ΔF for each picture element 4 of the current printing sheet 3 from the color vectors F determined on this and the OK sheet 3.

The input variables to be determined for the printing press control device 9, that is to say the zonal relative changes in layer thickness for the individual printing inks involved in the printing, are also shown vectorially for the following and are referred to collectively as the layer thickness change vector ΔD: ΔD = (ΔD c , ΔD G , ΔD m , ΔD s )

The indices c, g, m and s stand for the printing inks cyan, yellow, magenta and black, the correspondingly indexed components of the vector are the relative changes in layer thickness for the printing ink indicated by the index. The current layer thicknesses themselves can be represented as layer thickness vector D: D = (D c , D G , D m , D s ) where the indices have the same meaning.

As is known, an offset printing machine 1 is designed zonally, i.e. the printing takes place in a series of parallel zones (typically 32), at which Printing machine 1 separate coloring organs are provided for each zone, the Regulation - at least for the interests of the present invention - independent of one another. The mutual influence of neighboring pressure zones and their Consideration in the printing press control is not the subject of present invention and is therefore disregarded. The following Comments on the actual control of the printing press 1 or the calculation of the corresponding input variables for the press control relate each to a pressure zone and apply equally to all pressure zones.

According to the teaching of, for example, EP-B2 0 228 347 mentioned at the outset and with the additional consideration of the printing ink black according to the invention, the relative layer thickness changes ΔD of the individual printing inks required for the compensation of a color deviation from the reference (OK sheet 3) can be determined from the individual printing inks current print sheet 3 determined color distance vectors ΔF for reference (OK sheet 3) according to the equation ΔF = S * ΔD calculate, where S is a so-called sensitivity matrix, which as coefficients is the partial derivatives (gradients) of the four components L, a, b, I of the color vector F according to the four components D _c , D _g , D _m , D _s des Layer thickness vector D contains:

The coefficients of the sensitivity matrix S are usually referred to as color value gradients. In the explanations below, the summary term sensitivity matrix is used to represent these 16 color value gradients.
The sensitivity matrix S is a linear substitute model for the relationship between the changes in the layer thicknesses of the printing inks involved in the printing and the resulting changes in the color impression of the image element 4 printed with the changed layer thickness values. The sensitivity matrix S is not the same for all color locations in the color space, but applies Strictly speaking, only in the immediate vicinity of a color location, ie for each measured color vector F of the individual picture elements 4 is in the equation ΔF = S * ΔD strictly speaking, to use your own sensitivity matrix S.

It should be noted that it is possible to select the sensitivity matrix S only from the To form components L, a, b of a three-dimensional color vector F. On the Component I can be omitted if there are several Image elements 4 in relation to the flat coverage of the printing inks involved is independent of each other, which is the case in most cases.

Provided that the sensitivity matrices S are known, the matrix equation can be ΔF = S * ΔD according to the known rules of the matrix calculus according to ΔD ( ΔD = S -1 * ΔF ). The determination of the sensitivity matrices is discussed further below.

According to the above explanations, each printing zone comprises a large number, typically approximately 4000, individual picture elements. Experience has shown that the interference that occurs during printing generally does not have the same effect on the individual image elements or that not all image elements are affected by the same interference. The change in layer thickness calculated on the basis of a picture element can therefore, for example, lead to a complete compensation of the color deviation for one picture element, but for the other picture elements (in the same zone) either be insufficient or cause a change in direction or an enlargement of the color deviation. Since a different layer thickness change vector ΔD could result for each picture element in extremis, the matrix equation can be used ΔF = S * ΔD cannot be resolved independently for each picture element. The individual matrix equations for the individual picture elements must therefore be combined to form a matrix equation system which is overdetermined according to the reduced number of picture elements and which is to be solved according to the known methods of compensation calculation using a framework or secondary condition. In the case of 4000 picture elements, there is a system of 4000 matrix equations or 16000 simple algebraic equations with the four unknowns ΔD _c , ΔD _g , ΔD _m and ΔD _s . As a secondary condition for the solution of this system of equations, it is practically required that the mean quadratic error should be minimal. The mean square error is understood to mean the mean value of the squares of the color distances ΔE of the individual picture elements remaining after the corrected layer thicknesses have been applied.

The 4000 matrix equations mentioned can be summarized as follows: {ΔF} = {S} * ΔD

Here {ΔF} means a column vector with 16000 components (ΔL ₁ , Δa ₁ , Δb ₁ , ΔI ₁ , ΔL ₂ , Δa ₂ , Δb ₂ , ΔI ₂ ....... ΔL ₄₀₀₀ , Δa ₄₀₀₀ , Δb ₄₀₀₀ , ΔI ₄₀₀₀ ), {S} a matrix with 4 rows and 4000 columns and ΔD a column vector with the four unknowns ΔD _c , ΔD _g , ΔD _m and ΔD _s as components. The indices of the components of {ΔF} relate to the picture elements 4 1-4000, ie the components of {ΔF} are the determined components of the color distance vectors ΔF of the individual picture elements 4 compared to the corresponding picture elements 4 of the OK sheet. The rectangular matrix {S} results from a series of the 4000 sensitivity matrices S of the individual picture elements 4, ie {S} = (S ₁ S ₂ .... S ₄₀₀₀ ).

According to the rules of the equalization calculation and with the mentioned constraint, the solution of this system of equations can generally be represented as follows: ΔD = {Q} * {ΔF}

In it {Q} is a rectangular matrix with 4000 columns and 4 rows, which is calculated as follows: {Q} = {S} T * {S} -1 * {S} T where {S} ^T and {S} ^{-1 is} the transposed or inverse matrix to {S}.

As can be seen, the calculation of the layer thickness change vector ΔD is based on Although this is possible in principle, it requires enormous computing effort and corresponding expenditure of time that goes far beyond the limits of what is practically possible exceeds. In particular, a sufficiently fast control, such as in practice, especially in modern high-performance printing presses 1 is not feasible. The computing effort for determining the 4000 Sensitivity matrices (64,000 coefficients in total) for each Image elements 4 are not considered at all and move the Feasibility even further away.

This is where the invention begins. The most important basic idea of the invention is that the individual picture elements 4 according to certain criteria Groups or classes are summarized within which the Color difference vectors and the sensitivity matrices are summed and averaged and only continue with the mean values. In this way you can Equation system for the calculation of the layer thickness change vector considerably simplify (typically 81 instead of 4000 matrix equations per pressure zone) and with reasonable computing effort for practice sufficiently fast (<1 minute for the loosen the entire sheet 3). More details are given below.

The visual color impression (metrologically the color value, color location or color vector) of a picture element 4 is in offset raster printing by the percentage Raster values (area coverage) of the printing inks involved and, to a lesser extent Mass determined by the layer thickness of the printing inks. The grid values or Area coverage (0-100%) are due to the underlying printing plates fixed and practically unchangeable. Influenced the color impression and can therefore only be regulated via the layer thicknesses of the printing inks involved become. The terms "grid value" and "area coverage" are given below used synonymously. The totality of all possible combinations R of percentage screen values of the printing inks involved (usually cyan, yellow, Magenta, black) is referred to below as a grid space (four-dimensional).

Under given printing conditions (characteristics of printing press 1, nominal Layer thicknesses, material to be printed, printing inks used etc.) corresponds to each Raster value combination R a precisely defined color impression or color vector F the picture element 4 printed with this raster value combination R; so it exists a clear assignment of raster value combination R to color location or color vector F; the grid space can be clearly mapped onto the color space, although the color space is not completely occupied because it also contains non-printable color locations contains. Conversely, there is generally no clear relationship. The one Any raster value combination R belonging to color vector F can be empirically determined by Sample prints determined or using a suitable model that the Printing process sufficiently accurate under the given printing conditions describes, can be calculated. A suitable model is e.g. through the well-known Neugebauer equations for offset printing. The model sets the Knowledge of the reflectance spectra of single-color full tones, some Overlaying full tones and some grid fields all on the print involved inks with the nominal layer thicknesses of the inks ahead. These reflectance spectra can be measured very easily using a test print. If the characteristics of the printing press 1 are known, simple ones are sufficient Measurements on solid tones.

Using the model mentioned, it is possible in a manner known per se to determine the (16) coefficients of the sensitivity matrix S belonging to this raster value combination for any raster value combination R. All that is necessary is to change the nominal layer thicknesses of the printing inks involved, preferably individually in each case by, for example, 1%, and to use these changed layer thicknesses to calculate the associated color vectors and corresponding color spacing vectors compared to the color vector resulting from the nominal layer thicknesses. These color distance vectors ΔF and the underlying layer thickness change vectors ΔD are in the equation ΔF = S * ΔD used and this resolved according to the coefficients of the sensitivity matrix S.

When determining the coefficients of the sensitivity matrix S, the Area coverage values of the picture elements 4 are used. Are the Area coverage values from prepress are already known, so there is no need Measurement on test prints (exception: full tones).

According to the invention, there are now a limited number of possible Raster value combinations R the associated color vector F and the associated Sensitivity matrix S calculated in advance and stored in a table. This the entirety of all sensitivity matrices S and color vectors F calculated in this way containing table is referred to below as a raster color table RFT.

For the calculation of the layer thickness change vectors ΔD from the equation ΔF = S * ΔD As explained above, knowledge of the sensitivity matrix S belonging to the respective color locus or color vector F is required. In order to achieve this, according to the invention, the associated raster value combination R is calculated from the color vector F of the respective picture element according to a particularly advantageous calculation method, which will be explained in more detail below, and the associated sensitivity matrix S is calculated from the raster value combination R from the predicted Raster color table taken from RFT. In this way it is possible to quickly determine the required sensitivity matrices without undue computation.

According to a further idea of the invention, a number of, for example, 1296 equally spaced discrete screen value combinations R _iR (6 discrete screen percentages A _C , A _G , A _M , A _S for the printing colors cyan, yellow, magenta, black) are defined in the screen space: i 0 1 2nd 3rd 4th 5 A _C 0 20th 40 60 80 100% A _G 0 20th 40 60 80 100% A _M 0 20th 40 60 80 100% A _S 0 20th 40 60 80 100%

These 1296 discrete raster value combinations R _iR are numbered according to the following formula with a unique raster index iR: iR = i (A C. ) * 5 0 + i (A G ) * 5 1 + i (A M ) * 5 2nd + i (A S ) * 5 3rd

I (A _C ) .... is the value of the index i for the respective discrete screen value of the respective printing ink. A sensitivity matrix S _{iR is} calculated for each of these 1296 discrete raster value combinations R _iR and stored in the raster color table RFT. The calculated color vector F _iR belonging to the discrete raster value combinations R _iR is also stored in the table RFT. The raster color table RFT thus contains a total of 1296 color vectors F _iR and 1296 associated sensitivity matrices S _iR .

The grid space is preferably quantized in two stages. In the first stage, for only 256 discrete halftone value combinations (corresponding to four discrete halftone percentage values 0%, 40%, 80%, 100% for each of the printing colors cyan, yellow, magenta, black), the associated color vectors and the are based on the offset printing model associated sensitivity matrices. In the second stage, the associated color vectors and sensitivity matrices for the missing raster percentage values 20% and 60% are calculated by linear interpolation from the color vectors and sensitivity matrices of the 16 nearest discrete raster value combinations. This again results in a total of 1296 discrete raster value combinations R _iR with 1296 associated discrete color vectors F _iR and 1296 associated sensitivity matrices S _iR . Of course, the grid space could also be reduced to a different number of discrete grid combinations, for example about 625 or 2401, but the number 1296 represents an optimal compromise between accuracy and computing effort in practice.

A sensitivity matrix S _iR whose associated discrete raster value combination R _{iR is} closest to the raster value combination R calculated from the color vector F is now assigned to a color vector F determined for a picture element 4. In other words, the calculated raster value combination is replaced by R each closest discrete halftone value combination R _iR and receives associated with the precalculated to this discrete halftone value combination R _iR sensitivity matrix S _iR.

In another view, the rest area is quantized by dividing it into a number of subspaces. All color vectors F, the calculated associated raster value combinations R of which fall into one and the same of these subspaces, are assigned the same sensitivity matrix S _{iR previously} calculated for this subspace. The subspaces are defined by the following six value ranges of the percentage raster portions (area coverage) of the four printing inks involved:
0 .... 10, 10 .... 30, 30 .... 50, 50 .... 70, 70 .... 90, 90 .... 100%

According to a further aspect of the invention, the (including infrared value I four-dimensional) color space is also subjected to quantization, ie divided into a number of subspaces, for determining the raster value combination R from the color vector F. For this purpose, a number of discrete color locations, each with discrete coordinate values, are defined in the color space. The four-dimensional color space can be quantized, for example, such that each dimension L, a, b, I of the color space can only assume 11 discrete values, resulting in a total of 14641 discrete color locations F _iF : i 0 1 2nd 3rd 4th 5 6 7 8th 9 10th L 0 10th 20th 30th 40 50 60 70 80 90 100 a -75 -60 -45 -30 -15 0 15 30th 45 60 75 b -45 -30 -15 0 15 30th 45 60 75 90 105 I. 0 10th 20th 30th 40 50 60 70 80 90 100

These 14641 discrete color loci F _iF are numbered according to the following formula with a unique color locus index iF: iF = i (L) * 11 0 + i (a) * 11 1 + i (b) * 11 2nd + i (I) * 11 3rd

For these discrete color locations F _{iF of} the color space, the associated halftone value combinations R _{iF are} calculated using the special calculation method explained below and, unless they coincide with a discrete halftone value combination R _iR , are replaced by the closest discrete halftone value combination R _iR . This results in a clear, pre-calculated mapping of the 14641 discrete color locations F _{iF of} the (four-dimensional) color space onto the 1296 discrete raster value combinations R _{iR of} the raster space. As already mentioned, this mapping is calculated in advance and stored in an assignment table referred to below as the raster index table RIT.

For the purposes of determining the raster value combinations R from the color vectors F determined for the picture elements 4, each color vector F determined for a picture element 4 is replaced by the closest discrete color location F _iF . The discrete raster value combination R _iR assigned to this discrete color location F _iF is then taken from the raster index table RIT and the corresponding sensitivity matrix S _{iR is} read out from the raster color table RFT and assigned to the color vector F. In this way, the sensitivity matrix S can be determined with comparatively little computing effort and correspondingly quickly for any determined color vector F, although this can only be selected from one of the 1296 precalculated sensitivity matrices S _iR . In practice, however, this is sufficient.

For the above, it was assumed that the associated raster value combinations R can be calculated from the color vectors F. How this can be carried out particularly advantageously according to the invention is the subject of the explanations below.
First, the color space is divided into 81 sub-areas T _iT as follows: i 0 1 2nd L (0..120) 0..20..40 40..60..80 80..100..120 a (-90 .. + 90) -90 ..- 60 ..- 30 -30..0 .. + 30 +30 .. + 60 .. + 90 b (-60 .. + 120) -60 ..- 30..0 0 .. + 30 .. + 60 +60 .. + 90 .. + 120 I (0..120) 0..20..40 40..60..80 80..100..120

The total of 81 sub-areas T _iT are clearly numbered by a sub-area index iT defined according to the following formula: iT = i (L) * 3 0 + i (a) * 3 1 + i (b) * 3 2nd + i (I) * 3 3rd

Within each sub-area T _iT , the relationship between the color vector F and the associated raster value combination R, written as raster vector A, is approximated linearly by the following matrix equation: A = U iT * F

A means the raster vector with the raster percentage values A _C , A _G , A _M , A _{S of} the four printing inks involved as components, and U _iT a conversion matrix with 16 coefficients, which shows the partial derivatives (gradients) of the components of the raster vector according to the components of the color vector are. If the conversion matrices U _{iT of} the individual partial areas T _{iT are} known, the associated raster vector A or the associated raster value combination R can thus be calculated for each color vector F.

The problem is therefore reduced to the calculation of the conversion matrices U _iT for the individual sub-areas T _iT or more precisely for the color vectors F _iT from their centers. The conversion matrices are calculated using a weighted linear compensation calculation using the values from the raster-color table RFT explained above, that is to say the 1296 discrete raster value combinations R _iR and the associated discrete color vectors F _iR . For the compensation calculation, essentially only the inversion of a 4x4 matrix is required for each sub-area T _iT . The

Weight of the support points, ie the discrete color locations F _{iR of} the raster color table RFT, for the compensation calculation is determined according to a suitable function with the color distance between the support points and the respective color vector F _iT as parameters. The compensation calculation is linear, ie there are discontinuities at the transitions of the individual sub-areas T _iT , which are insignificant in practice.

In the following the actual control procedure for the coloring of the Printing machine 1 described in more detail.

At the start of a production run, the raster color table RFT and the raster index table RIT are calculated and saved in accordance with the above explanations for the prevailing printing conditions. If already determined and saved on a storage medium, the tables RFT, RIT can of course also be called up from this storage medium. On the basis of the two tables RFT, RIT, it is possible without substantial computing effort to assign the color vectors F determined for the individual picture elements 4 to the discrete sensitivity matrix S that applies in each case.
Now a current print sheet 3 is removed from the current printing process and measured by the scanning device 2 in the manner described, with the color vector F and the color distance vector ΔF for each picture element 4 from the corresponding picture element 4 of a previously measured analog OK sheet in the computer 5 23 is determined. The total number of picture elements 4 is, for example, around 130,000, so that with the usual 32 printing zones, the color vectors and color distance vectors of around 4,000 picture elements 4 each have to be processed per printing zone. The following explanations apply equally to one pressure zone and to all pressure zones.

A very important aspect of the present invention, as has been said before mentioned above, in the measure that the picture elements 4 according to certain criteria are classified and the measurement data belonging to each class Image elements 4 are averaged, in which case only the mean values are processed further become. The determined color vectors F and Color distance vectors ΔF understood. To classify the picture elements 4 Sensitivity classes formed. The sensitivities are for each sensitivity class (Sensitivity matrices S) and also the color vectors F similar, and therefore is one Averaging allowed. The one required for the control of the printing press 1

Layer thickness change vector ΔD is then calculated so that the mean quadratic error should be minimal across all sensitivity classes. Under middle quadratic error is the mean of the squares according to the Application of the corrected layer thicknesses remaining mean color distances of the Understanding picture elements 4 of the individual classes.

The areas of the sensitivity classes are preferably defined in the grid space. For example, 16-256 classes can be provided. The more classes there are, the fewer errors arise from averaging, but the more the computing effort increases. The definition of 81 classes, which result from dividing the grid space into 81 subspaces according to the following scheme, has proven to be a practical compromise: n 0 1 2nd A _C 0% .... 30% 30% .... 70% 70% .... 100% A _G 0% .... 30% 30% .... 70% 70% .... 100% A _M 0% .... 30% 30% .... 70% 70% .... 100% A _S 0% .... 30% 30% .... 70% 70% .... 100%

These 81 subspaces or sensitivity classes K _iK are clearly numbered by a class index iK as follows: iK = n (A C. ) * 3 0 + n (A G ) * 3 1 + n (A M ) * 3 2nd + n (A S ) * 3 3rd

As explained further above, the grid space comprises 1296 discrete grid value combinations R _iR . Exactly 16 raster _{value combinations} R _iR thus fall in each of the 81 subspaces and, accordingly, 16 (similar) sensitivity matrices S _iR fall into each sensitivity class K _iK .

For each picture element 4, the color vector F determined for this now the procedure described above using the raster index table RIT the associated raster index iR and from this the affiliation to one of the 81st

Sensitivity _classes K _iK determined. Using the raster index iR and the raster color table RFT, the sensitivity matrix S associated with the color vector F of the picture element 4 is further determined. After these steps, the color vector F, the color distance vector ΔF, the raster index iR, the sensitivity matrix S and the class index iK are thus available for each of the approximately 4000 image elements 4 of a printing zone. The raster index iR defines the raster value combination R, ie the percentage raster portions (area coverage) of the printing inks involved for the image element 4, the class index iK defines the affiliation of the image element 4 to a specific sensitivity class.

Next, the picture elements 4 or their color distance vectors ΔF become one Weighting process subjected to the influence of area coverage and of Positioning errors are taken into account.

For the subsequent averaging, it is advantageous if picture elements 4 with relatively small area coverage values are taken into account less or not, in particular picture elements 4 with area coverage values below 10% should be disregarded. As a result, a first weighting factor g1 that is dependent on area coverage can be defined as follows: g1 = 1 for area coverage> = 10% and g1 = 0 for area coverage <10%

Since the color values L, a, b, I are approximately proportional to the area coverage, the first weighting factor is preferably defined as follows on the basis of the color distance ΔE of the picture element to an unprinted location on the printing sheet 3 (Papierwiess): g1 = 1 for ΔE p 2nd > = 5 2nd and g1 = 0 for ΔE p 2nd <5 2nd

In it, ΔE _p ^{2 is} the square of the color distance of the picture element 4 from the unprinted position of the printed sheet 3 (paper white).

Another variant for the determination of the weight factor g ₁ is that it receives the value 1 as the maximum value if the sum of the area coverings of the respective picture element 4 falls below a predetermined threshold value, preferably the value 250. Otherwise, the weighting factor g1 is given a smaller value, in particular a value of 0. A combination of the two variants mentioned above is also conceivable.

The influence of positioning errors is taken into account by a second weighting factor g2. It is assumed that picture elements 4 are relatively insensitive to positioning errors in a homogeneous environment. A homogeneous environment is understood to mean that the color differences between the picture element 4 and its 8 neighboring picture elements 4 are relatively small. In this case the second weighting factor is set to g2 = 1. The second weight factor is reduced with increasing color distances. The second weight factor g2 can be determined as follows, for example: g2 = 1 for ΔE M <= 8 and g2 = (8 / ΔE M ) for ΔE M > 8

In this, ΔE ^{M means} the sum of the color distances between the picture element 4 and its 8 neighboring picture elements 4. A preferred definition of the second weight factor g2, which is less computationally complex, is given by the following relationship: g2 = 1 for ΔE M2 <= 8 and g2 = (8 / ΔE M2 ) 0.5 for ΔE M2 > 8

In this, ΔE ^{M2 means} the sum of the squares of the color distances of the picture element 4 from its 8 neighboring picture elements 4.

When determining the weight factor g2, the difference between the area coverage values and the neighboring picture elements 4 can also be used, with an increasing difference the weight factor g2 likewise receiving a smaller value going towards 0.
The two weight factors g1 and g2 become an individual combined weight factor g for each picture element 4 according to g = g1 * g2 combined. With these individual combined weight factors g, the color distance vectors ΔF of the individual picture elements 4 and the associated sensitivity matrices S are weighted multiplicatively. The weighted color distance vectors and sensitivity matrices of the individual picture elements 4 are referred to below as ΔF _g and S _g .

Subsequently, the averaging and normalization are carried out for all picture elements 4 of one sensitivity class according to the following formulas: ΔF MK = (Σ k (ΔF G )) / Σ k (g); S MK = (Σ k (P G )))) / Σ k (G)

The totals are generated across all picture elements in a class.

After this averaging, 81 average color distance vectors ΔF _MK and 81 average sensitivity matrices S _MK are available per printing zone. These are described in the basic relationship as described above ΔF = S * ΔD used and lead to a system of 81 matrix equations, which must be solved for the unknown layer thickness change vector ΔD. The resolution is again carried out by means of a weighted linear compensation calculation with the additional condition that the mean square error should be minimal, whereby the mean square error means the mean value of the squares of the mean color distances ΔE _{MK of} the individual sensitivity classes remaining after application of the layer thicknesses corrected by ΔD becomes.

The system of equations is as follows: {ΔF e.g. } = {P e.g. } * ΔD

Here {ΔF _z } means a column vector with 4x81 components, which results from the stacking of the 81 vectors ΔF _MK with their 4 components each. {S _z } is a matrix with 4 rows and 81 columns, which results from the 81 sensitivity matrices S _{MK being arranged} horizontally side by side. ΔD is a column vector with the four unknowns ΔD _c , ΔD _g , ΔD _m and ΔD _s as components.

According to the rules of the equalization calculation and with the mentioned constraint, the solution of this system of equations can generally be represented as follows: ΔD = {Q e.g. } * {ΔF e.g. }

In it, {Q _z } is a rectangular matrix with 81 columns and 4 rows, which is calculated as follows: {Q e.g. } = {P e.g. } T * {P e.g. } -1 * {P e.g. } T where {S _z } ^T and {S _z } ^{-1 is} the transposed or inverse matrix to {S _z }.

As a result of all these calculations, the desired layer thickness change vector ΔD with its components ΔD _c , ΔD _g , ΔD _m and ΔD _s are obtained for each printing zone, which are supplied to the control device 9 as input variables and thus cause the required adjustment of the coloring elements of the printing press 1 that the mean square error mentioned is minimized in each pressure zone.

Claims (14)

Method for controlling the ink application in a printing press, in which a printed sheet (3) printed with the printing press (1) is measured colorimetrically in a number of picture elements (4) with respect to a selected color coordinate system, from the color vectors (F) obtained for each picture element Color distance vectors (ΔF) are calculated for specified color vectors based on the same color coordinate system or determined from a reference printed sheet, these color difference vectors (ΔF) with the aid of sensitivity matrices (S) in input variables, in particular layer thickness change vectors (ΔD), for a control device ( 9) are converted for the coloring elements of the printing machine (1), and the regulation of the color guidance of the printing press (1) is carried out on the basis of the input variables converted from the color distance vectors (ΔF), in particular layer thickness change vectors (ΔD),
characterized, that for each measured image element (4) of the printed sheet (3) a separate sensitivity matrix (S) is determined, that the image elements 4 are classified according to sensitivity _classes (K _iK ), that the color distance vectors (ΔF) and the sensitivity matrices (S ) the image elements 4 belonging to a sensitivity class are averaged for each sensitivity class (K _iK ), and that the input variables mentioned, in particular 'layer thickness change vectors (ΔD), are calculated from the averaged color spacing vectors (ΔF _MK ) and the averaged sensitivity matrices (S _MK ) of all sensitivity _classes (K _iK ). Method according to claim 1,
characterized, that the sensitivity matrices (S) are determined from previously known area coverage values. Method according to claim 1,
characterized, that at least one measured value (I) in the near infrared range is obtained for each picture element 4, that the color vector (F) determined for each picture element 4 is four-dimensional, with three components of the color vector (F) being the coordinate values of an approximately equilibrium-dependent color space and the fourth Component is formed from the at least one measured value (I) in the near infrared range, that the color distance vector (ΔF) determined for each picture element 4 is correspondingly four-dimensional, and that the sensitivity matrix (S) determined for each picture element 4 is determined by the gradients of four components of the four-dimensional color vector (F) are formed according to the printing inks involved in the printing. Method according to claim 1,
characterized, that at least one measured value (I) is obtained in the near infrared range for each picture element 4, that the color vector (F) determined for each picture element 4 is four-dimensional, three components of the color vector (F) being the coordinate values of an approximately equally spaced color space and fourth component is formed from the at least one measured value (I) in the near infrared range, that the color distance vector (ΔF) determined for each picture element 4 is three-dimensional, and that the sensitivity matrix (S) determined for each picture element 4 is determined by the gradients of the three components of the three-dimensional color vector (F) is formed according to the printing inks involved in the printing. Method according to one of the preceding claims,
characterized, that the color distance vectors (ΔF) and the sensitivity matrices (S) of the picture elements 4, each belonging to a sensitivity class (K _iK ), are weighted averaged for each sensitivity class, with each picture element being assigned 4 weight _factors (g1; g2), which are derived from the area coverage of the Image element 4 and / or the color distances of the image element 4 to its neighboring image elements 4 can be determined. Method according to claim 5,
characterized, that the area coverage of each picture element is determined with respect to the printing inks involved, that the weight factor (g1) of a picture element 4 receives the value 1, whom the average value or one of the area coverage of the picture element 4 exceeds a predetermined first threshold value, in particular the value 10%, and that the weight factor (g1) otherwise receives a smaller value, in particular the value 0. Method according to claim 5,
characterized, that the area coverage of each picture element (4) is determined with regard to the printing inks involved, that the weight factor (g1) of a picture element (4) receives a maximum value, in particular the value 1, if the sum of the area coverage of the respective picture element (4) has a predetermined threshold value, in particular falls below the value 250, and that the weight factor (g1) otherwise receives a smaller value, in particular the value 0. Method according to claim 6 or 7,
characterized, that instead of the area coverage for each picture element (4), the color distance to an unprinted point on the printed sheet (3) is determined, that the weight factor (g1) of a picture element (4) receives the value 1 if the color distance of the picture element has a predetermined second threshold value, in particular exceeds the value 5, and that the weight factor (g1) otherwise receives a smaller value, in particular the value 0. Method according to claim 5,
characterized, that for each picture element (4) the color distances to its immediately neighboring picture elements (4) are determined, that the weight factor (g2) of a picture element (4) receives the value 1 if the sum of the color distances has a predetermined third threshold value, in particular the value 8 , falls below, and that the weight factor (g2) otherwise receives a smaller value towards 0 with increasing sum of the color distances or with increasing difference in area coverage to the neighboring picture elements (4). Method according to claim 9 and one of claims 6 and 8,
characterized, that for each picture element (4) a weight factor (g) is determined, which is obtained by multiplying the weight factor (g2) calculated on the basis of the color distances of the picture element (4) from its neighboring picture elements (4) by that due to the area coverage or the color difference of the Image element (4) to an unprinted point on the printed sheet (3) gives the calculated weight factor (g1). Method according to one of the preceding claims,
characterized, that for a predetermined first number of discrete raster value combinations (R _iR ) of the printing inks involved in the printing, an associated sensitivity matrix (S _iR ) is calculated and stored in a raster color table (RFT) that for each picture element (4) the associated halftone value combination (R) is calculated for this color vector (F), and that the image element (4) is assigned that sensitivity matrix (S _iR ) from the halftone color table (RFT) whose associated discrete halftone value combination ( R _iR ) is the closest to the raster value combination (R) calculated for the picture element (4). A method according to claim 11,
characterized, that a second number of discrete color locations (F _iF ) is defined in the color space expanded to four dimensions by the infrared component (I), that the associated halftone value combination of the printing inks involved in the printing is calculated for each of these discrete color locations, that for each discrete color location the associated calculated grid value combination is replaced by the closest discrete grid value combination (R _iR ), and the assignments of the discrete color locations (F _iF ) to the discrete grid value combination (R _iR ) are stored in a grid index table (RIT) . Method according to claim 12,
characterized, that for the determination of the sensitivity matrix of a picture element (4) the four-dimensional color vector (F) determined for it is replaced by the closest discrete color location (F _iF ), that the raster index table (RIT) assigns the raster value combination assigned to this discrete color location ( R _iR ) that the sensitivity matrix (S _iR ) belonging to this raster value combination (R _iR ) is taken from the raster color table (RFT) and that this sensitivity matrix (S _iR ) is derived from the picture element (4th ) is assigned. Method according to one of the preceding claims,
characterized, that the sensitivity matrices (S _iR ) are calculated with the aid of a mathematical model of the underlying printing press (1) from measured values on solid areas printed with the printing press (1) and taking into account the characteristics of the printing press.

Images