Test pattern to compensate for lateral offset when detecting a compromised nozzle

Abstract

A test pattern (200, 300) for detecting a damaged nozzle (21, 22) of an inkjet printing device (100) is described. The test pattern (200, 300) has a matrix with matrix points (222), wherein the columns (223) of the matrix differ from one another such that, due to the differently designed columns (223), a lateral offset of up to ±V columns (223) can be detected and compensated. Thus, a damaged nozzle (21, 22) can be detected in a particularly reliable and robust manner.

Description

[0001] The invention relates to the detection of at least one impaired nozzle of a print head of an inkjet printing device.

[0002] An inkjet printing device for printing on a recording medium can comprise at least one print bar with one or more print heads, each print head typically having a plurality of nozzles. The nozzles are each configured to eject ink droplets to print pixels of a print image onto the recording medium. During printing, one or more nozzles of a print head may become impaired, e.g., due to contamination.

[0003] US 2017 / 0 120 647 A1 describes a method for evaluating the ejection performance of an inkjet head. DE 10 2019 211 687 A1 describes a method for print quality analysis in an inkjet printing machine using a computer. DE 10 2017 993 B3 describes a method for detecting defective print nozzles in an inkjet printing machine using a computer. For detection, a multi-line nozzle test pattern is printed. This pattern consists of a specific number of horizontal lines of periodically vertically printed, equally spaced lines arranged one below the other.

[0004] This document addresses the technical problem of enabling particularly efficient, reliable, and robust detection of a damaged nozzle of a print head. This problem is solved by the features of independent claim 1 and by the features of independent claim 9, respectively.

[0005] According to one aspect of the invention, a test pattern for detecting a defective nozzle of a print head of an inkjet printing device is described. The print head comprises K nozzles, e.g., with K>10, which are configured to print corresponding K pixels in corresponding K columns of a line of a print image on a recording medium.

[0006] The test pattern comprises, in particular, an N x M matrix with Z matrix points, wherein the matrix has N rows, M columns, and Z = N x M matrix points. Therefore, the test pattern has a sequence of N matrix points in each of the M columns, of which one or more are printed matrix points and one or more are non-printed matrix points, in particular such that the sum of the one or more printed matrix points and the one or more non-printed matrix points is exactly N, wherein N is preferably N > 1.

[0007] The pattern can be as wide as the print head 103, but it will typically be narrower and repeated periodically until the entire print head or the area to be examined is covered. It should also be noted that not only can the evaluation be based on multiple (>=1) patterns printed next to each other. When training the network, one can also choose to use >=1 patterns. The network is constructed accordingly as a CNN (Convolution Neural Network) with subsequent pooling and flattening such that the pattern width is not included in the training parameters, only the pattern height. This allows the network to be trained and operated with any number of patterns repeated in width.

[0008] A printed matrix dot in a specific column and in a specific row indicates that the nozzle corresponding to the specific column prints at least one pixel when printing the specific row of the test pattern. A non-printed matrix dot in the specific column and in the specific row indicates that the nozzle corresponding to the specific column does not print a pixel when printing the specific row of the test pattern.

[0009] The sequences of N matrix points in the M columns differ from each other in such a way that a lateral offset of up to ±V columns of the test pattern can be detected and compensated, with V ≥ 1.

[0010] According to a further aspect of the invention, a method and a corresponding control unit for an inkjet printing device are described. The method is designed to detect a defective nozzle of a print head of the inkjet printing device. The print head comprises K nozzles, e.g., with K>10, which are designed to print corresponding K pixels in corresponding K columns of a line of a print image on a recording medium.

[0011] The method comprises causing a test pattern formed as described in this document to be printed by the print head onto a recording medium, and thus a test print image corresponding to the test pattern to be printed onto the recording medium.

[0012] The method further comprises acquiring sensor data relating to the test pattern printed on the recording medium, i.e., relating to the test print image. Furthermore, the method comprises detecting at least one impaired nozzle of the print head based on the sensor data. The at least one impaired nozzle can be detected in a particularly robust and reliable manner using a pre-trained neural network.

[0013] In the following, exemplary embodiments of the invention are described in more detail with reference to the schematic drawings. In the drawings: Fig. 1a is a block diagram of an exemplary inkjet printing device; Fig. 1b an example test print image, Fig. 2 an exemplary compact test pattern; Fig. 3 a test pattern composed of several basic test patterns; and Fig. 4 is a flowchart of an exemplary method for detecting a degraded nozzle of a printhead.

[0014] The Fig. The printing device 100 shown in Figure 1a is designed for printing on a sheet-, sheet-, plate-, or tape-shaped recording medium 120. The recording medium 120 can be made of paper, cardboard, carton, metal, plastic, textiles, a combination thereof, and / or other suitable and printable materials. The recording medium 120 is transported along the transport and / or printing direction 1 (represented by an arrow) through the printing unit 140 of the printing device 100.

[0015] In the illustrated example, the printing unit 140 of the printing device 100 comprises two printing bars 102, wherein each printing bar 102 can be used for printing with ink of a specific color (e.g., black, cyan, magenta, and / or yellow, and possibly MICR ink). Furthermore, the printing device 100 typically comprises at least one fixing or drying unit (not shown) configured to fix a print image printed on the recording medium 120.

[0016] A printing bar 102 may comprise one or more print heads 103, which may be arranged in several rows next to one another, in order to print the pixels of different columns 31, 32 of a print image onto the recording medium 120. In the Fig. In the example shown in Figure 1a, a printing bar 102 comprises five print heads 103, each print head 103 printing the pixels of a group of columns 31, 32 of a print image onto the recording medium 120.

[0017] Each print head 103 of the printing unit 140 comprises in the Fig. 1a, a plurality of nozzles 21, 22, each nozzle 21, 22 being configured to fire or impinge ink droplets onto the recording medium 120. A print head 103 of the printing unit 140 can, for example, comprise several thousand effectively used nozzles 21, 22 arranged along several rows transversely to the transport direction 1 of the recording medium 120. By means of the nozzles 21, 22 of a print head 103 of the printing unit 140, pixels of a line of a print image can be printed onto the recording medium 120 transversely to the transport direction 1, i.e., along the width of the recording medium 120.

[0018] The printing device 100 further comprises a control unit 101, e.g. a control hardware and / or a controller, which is configured to control the actuators of the individual nozzles 21, 22 of the individual print heads 103 of the printing unit 140 in order to apply the print image to the recording medium 120 as a function of print data.

[0019] The printing unit 140 of the printing device 100 thus comprises at least one printing bar 102 with K nozzles 21, 22, which can be controlled with a specific line cycle to print a line (transverse to the transport direction 1 of the recording medium 120) with K pixels or K columns 31, 32 of a print image onto the recording medium 120 (with K>1000). In the example shown, the nozzles 21, 22 are immovably or permanently installed in the printing device 100, and the recording medium 120 is guided past the stationary nozzles 21, 22 at a specific transport speed.

[0020] Furthermore, the printing device 100 comprises a sensor unit 150 configured to capture sensor data, in particular image data, relating to a print image printed on the recording medium 120. For this purpose, the sensor unit 150 can be arranged behind the one or more printing bars 102 with respect to the transport direction 1. The sensor unit 150 can comprise, for example, an image or line scan camera.

[0021] The print quality of a print image printed by printing device 100 may be impaired. For example, a printed image may exhibit visible streaking in the transport direction 1. Such streaking may be caused by different inking of individual nozzles 21, 22, by isolated nozzle failures, and / or by a transverse offset of the ink ejection of a nozzle 21, 22.

[0022] To detect a print image impairment and / or to determine the cause of an impairment, as shown in the example in Fig. 1b, a test print image 160 is printed on a recording medium 120. The test print image 160 can, for example, be designed such that only partial sets of the nozzles 21, 22 of the printing device 100 are controlled in sections in order to enable reliable detection of a damaged nozzle 21, 22. The Fig. The test print image 160 shown in Figure 1b comprises several consecutive groups 161 of lines 162, wherein the individual lines 162 were each printed by a single nozzle 21, 22. The lines 162 of a specific group 161 have a specific spacing transverse to the transport direction 1, which makes it possible to reliably distinguish the individual lines 162 and the corresponding nozzles 21, 22 from one another. The lines 162 in the different groups 161 are printed by different nozzles 21, 22. This is done in particular such that each nozzle 21, 22 of the printing device 100 generates exactly one line 162 of the test print image 160.

[0023] Printing test print images 160 results in increased material consumption for ink and recording medium 120. Material consumption can be reduced by reducing the area of ​​the test print images 160, but this can have a negative impact on the quality of detection and / or identification of impaired nozzles 21, 22. This document describes a compact test print image or test pattern that enables reliable detection and / or identification of impaired nozzles 21, 22.

[0024] In particular, a test pattern is described which, on the one hand, is as compact as possible and thus, as a test print image, takes up as little space as possible on the recording medium 120, and yet still makes it possible to clearly assign the consequences of one or more incorrectly functioning nozzles 21, 22 during printing of the test pattern to the one or more responsible, incorrectly functioning nozzles 21, 22. Fig. 2 shows an example test pattern 200.

[0025] The test pattern 200 has, within each individual row 221, a plurality of matrix points 222 for the corresponding plurality of nozzles 21, 22 of a print head 103 or a print bar 102 of the printing device 100. In other words, in each individual row 221, exactly one matrix point 222 can be provided for each individual nozzle 21, 22. The matrix points 222 for a specific nozzle 21, 22 are each arranged in a column 31, 32, 223 of the test pattern 200.

[0026] The test pattern 200 can thus correspond to a raster or a matrix with N x M matrix points 222, with M columns 223 and N rows 221. The individual matrix points 222 can each indicate whether or not at least one pixel is printed by the respective nozzle 21, 22 in the respective row 221. A matrix point 222 can thus be a printed matrix point that is Fig. 2 is shown in black, or a non-printed matrix point that is shown in Fig. 2 is displayed in white.

[0027] The test pattern 200 preferably has one or more pattern properties. The test pattern 200 can include a pattern property relating to a required minimum spacing between printed matrix dots 222 within the individual rows 221 of the test pattern 200. The minimum spacing between two printed matrix dots can, for example, be at least one column 223 or more. Preferably, the minimum spacing between two printed matrix dots in a row 221 is at least four columns. For example, the test pattern 200 can have the pattern property that between two printed matrix dots 222 of the individual rows 221 of the test pattern 200, there is always at least one non-printed matrix dot 222 arranged, or there are always at least two non-printed matrix dots 222 arranged.By requiring a minimum distance between printed matrix dots 222 within the individual rows 221 of the test pattern 200, it can be achieved that the individual columns 223 of the test pattern 200 and thus the individual nozzles 21, 22 of the print head 103 or the print bar 102 can be distinguished from one another at any time, because the ink of adjacent printed matrix dots 222 within a row 221 does not flow together or these dots do not optically merge due to the imaging properties of the scanner or can still be distinguished by the scanner.

[0028] A further pattern property may relate to the frequency and / or number of printed matrix dots 222 in the different columns 223 of the test pattern 200. Preferably, all columns 223 of the test pattern 200 have the same number of printed matrix dots 222. This allows for a uniform detection quality when detecting impairments for all nozzles 21, 22 of the print bar 102 or the print head 103.

[0029] The test pattern 200 can have a spatial coding that makes it possible to detect and compensate for a lateral offset of the test pattern 200 transverse to the transport direction 1 due to the optical properties of the scanner, media deformation, and other interference. In particular, the spatial coding can ensure a clear assignment of a non-functioning nozzle 21, 22 within the entire width of the pattern 200. The spatial coding can be achieved, for example, by having all columns 223 of the test pattern 200 differ from one another. In other words, the different columns 223 of the test pattern 200 can each have different sequences of printed and non-printed matrix dots 222.

[0030] When designing the test pattern 200, a maximum possible lateral offset V can be taken into account. In particular, it can be taken into account that the test pattern 200 can be offset by a maximum of ±V columns 223 due to measurement errors of the sensor unit 150 and / or due to impairments of the nozzles 21, 22, e.g., with V between 5 and 20. The test pattern 200 can then have a spatial coding that is limited to detecting and compensating for a lateral offset of up to ±V columns 223. This can be achieved by a test pattern 200 that has at least 2*V columns 223 in a unique sequence. By reducing the number of different columns 223 of the test pattern 200, the required number of rows 221 of the test pattern 200 can also be reduced.In particular, 2*V columns 223 with different sequences of printed and non-printed matrix points 222 can already be defined by using log2(2*V) rows 221. The test pattern 200 can thus have, if necessary, exactly a number N of rows 221, where N corresponds to the next higher natural number above log2(2*V). The test pattern can have a maximum number of rows N ≤ 2*V columns 223 in a unique sequence. The test pattern would then be very compact.

[0031] Such a test pattern 200 can be used as a basic test pattern, which can be repeatedly arranged transversely to the transport direction 1 in order to provide the overall test pattern 300 for the print head 103 or for the print bar 102, as shown by way of example in Fig. 3. The overall test pattern 300 makes it possible to detect and / or compensate for a maximum possible lateral offset 301 of ±V columns 223 at any location. The overall test pattern 300 can correspond to a sequence of M, possibly identical, basic test patterns 200.

[0032] The basic test pattern 200 is preferably constructed such that one or more of the pattern properties described in this document are also valid for the overall test pattern 300 with the sequence of two or more basic test patterns 200. This preferably applies in particular to the pattern property with respect to the minimum spacing between printed raster cells 222 of the individual rows 221. The basic test pattern 200 can thus be designed to be periodically continuable, at least along the row direction.

[0033] In connection with the spatial coding, the test pattern 200 can be continued periodically. The test pattern 200 typically does not have to enable a unique assignment across the entire plurality of nozzles 21, 22 of a print head 103, because a rough orientation regarding the location of the imaged area already exists based on an image from the sensor unit 150. The functionality of the spatial coding therefore typically only has to cover the fine range. The functionality of the spatial coding typically only has to occur over a range that corresponds to the inaccuracy of the rough orientation based on the assignment across the imaged area. If an assignment of the location of a nozzle 21, 22 to an imaged area is only possible with an accuracy of, for example, V = 10 print image division units 222, the location only has to be uniquely assigned within a range of 2 * V = 20 print image division units 222.Following this area, the test pattern 200 can be repeated periodically, as shown in the example in . Fig. 3. As a result, the test pattern 200 has a reduced width, ie, a reduced number of columns 223, and a reduced height, ie, a reduced number of rows 221, so that a particularly compact test pattern 200 can be provided.

[0034] The test pattern 200, 300 can be designed such that the test pattern 200, 300 has exactly one printed matrix dot for each nozzle 21, 22, i.e., in each column 223. This makes it possible to provide a particularly compact test pattern 200, 300. Typically, however, it is advantageous for the test pattern 200, 300 to have, as a further pattern property, a certain redundancy of printed matrix dots for the individual nozzles 21, 22 in order to be able to detect nozzles 21, 22 that sporadically print unreliably and / or improperly. Therefore, it can be advantageous for the test pattern 200, 300 to have several printed matrix dots in each of the individual columns 223.

[0035] Another exemplary pattern property is that the test pattern 200, 300 does not have any printed matrix dots that are directly adjacent to each other in the diagonal direction, i.e., that are arranged in both directly adjacent columns 223 and in directly consecutive rows 221. Thus, the detection quality for detecting a degraded nozzle 21, 22 can be further increased.

[0036] The test pattern 200, 300 can be determined using a computational model that is designed, e.g., using a trial-and-error approach, to generate a test pattern 200, 300 that satisfies one or more, in particular all, of the pattern properties described in this document.

[0037] Fig. 2 shows an exemplary test pattern 200 which is suitable for reliably detecting incorrectly functioning nozzles 21, 22. The nozzle arrangement 102, 103 is arranged orthogonally to the printing direction 1. The columns 223 in the test pattern 200, 300 thus correspond to the individual nozzles 21, 22 of the nozzle arrangement 102, 103, in particular of the print head 103 or the print bar 102. An exemplary column 201 of the test pattern 200 from Fig. 2 indicates that the corresponding nozzle 21, 22 is actuated three times within the test pattern 200, namely in rows 2, 8, and 12. Each actuation can be a single pressure point, or a series of Q pressure points in succession. Any actuation in any other column 202, 223 of the test pattern 200 is Fig. 2 performed equally frequently.

[0038] For all columns 201, 202, and 223 of the test pattern 200, it is also true that for each pressure point 222, a pressure point 222 directly to the left or right of it is never pressed, because otherwise they would be too close together and would merge into a single unit, which would complicate the evaluation. The same condition also applies if the basic test pattern 200 is periodically repeated to the left or right to form an overall test pattern 300, as exemplified in Fig. 3 shown.

[0039] The test pattern 200 has a certain number of rows 221 such that each column 201, 202, 223 can be uniquely represented in order to satisfy the spatial coding and / or to fulfill one or more other pattern properties.

[0040] One of the possible malfunctions of a nozzle 21, 22 is the ejection of ink drops in an incorrect direction, which leads to an offset, in particular a transverse offset, of the print or image dots applied to the recording medium 120 and thus to print image defects such as streaks and / or color defects.

[0041] The test pattern 200 and the detection of an impaired nozzle 21, 22 with an incorrect orientation of the ink ejection based thereon can utilize the fact that misplaced print dots have the property that the misplaced print dots on the recording medium 120 merge with one or more print dots of neighboring other nozzles 21, 22 and / or that the misplaced print dots are merged with the one or more print dots of neighboring other nozzles 21, 22 by the optical properties of the scanning of the print image of the sensor unit 150.

[0042] The test pattern 200 can be designed such that a printed matrix dot in a specific column 201, 202 and in a specific row 221 within the same row 221 has one or more adjacent printed matrix dots (in one or more adjacent columns 223) at a defined distance 210. Preferably, the printed matrix dot in the specific column 201, 202 and in the specific row 221 has, on both sides, an adjacent printed matrix dot within the same row 221, wherein the adjacent printed matrix dots each have a defined distance 210, e.g., of one, two, or three columns 223, from the printed matrix dot. One or more non-printed matrix dots are arranged between the printed matrix dot and a directly adjacent printed matrix dot within the same row 221.

[0043] The printed matrix point in the second row 221 and in the column 201 of the test pattern 200 from Fig. 2 has on both sides, ie, right and left, a directly adjacent printed matrix dot within the second row 221. The distance 210 corresponds to two columns 223, in each of which a non-printed matrix dot is arranged.

[0044] The test pattern 200 can be designed such that the above conditions are met for all columns 223 of the test pattern 200.

[0045] The test pattern 200 can thus be designed such that each nozzle 21, 22 has the same number of neighboring nozzles 21, 22 that are actuated simultaneously, at a uniform distance 210. In particular, each nozzle 21, 22 can have the same number of simultaneously actuated neighboring nozzles that are located at a distance 210 or, if appropriate, at several different distances 210 from one another. For example, within the test pattern 200, each nozzle 21, 22 can have a neighboring nozzle 21, 22 to the left and right that is actuated simultaneously, at a distance of two nozzle pitches. In addition, there can be neighboring nozzles that are actuated simultaneously and are located at other distances 210, for example, at a distance of three nozzle pitches.By adhering to this formation rule for all nozzles 21, 22 when constructing the test pattern 200, it can be ensured that all nozzles 21, 22 are equally involved in the phenomenon of pressure dots merging in the event of misplacement. This enables detection of misplacements with uniform sensitivity.

[0046] Fig. 4 shows a flowchart of an exemplary (possibly computer-implemented) method 400 for detecting a degraded nozzle 21, 22 of a typically stationary print head 103 of an inkjet printing device 100. The print head 103 comprises K nozzles 21, 22, with K>10, typically K>100 or K>500 or K>1000, which are configured to print corresponding K pixels in corresponding K columns 31, 32 of a line of a print image on a recording medium 120. The method 400 can be executed by a control unit 101 of the printing device 100.

[0047] The method 400 comprises causing 401 a test pattern 200, 300, configured as described in this document, to be printed by at least one print head 103 onto a recording medium 120. The test pattern 200, 300 corresponds to a printed test print image on the recording medium 120.

[0048] It can be arranged that for each page of a useful print image to be printed, the test pattern 200, 300 is additionally printed on the recording medium 120. This allows a damaged nozzle 21, 22 to be detected particularly quickly.

[0049] The method 400 further includes the acquisition 402 of sensor data relating to the test pattern 200, 300 printed on the recording medium 120, ie, relating to the test print image. The sensor data can be acquired using a sensor unit 150, e.g., a camera or a line scanner, of the printing device 100.

[0050] Furthermore, method 400 includes detecting 403 at least one impaired nozzle 21, 22 of print head 103 based on the sensor data. The impaired nozzle 21, 22 can be detected particularly efficiently and reliably using a pre-trained neural network.

[0051] The neural network can be configured to receive the sensor data relating to the test pattern 200, 300 as input data at one input side. Furthermore, the neural network can be configured to receive the test pattern at another input, for the purpose of comparison with the sensor data, as input data at one input side. Furthermore, the neural network can be configured to specify, as output data at an output side, one or more nozzles 21, 22, each of which, with a certain probability, exhibits an impairment. The test pattern 200, 300 described in this document is particularly advantageously suited for evaluation using a neural network.

[0052] The neural network may have been trained in advance using training data, e.g., using a learning algorithm. The training data may comprise a plurality of data sets, e.g., 1,000 or more, or 10,000 or more data sets, wherein the individual data sets each comprise test input data and corresponding test output data for the neural network. The test input data may comprise test sensor data relating to a printed test pattern 200, 300, i.e., relating to a test print image, which may also be part of the test input data, and the corresponding test output data may indicate the one or more nozzles 21, 22 that were impaired during the printing of the test pattern 200, 300 for which the test sensor data was acquired.

[0053] The neural network can be trained using the training data in such a way that a specific error function, which depends on the detection accuracy for detecting impaired nozzles 21, 22, is reduced, in particular minimized, for the training data. For this purpose, a so-called backpropagation algorithm can be used, which allows the parameters of the neural network to be adjusted to reduce the error function.

[0054] Within the scope of method 400, additional training and / or adaptation of the neural network may be performed if necessary. In particular, during operation of the printing device 100, additional training data may be determined, which can be used to further train the neural network.

[0055] Within the scope of method 400, it can be effected that, in addition to the test pattern 200, 300, a master test pattern, ie a corresponding master print image, is printed by the print head onto the recording medium. The master test pattern, ie the corresponding master print image, can occupy a print area on the recording medium that is larger by a factor of 2 or more, in particular by a factor of 4 or more, than the test pattern 200, 300. The master test pattern, ie the master print image, can, for example, correspond to the print area associated with Fig. 1b. The master test pattern can be configured such that the master test pattern enables the detection of one or more impaired nozzles 21, 22 with a detection probability that is higher than the detection probability when using the test pattern 200, 300. It can also be used to generate training data for the method 400 in order to initially train the network or to extend the training to specific conditions, such as previously unused substrates or inks.

[0056] The method 400 may include capturing 402 master sensor data relating to the master test pattern printed on the recording medium 120, ie, relating to the master print image. The master sensor data may be captured using the sensor unit 105 of the printing device 100.

[0057] Furthermore, the method 400 can include determining training data for training the neural network based on the master sensor data. The master sensor data can be used to detect one or more impaired nozzles 21, 22 with a relatively high detection probability. These one or more nozzles 21, 22 can be used as test output data for the neural network. Furthermore, the corresponding sensor data relating to the test pattern 200, 300 can be used as test input data for the neural network. These test input data and test output data can be used as a data set to further train the neural network using a learning algorithm.

[0058] This allows repeated training of the neural network during operation of the printing device 100. For this purpose, a master test pattern can be printed at a specific repetition rate, e.g., every 10 or more, or 50 or more, or 100 or more pages of useful print images, and used to train the neural network. Thus, the quality of detecting impaired nozzles 21, 22 can be further increased based on the test patterns 200, 300 described in this document.

[0059] Not yet mentioned is the method of deliberately not activating nozzles when printing test patterns 200 and 300, thereby simulating nozzle failures. This method allows for a large amount of training material to be obtained. One A3 page per color is sufficient for a complete training session. However, the data obtained with pattern 160 also includes sideshooters (laterally offset print dots) that cannot be deliberately induced. The user is always dependent on the small number of sideshooters that randomly occur with low frequency.

[0060] This document thus describes a compact test pattern 200, 300 for detecting a defective nozzle 21, 22 of a possibly stationary print head 103 of an inkjet printing device 100. The print head 103 comprises K nozzles 21, 22, e.g., with K>10, which are configured to print corresponding K pixels in corresponding K columns 31, 32 of a line of a print image on a recording medium 120. The print head 103 typically comprises K>100, or K>500, or K>1000 nozzles 21, 22. A one-to-one relationship can exist between the nozzles 21, 22 and the columns 31, 32.

[0061] The test pattern 200, 300 comprises a matrix with matrix points 222, wherein the matrix of matrix points 222 has, in particular exactly, a number M of columns 223 for a subset or all corresponding K nozzles 21, 22 and, in particular exactly, a number N of rows 221, with N>1. The individual matrix points 222 of the matrix of matrix points 222 can each correspond to Q pixels in Q consecutive lines of a print image, with Q ≥ 1, in particular Q ≥ 2 or Q ≥ 10, preferably Q = 3. In other words, a matrix point 222 of the test pattern 200, 300 can optionally correspond to several pixels in consecutive lines of the test print image printed on the recording medium 120. A matrix point 222 of the test pattern 200, 300 can thus correspond to a line 162 with Q pixels each of the corresponding test print image.

[0062] The individual matrix points 222 can thus each correspond to exactly one pixel, if necessary, in order to provide the most compact test pattern 200, 300 or test print image possible. On the other hand, the detection probability for detecting impaired nozzles 21, 22 can typically be increased by using Q>1.

[0063] The test pattern 200, 300 can have in each of the K columns 223 a sequence of N matrix points 222, of which one or more are printed matrix points and one or more are non-printed matrix points. A printed matrix point is in the Fig. 2 and Fig. 3 as a black and / or filled square. A non-printed matrix point is shown in the Fig. 2 and Fig. 3 shown as a white and / or unfilled square. Final version

[0064] A printed matrix dot in a specific column 223 and in a specific row 221 indicates that the nozzle 21, 22 corresponding to the specific column 223 prints at least or exactly one pixel when printing the specific row 221 of the test pattern 200, 300. In particular, the printed matrix dot can indicate that Q pixels are printed on the recording medium 120 in Q rows of the test print image corresponding to the test pattern 200, 300, where Q rows of the test print image correspond to exactly one row 221 of the test pattern 200, 300.

[0065] A non-printed matrix dot in the specific column 223 and in the specific row 221 indicates, respectively, that the nozzle 21, 22 corresponding to the specific column 223 does not print any pixels when printing the specific row 221 of the test pattern 200, 300. In particular, the non-printed matrix dot can indicate that no pixels are printed on the recording medium 120 in any of the Q rows of the test print image corresponding to the test pattern 200, 300.

[0066] In a preferred example, the test pattern 200, 300 is designed such that a uniform number of printed matrix dots is arranged in each of the individual columns 223 of the test pattern 200, 300. For example, the K sequences of N matrix dots 222 each can have exactly R printed matrix dots 222. Typically, R > 1 and / or R < N. If necessary, R can be between N / 5 and N, preferably R = N / 4. By using a uniform number of printed matrix dots in the individual columns 223 of the test pattern 200, 300, a uniform detection probability can be achieved for all K nozzles 21, 22 of the print head 103.

[0067] The test pattern 300 can have a number G of basic test patterns 200 arranged one behind the other in the row direction, which runs transversely to the transport or printing direction 1, with G > 1, e.g. G = 2, or G = 3, or G ≥ 4. The number G of basic test patterns 200 can each comprise a partial matrix with K / M columns 223 and N rows 221 of matrix points 222. Furthermore, the number G of basic test patterns 200 can be identical. In one example, the overall test pattern 300 can thus be composed of G identically constructed basic test patterns 200 arranged next to one another with respect to the printing direction 1. In this way, a particularly compact test pattern 300 can be provided for reliable detection of impaired nozzles 21, 22.

[0068] When printing the test print image corresponding to the test pattern 200, 300 and / or when acquiring the sensor data related to the test print image, a lateral offset 301 transverse to the printing direction 1 may occur, which impairs the detection quality for detecting an impaired nozzle 21, 22. In particular, such a lateral offset 301 may result in a corresponding lateral offset for the nozzle 21, 22 that is detected as impaired.

[0069] The K sequences of N matrix points 222 in the corresponding K columns 223 can differ from one another in such a way that a lateral offset 301 of up to ±V columns 223 of the test pattern 200, 300 can be detected and compensated, with V ≥ 1, in particular V ≥ 2 or V ≥ 5 or V ≥ 10. Thus, by designing the K sequences of N matrix points 222 differently, it can be achieved that a lateral offset 301 of up to ±V columns 223 can be detected and compensated. The individual sequences can differ in how printed matrix points and non-printed matrix points are arranged within the respective column 223. By varying the sequences of matrix points 222, a particularly high detection quality can be achieved in a compact manner.

[0070] Thus, a test pattern 200, 300 for detecting a damaged nozzle 21, 22 of an inkjet printing device 100 is described. The test pattern has a matrix with matrix points 222, wherein the columns 223 of the matrix differ from one another such that, due to the differently designed columns 223, a lateral offset of up to ±V columns 223 can be detected and compensated. Thus, a damaged nozzle 21, 22 can be detected in a particularly reliable and robust manner.

[0071] As already explained above, the test pattern 200, 300 can have M basic test patterns 200 arranged one behind the other in the row direction. The sequences of N matrix points 222 in the K / M columns 223 of the individual basic test patterns 200 can differ from one another in such a way that a lateral offset 301 of up to ±V columns 223 of the test pattern 200, 300 can be detected and compensated. It can be ensured that K / M≥2*V. The detection of the lateral offset 301 can thus be enabled by the individual basic test patterns 200, so that the detection quality can be further increased and / or so that a particularly compact test pattern 300 can be provided.

[0072] The test pattern 200, 300 may have N = 2*V or fewer rows 221. Thus, a particularly compact test pattern 200, 300 can be provided, which further enables the detection and compensation of a lateral offset 301 of up to ±V columns 223.

[0073] Alternatively or additionally, the test pattern 200, 300 can be designed such that the number and / or frequency of printed matrix dots 222 in different rows 221 is different. A variation in the number and / or frequency of printed matrix dots 222 can thus be effected in the different rows 221. The number and / or frequency of printed matrix dots 222 can in particular, at least on average, decrease either with an increasing or decreasing row index for the rows 221 of the test pattern 200, 300, i.e., within the respective column 223. By varying the number and / or frequency of printed matrix dots 222 in different rows 221 of the test pattern 200, 300, the detection quality of the detection of impaired nozzles 21, 22 can be further increased.

[0074] Alternatively or additionally, the test pattern 200, 300 can be designed such that in the individual rows 221 of the test pattern 200, 300, a minimum number T of non-printed matrix dots is arranged between two directly adjacent printed matrix dots in the row direction, e.g., with T≥1 or T≥2. It can thus be ensured that within the individual rows 221, at least one non-printed matrix dot is arranged between two directly adjacent printed matrix dots. By providing a distance 210 between printed matrix dots that corresponds to a specific minimum number T of non-printed matrix dots, the detection quality of impaired nozzles 21, 22 can be further increased.

[0075] A possible impairment of a nozzle 21, 22 is that the ink ejected by the nozzle 21, 22 is ejected with a portion transverse to the printing direction 1, which leads to the pixels printed by the impaired nozzle 21, 22 within the test print image having a transverse offset transverse to the printing direction 1. The transverse offset can, based on Fig. 2, to the left or to the right side. A transverse offset to the left side can lead to • the pixels in a column 223 located to the left of the column 201 of the affected nozzle 21, 22 have an increased amount of ink and / or an increased color intensity; and / or • a column 223 located to the right of column 201 of the affected nozzle 21, 22 has a reduced color intensity.

[0076] The same applies to a transverse offset to the right side.

[0077] The increased and / or reduced color intensity of the pixels in the individual columns 223 can be used to detect a damaged nozzle 21, 22.

[0078] The test pattern 200, 300 can be designed such that in each of the K columns 223, ie in particular in all columns 223, of the test pattern 200, 300 at least one first printed matrix dot is arranged, which has at least one directly adjacent first printed matrix dot within the same row 221, which is spaced from the first printed matrix dot by exactly a first number L1 of columns 223, where, for example, L1 ≥ 2 or L1 ≥ 3. In principle, L1 >= 1 is also possible, provided the dots can be printed sufficiently small. If the first printed matrix dots are spaced by a number L1 of columns 223, exactly L1-1 non-printed matrix dots can be arranged between the two directly adjacent first printed matrix dots.

[0079] By providing a defined spacing between printed matrix dots 222 of the test pattern 200, 300, changes in the color intensity of the pixels of the corresponding test print image can be detected in a particularly reliable and robust manner, thereby increasing the detection quality of impaired nozzles 21, 22.

[0080] Thus, a test pattern 200, 300 for detecting a damaged nozzle 21, 22 of an inkjet printing device 100 is described. The test pattern has a matrix with matrix dots 222, wherein the printed matrix dots 222 in the individual columns 223 each have a defined spacing from one another in the row direction. Thus, a damaged nozzle 21, 22 with a transverse offset can be detected in a particularly reliable and robust manner.

[0081] The test pattern 200, 300 can be designed such that in K-k1 columns 223 of the test pattern 200, 300, a first printed matrix dot is arranged, which on both sides has a directly adjacent first printed matrix dot within the same row 221, which is spaced from the first printed matrix dot by exactly the first number L1 of columns 223. In particular, k1 can be 2*L1. The test pattern 200, 300 can thus have a defined spacing of printed matrix dots 222 on both sides. This can apply to all matrix dots 222, except for matrix dots 222 that are arranged in one or more columns 223 at the edge of the test pattern 200, 300.

[0082] Providing a defined spacing of printed matrix dots 222 on both sides results in an increase in color intensity on one side of the affected nozzle 21, 22 due to the transverse offset, and a corresponding reduction in color intensity on the other side of the affected nozzle 21, 22. This effect can be used to detect the affected nozzle 21, 22 in a particularly reliable manner.

[0083] The test pattern 200, 300 can be designed such that in each of the K columns 223, ie in particular in all columns 223, of the test pattern 200, 300, a second printed matrix dot is arranged, which has at least one directly adjacent second printed matrix dot within the same row 221, which is spaced from the printed matrix dot by exactly a second number L2 of columns 223, e.g., with L2≥2. The first number L1 of columns 223 and the second number L2 of columns 223 are different.

[0084] Preferably, the test pattern 200, 300 can be designed such that in K-k2 columns 223, i.e. in particular in all columns (preferably except at the edge), of the test pattern 200, 300, a second printed matrix dot is arranged, which on both sides has a directly adjacent second printed matrix dot within the same row 221, which is spaced from the second printed matrix dot by exactly the second number L2 of columns 223. In this case, k2 can depend on the second number L2 of columns 223, in particular k2 = 2 * L2. The test pattern can also be arranged at the edge, because the pattern is constructed in such a way that when it is periodically continued, the formation rules are also fulfilled in the edge zones.

[0085] This allows for several different spacings of printed matrix dots, on at least one side or on both sides. By varying defined spacings, the recognition quality can be further improved. List of reference symbols 1 Transport direction (of the recording medium) 21, 22 nozzle 31, 32 columns (of the print image) 100 printing device 101 Control unit / control device 102 pressure bars 103 Printhead 120 recording media 140 printing units 150 sensor unit 160 test print image 161 group of lines 162 Line 200 test samples 201, 202 column 210 distance 221 line 222 matrix point 223 column 300 test samples 301 lateral offset 400 Methods for Detecting a Degraded Nozzle 401-403 Procedural steps

Claims

[1] Test pattern (200, 300) for detecting a damaged nozzle (21, 22) of a print head (103) of an inkjet printing device (100); wherein the print head (103) comprises K nozzles (21, 22), with K > 10, which are designed to print corresponding K pixels in corresponding K columns (31, 32) of a line of a print image on a recording medium (120); wherein - the test pattern (200, 300) comprises a matrix with matrix points (222); - the matrix of matrix points (222) has K columns (223) for the corresponding K nozzles (21, 22) and N rows (221), with N > 1; - the test pattern (200, 300) has in each of the K columns (223) a sequence of N matrix points (222), of which one or more are printed matrix points and one or more are non-printed matrix points; - a printed matrix dot in a specific column (223) and in a specific row (221) indicates that the nozzle (21, 22) corresponding to the specific column (223) prints a pixel when printing the specific row (221) of the test pattern (200, 300); - a non-printed matrix dot in the specific column (223) and in the specific row (221) indicates that the nozzle (21, 22) corresponding to the specific column (223) does not print a pixel when printing the specific row (221) of the test pattern (200, 300); and - the sequences of N matrix points (222) in the K columns (223) differ from one another in such a way that a lateral offset (301) of up to ± V columns (223) of the test pattern (200, 300) is detectable and compensated, with V ≥1, - the test pattern (200, 300) is designed such that in each of the K columns (223) of the test pattern (200, 300) a first printed matrix dot is arranged, which has at least one directly adjacent first printed matrix dot within the same row (221) which is spaced from the first printed matrix dot by exactly a first number L1 of columns (223); and - L1 ≥2 or L1 ≥3, - the test pattern (200, 300) is designed such that in each of the K columns (223) of the test pattern (200, 300) a second printed matrix dot is arranged, which has at least one directly adjacent second printed matrix dot within the same row (221) which is spaced from the second printed matrix dot by exactly a second number L2 of columns (223); - L2 ≥2; and - the first number L1 of columns (223) and the second number L2 of columns (223) are different. [2] Test pattern (200, 300) according to claim 1, wherein - the test pattern (200, 300) has M basic test patterns (200) arranged one behind the other in the row direction, with M>1; - the number G of basic test patterns (200) each comprise a sub-matrix with L columns (223) and N rows (221) of matrix points (222), where L is the number of columns of the test pattern (200); - the sequences of N matrix points (222) in the K / M columns (223) of the individual basic test patterns (200) differ from each other in such a way that a lateral offset (301) of up to ±V columns (223) of the test pattern (200, 300) can be detected and compensated; and - L ≥2*V. [3] Test pattern (200, 300) according to claim 2, wherein the basic test patterns (200) are identical. [4] Test pattern (200, 300) according to one of the preceding claims, wherein the test pattern (200, 300) comprises a number of rows of N ≤2*V (221). [5] Test pattern (200, 300) according to one of the preceding claims, wherein - the test pattern (200, 300) is designed such that in the individual lines (221) of the test pattern (200, 300) a minimum number T of non-printed matrix dots is arranged between two printed matrix dots directly adjacent in the line direction; and - the minimum number T of non-printed matrix points is preferably T ≥1 or T ≥2. [6] Test pattern (200, 300) according to claim 5, wherein - the test pattern (200, 300) is designed such that in Kk columns (223) of the test pattern (200, 300) a first printed matrix dot is arranged, which has on both sides a directly adjacent first printed matrix dot within the same row (221), which is spaced from the first printed matrix dot by exactly the first number L1 of columns (223); - k depends on the first number L1 of columns (223); and - in particular k=2* L1. [7] Method (400) for detecting a damaged nozzle (21, 22) of a print head (103) of an inkjet printing device (100); wherein the print head (103) comprises K nozzles (21, 22), with K>10, which are designed to print corresponding K pixels in corresponding K columns (31, 32) of a line of a print image on a recording medium (120); wherein the method (400) comprises - causing (401) a test pattern (200, 300) formed according to claim 1 to be printed by the print head (103) onto a recording medium (120); - detecting (402) sensor data relating to the test pattern (200, 300) printed on the recording medium (120); and - detecting (403) an impaired nozzle (21, 22) of the print head (103) on the basis of the sensor data, in particular using a previously trained neural network.

Images