阅读说明：本技术 印刷图像中的缺失喷嘴探测 (Missing nozzle detection in printed images ) 是由 J·克里格 F·许曼 于 2019-12-11 设计创作，主要内容包括：一种用于通过计算机(9)确定印刷过程中的印刷错误的方法,该印刷过程在用于处理印刷任务的喷墨印刷机(7)中执行,在印刷过程期间,借助摄像机系统(10)检测所产生的印刷产品(2)并将该印刷产品数字化,将如此产生的摄像机图像(13)提供给所述计算机上的探测算法,在识别出印刷错误(14)时将报告发送给机器控制装置(6),机器控制装置通过废页分送机构将该印刷产品剔出,所述探测算法对所述摄像机图像的分色进行分离,在分色中探测所述印刷错误(14),使各个分色的图像关联成候选图像(21),对候选图像(21)进行过滤,最后将剩余的所探测的印刷错误输入到列表中,并且将该列表发送给印刷机(7)的机器控制装置。(A method for determining a printing error in a printing process by means of a computer (9), which printing process is carried out in an inkjet printer (7) for processing printing jobs, during which printing process a produced printed product (2) is detected and digitized by means of a camera system (10), the so produced camera image (13) is supplied to a detection algorithm on the computer, a report is sent to a machine control device (6) upon recognition of a printing error (14), which rejects the printed product by means of a waste page dispensing mechanism, the detection algorithm separates the color separations of the camera image, detects the printing error (14) in color separations, correlates the images of the individual color separations into candidate images (21), filters the candidate images (21), and finally inputs the remaining detected printing errors into a list, and sends the list to a machine control of the printing press (7).)

1. A method for determining a printing error in a printing process by means of a computer (9), which printing process is carried out in an inkjet printer (7) for processing printing jobs, wherein during the printing process a generated printed product (2) is detected by means of a camera system (10) and digitized, the camera image (13) thus generated is supplied to a detection algorithm on the computer (9), a report is sent to a machine control device (6) upon recognition of a printing error (14), which machine control device optionally ejects the printed product (2) by means of a waste dispensing mechanism,

it is characterized in that the preparation method is characterized in that,

the detection algorithm separates the color separations of the camera image (13), detects the printing errors (14) in the color separations, correlates the images of the individual color separations into candidate images (21), filters the candidate images (21), finally enters the remaining detected printing errors (14) into a list, and sends the list to a machine control (9) of the printing press (7).

2. The method of claim 1, wherein the first and second light sources are selected from the group consisting of,

it is characterized in that the preparation method is characterized in that,

the printing errors (14) relate to white line errors or black line errors (14) which are caused by faulty printing nozzles of the inkjet printer (7).

3. The method of claim 2, wherein the first and second light sources are selected from the group consisting of,

it is characterized in that the preparation method is characterized in that,

before being sent to a machine control (9) of the printing press (7), the computer (9) filters out, in a further method step, a false white-line error or a false black-line error (14b) from the list of white-line errors or black-line errors (14) by using a specific test method.

4. The method according to claim 2 or 3,

it is characterized in that the preparation method is characterized in that,

the computer (9) determines the faulty printing nozzle which is caused by the remaining list of detected white-line errors or black-line errors (14) and compensates the white-line errors or black-line errors (14) in accordance therewith by means of a suitable compensation method, respectively.

5. The method of claim 4, wherein the first and second light sources are selected from the group consisting of,

it is characterized in that the preparation method is characterized in that,

for a particular test method, the computer (9) generates a reference image from pre-print data of the print job, applies the detection algorithm to the reference image, thereby either obtaining knowledge about the resulting candidate of a pseudo white-line error or a pseudo black-line error (14b) and removing the candidate from the list of white-line errors or black-line errors (14); or to obtain knowledge about a region in the camera image (13) having a possible false white line error or false black line error (14b), and thus not to apply the detection algorithm to the region in the camera image.

6. The method of claim 5, wherein the first and second light sources are selected from the group consisting of,

it is characterized in that the preparation method is characterized in that,

the computer (9) generates the reference images at a plurality of sizes and/or resolutions, the computer uses the detection algorithm for the different reference images a number of times accordingly, and summarizes and uses the cognition obtained thereby.

7. The method of claim 6, wherein the first and second light sources are selected from the group consisting of,

it is characterized in that the preparation method is characterized in that,

the algorithm is not used for or excludes results from: the region is characterized by a large variation in the gray-scale values in a bounded local environment in the reference image.

8. The method according to any one of the preceding claims,

it is characterized in that the preparation method is characterized in that,

the list of white-line errors or black-line errors (14) is generated by the computer (9) by means of column sums in the filtered candidate image (21) by using limit values for the respectively found column sums in the candidate image (21).

9. The method according to any one of the preceding claims,

it is characterized in that the preparation method is characterized in that,

the computer (9) associates the color-separated candidate images (21) by means of a mathematical OR operation.

10. The method according to any one of the preceding claims,

it is characterized in that the preparation method is characterized in that,

-performing by means of morphological operations a filtering of said candidate images (21) by said computer (9).

11. The method according to any one of the preceding claims,

it is characterized in that the preparation method is characterized in that,

the detection algorithm is applied by the computer (9) to the generated camera image (13) a plurality of times, wherein the method is parameterized differently in order to detect differently pronounced black-line errors or white-line errors (14) and to logically relate the results of all color separations of all applications of the method to one another.

12. The method of claim 11, wherein the first and second light sources are selected from the group consisting of,

it is characterized in that the preparation method is characterized in that,

the generated camera image (13) is limited in each pixel to a maximum gray value in each case for each of the differently parameterized applications of the method.

13. The method according to any one of the preceding claims,

it is characterized in that the preparation method is characterized in that,

candidate images (21) of color channels are generated by dividing the generated camera image (13) into horizontal strips (15), wherein each strip (15) is reduced to an image signal (18) by appropriately averaging each column of each strip, and then white or black lines (14) are searched in the image signal by a specific search method, each row thus analyzed resulting in a row of the white line candidate image (21).

14. The method of claim 13, wherein the first and second light sources are selected from the group consisting of,

it is characterized in that the preparation method is characterized in that,

the white or black line search method identifies a black or white line (14) at a location by considering a restricted neighborhood around the pixel under consideration in the image signal (18).

15. The method of claim 14, wherein the first and second light sources are selected from the group consisting of,

it is characterized in that the preparation method is characterized in that,

the search method first convolves the image signals (18) by means of different filter cartridges and converts the results into logical signals by comparing the results with possibly different limit values, respectively, and then converts the logical signals into white line candidate image signals or black line candidate image signals by means of logical associations.

Technical Field

The invention relates to a method for checking the printing quality of an inkjet printer by means of a camera and a computer.

The invention belongs to the technical field of digital printing.

Background

Certain printing errors occur in the operation of inkjet printers, which printing errors are specific to these printer types. The most common is the so-called white line error, which occurs when the individual printing nozzles of the inkjet print head used deviate from their expected standard behavior. If such deviations exceed certain limits, the relevant printing nozzles are usually disabled, since they can interfere with the printed image. But such disabled printing nozzles would produce a corresponding white line error. The reason for this is because such errors are most pronounced in the case of monochrome printed areas, since in this case the underlying, usually white, printing substrate is exposed. On the other hand, when printing with a light printing ink (for example, opaque white) on a dark background, this error is manifested as a so-called black-line error. However, even in colored image regions, where a plurality of printing nozzles of different printing heads print respective color separations in an overlapping manner, a failure or disabling of the involved printing nozzles can lead to a corresponding color distortion in the printed image to be produced. Since the printing nozzles output ink linearly in the printing direction, the resulting printing errors are also linear, which leads to the term white-line printing errors/black-line printing errors.

The reasons for these deviations in the operation of the printing nozzles may vary. One major problem is that if the respective print head is not used for a long time and is not properly stored temporarily in a stationary state, the ink may dry. In this case, the dried ink blocks the nozzle outlet, which leads to a deviation of the printing point of the relevant printing nozzle or in extreme cases even to complete failure. In any case, the printing nozzle no longer prints exactly where the printing spot should actually be, and the printing intensity also deviates from the standard value that is actually expected. Besides dry ink, the intrusion of dust particles and similar dirt can also cause white line errors.

In order to find out these white line errors, various solutions are known from the prior art. Of course, it is most common to print test patterns and to detect white lines by automatically detecting and analyzing the processed test patterns. However, this solution has the disadvantages that: depending on the position and size on the printed substrate, printing the test pattern can result in waste pages. There is thus a method of: the method detects the generated printed pattern itself and detects the occurrence of white line errors from these printed images. This method also has the following advantages: only those white lines that actually interfere with the printed image currently to be produced or the printing nozzles that cause these white lines are detected.

German patent application DE 2017220361 a1 discloses a method for detecting and compensating for failed printing nozzles in an inkjet printer by means of a computer, which method comprises the following steps: printing a current printing image; recording the printed print image by an image sensor and digitizing the recorded print image by a computer; adding the digitized color values of the recorded print images of each column over the entire print image height and dividing the accumulated color values by the number of pixels in order to obtain column characteristics; subtracting the optimized column characteristics without failed printing nozzles from the original column characteristics to obtain column characteristic difference values; setting a threshold value of the maximum value, and defining that the printing nozzle fails if the threshold value is exceeded; applying the maximum threshold to the column characteristic difference, whereby in the resulting column characteristic each maximum marks a failed print nozzle; the marked printing nozzles are compensated for in the subsequent printing process.

However, this method has disadvantages in that: it cannot be robustly put into practice. This method is based on the fact that: there is only a very small difference between the reference image and the camera image. But it is just in practice not always the case. The reasons for this are, for example: mis-calibrated cameras, non-optimal or outdated white balance, different paper types in the printing mechanism, or non-optimal colors. Furthermore, it is preferable to detect white lines using a single color occupied area in a printed image, and therefore, this method can be used only limitedly without such an area in the printed image.

A white line inspection system is known from US 9,944,104B 2. In order to detect white lines, a simple comparison of limit values is proposed, which assumes that the material to be examined is homogeneous at this location. For images that do not satisfy this condition, it is proposed to generate a signal by subtracting the locally aligned reference image generated from the pre-print data. But in addition differential image calculations are required.

In contrast, patent application EP 3300907a1 describes how a white line detection system achieves quality improvement by using different methods depending on the printing situation, in particular in order to avoid detecting weak and therefore not severe white lines or to avoid identifying white lines as defects that are not well compensated but are not visible to the human eye. Here, as in US 9,944,104B 2, a step of generating a reference image is necessary in order to generate reference data for finding a white line which is desired to be eliminated.

Further, U.S. patent application US 2012/092409 a1 discloses a system and a method for identifying missing inkjets in an inkjet imaging system. The system and the method detect missing inkjets in an inkjet imaging system. Here, the system generates a digital image of the printed document that does not contain test pattern data. The digital image is processed to detect a band of light and correlate the location of the band of light with the location of the ink ejection in the print head. An identification of the ink color associated with the associated ink ejection location is then obtained by analyzing the color separated image and/or the color difference. The object of the present invention is therefore to find a method for determining a printing error in a printing process of an inkjet printer, which is more efficient than the known methods from the prior art and enables a better and more reliable determination of printing errors (in particular white line errors).

Disclosure of Invention

This task is achieved by a method for determining, by a computer, a printing error in a printing process, the printing process being performed in an inkjet printer for processing a printing task, wherein, during the printing process, the produced printed product is detected and digitized by means of a camera system, the camera image thus produced is supplied to a detection algorithm on a computer, a report is sent to a machine control upon recognition of a printing error, and, thereafter, the machine control optionally ejects the printed product through the waste delivery mechanism (ausschleusen), it is characterized in that the detection algorithm separates color separations of the camera image, detects printing errors in the color separations, associates images of the color separations into candidate images, the candidate images are filtered and the remaining detected printing errors are then entered into a list and sent to the printer. Therefore, the core of the invention is that: the printing errors are determined directly from the generated camera image of the detected and digitized printed product. In this case, the printing errors are detected directly in the separated color separations, since they can be detected more easily in this case than in the combined camera image. However, it is important to identify printing errors in the generated camera image. For example, if the resolution of the generated camera image is too low, information about the corresponding printing error is lost and the entire detection algorithm falls into the blank. It is also important to note that the camera typically provides RGB images, so that the separation of the individual separations of the resulting camera image will of course provide individual RGB separations, rather than CMYK separations corresponding to the color space of the inkjet printer used. However, this does not constitute a problem for the method according to the invention, since the method has the primary task of determining the exact location of the respective printing error or reliably detecting a printing error that impairs the printing quality. Which separations in the machine color space (i.e., which inks and which printheads) are affected by nozzle failure can be determined by the computer through the corresponding color space transformations. To improve the detection algorithm, in addition to the subsequent detection in color separations, the images of the individual color separations are also correlated into a common candidate image, which is then further filtered in order to ensure that substantially only printing errors leading to waste pages are detected accordingly. To facilitate subsequent detection of print nozzles causing a print error, all columns in the candidate image that contain the detected print error are marked.

Advantageous and therefore preferred developments of the method result from the corresponding dependent claims and from the description with the aid of the corresponding figures.

A preferred embodiment of the method according to the invention is that the printing error relates to a white line error or a black line error, which is caused by a defective printing nozzle of the inkjet printer. The algorithm is therefore primarily used to identify the white line errors already described, since primarily these printing errors reduce the printing quality of the printing process so greatly that waste pages are produced.

A further preferred development of the method according to the invention consists in that the computer filters out, in a further method step, the false white-line errors or the false black-line errors from the list of white-line errors or black-line errors by using a specific test method before sending them to the printing press. It is important here that the detection algorithm does not output any false positive errors. It is particularly very easy to mark thin and bright lines in the printed image to be produced, such as a bar code, as pseudo white lines. The detection algorithm should therefore check in a further method step with the aid of specific tests: whether the detected white line is indeed a true white line, so as to thus exclude: the desired printed image component is erroneously detected as a white line error and thus causes an undesirable additional waste page.

A further preferred embodiment of the method according to the invention is characterized in that the computer determines the cause from a list of remaining detected white-line or black-line errorsThe nozzles are printed in a defect and, accordingly, the white-line errors or black-line errors are compensated by suitable compensation methods, respectively. Although the actual aim of the method according to the invention is: the following printed products are unambiguously identified in the form of printed sheets: the printed product has such a quality-reducing white-line error (which would render the produced printed sheet a waste page), but it is of course also possible to use the information about the white-line error ascertained by the detection algorithm to ascertain the faulty printing nozzle for the cause and thus to be able to compensate for it in a suitable compensation method. Finally, the compensation of faulty printing nozzles can be carried out: the relevant inkjet printer continues to be used to complete the ongoing print job without replacing the print head.

A further preferred development of the method according to the invention consists in that, for a specific test method, the computer generates a reference image from the preprinted data of the print job, applies a detection algorithm to this reference image, whereby either knowledge is obtained about the resulting candidate for a false white line error or a false black line error and the candidate is removed from the list of white line errors or black line errors; or to gain knowledge about the area of the camera image with possible false white line errors or false black line errors and then not to use the detection algorithm in that area. The simplest way to detect a false white line is: a reference image is generated from good data (e.g. pre-print data) and then examined: whether the found structure that has been detected as a white line is also present in the reference image. If this is the case, a pseudo white line is logically involved. With this knowledge of finding, the process can be done in two ways. The found false white line errors can be removed from the list very simply, which is of course the simplest method. However, if it is desired to avoid that the same false white line error is found again by the detection algorithm in the continuous running of the method according to the invention, it is preferable to exclude from the detection according to the invention the region in the camera image where such a false white line error occurs.

Another preferred embodiment of the method according to the invention consists in that the computer generates reference images in various sizes and/or resolutions, in that the computer applies the detection algorithm accordingly a plurality of times to the different reference images, and in summarizing and using the recognitions obtained thereby. This approach not only improves the reliability of the detection algorithm for the specific identification white line, but also improves the reliability of the detection algorithm for determining the false white line error.

Another preferred embodiment of the method according to the invention is that the algorithm is not used in the following region or results from this region are excluded: this region is characterized by a large variation in the gray values in the delimited local environment of the reference image. Such regions (e.g. bar codes) are particularly easily detected as false white line errors or false black line errors and must therefore be excluded from the algorithmic check.

A further preferred refinement of the method according to the invention consists in generating a list of white-line errors or black-line errors from the column sums in the filtered candidate image (by using a limit value for the column sums in the candidate image, which are respectively taken). The white/black line errors that actually cause interference typically extend over a larger area of the detected camera image. In order to prevent the following, only the print columns (in which the ascertained printing errors exceed a certain limit value) are marked in the candidate images: even a very small, brief pause of the individual printing nozzles leads to a detection of a printing error (although this pause causes no disturbance at all or even does not involve a false white line error/false black line error), which is very likely to occur in the case of very short white line errors.

A further preferred development of the method according to the invention consists in associating the individual color separation candidate images by means of a mathematical or operation. This combination of color separations into candidate images has proven to be the most suitable computationally.

A further preferred refinement of the method according to the invention consists in that the filtering of the candidate images is carried out by a computer by means of morphological operations (morpholoscher operations). This allows, in particular, very short printing errors or white lines, which are usually in any case pseudo-white lines, to be filtered out, or which do not affect the quality of the resulting printed product or printed sheet so strongly that a waste page decision must be made here.

A further preferred development of the method according to the invention consists in applying a detection algorithm to the generated camera image a plurality of times by the computer, wherein the method is parameterized differently in order to detect different pronounced black-line errors or white-line errors and to logically relate the results of all color separations of all applications of the method to one another. In addition to the multiple use of the detection algorithm for multiple reference images, which is an optional method step of the method according to the invention, the detection algorithm can also be used for multiple times for the generated camera images. This improves the accuracy of the detection algorithm in particular when filtering out false white line errors or false black line errors, and is also advantageous for the hit accuracy when finding true white line errors or black line errors.

A further preferred embodiment of the method according to the invention consists in that the camera image is limited in advance to a maximum gray value in each pixel for each of the different parameterized applications of the method, which has the advantage that bright outliers in the paper white region (Papierwei β -Bereichen) which could distort the mean value can be filtered out.

A further preferred development of the method according to the invention consists in generating the candidate images for the color channels by dividing the image into horizontal strips, wherein each strip is reduced to a row signal by an appropriate averaging for each column of each strip, and then white or black lines are searched for in the row signal by a specific search method, and thus each row thus analyzed results in a row of the white line candidate image. This is an important feature of the method according to the invention, since the detection of white/black lines in the strip by means of the detection algorithm is more efficient than if the processing by means of the algorithm had to be done in the whole image.

A further preferred development of the method according to the invention consists in that the computer identifies a black or white line at a location by evaluating a restricted neighborhood around the pixel under consideration in the row signal by means of a white line search method or a black line search method. The following classification can be made by means of an evaluation of the immediately adjacent pixels: whether the error found is indeed a true white line error or a black line error. Only so can the false white line error or false black line error be excluded.

A further preferred embodiment of the method according to the invention consists in that the search method first convolves the line signals by means of different filter cartridges (filterkernel) and converts the result into a logic signal by comparing it with possibly different limit values, respectively, and then converts the logic signal into a white-line candidate line signal or a black-line candidate line signal by means of a logical association.

Drawings

The invention and advantageous constructional and/or functional developments of the invention are described in more detail below with reference to the drawings, according to at least one preferred embodiment. In the drawings, elements corresponding to each other are provided with the same reference numerals, respectively.

The figures show:

fig. 1 shows an example of the structure of a sheet inkjet printer;

FIG. 2 illustrates an example of an image detection system for printing inspection;

FIG. 3 shows an example of a detected camera image;

FIG. 4 shows detected swaths of a camera image;

FIG. 5 shows a detected band of camera images with marked white lines;

FIG. 6 shows a magnified section in a strip of a detected camera image, the magnified section having marked white lines;

FIG. 7 shows an image composed of image strips with marked white line candidates;

FIG. 8 shows a marked white line region in a camera image;

FIG. 9 shows the interference with the column average due to a paper white area or a single bright pixel;

fig. 10 shows a schematic flow of a method according to the invention.

Detailed Description

The field of application of this preferred embodiment variant is an ink jet printer 7. An example of the basic structure of such a machine 7 is shown in fig. 1, which is formed by a feeder 1, which supplies a printing substrate 2 (typically a printing sheet 2) into a printing unit 4, in which the printing substrate is printed by a print head 5, as far as a delivery unit 3. The present invention relates to a sheet-fed ink-jet printer 7, which is controlled by a control computer 6. As already described, during operation of the printing machine 7, individual nozzles in the print head 5 of the printing unit 4 may fail. The result is then the appearance of white/black lines, or in the case of multicolor printing, distorted color values. An example of such white/black lines 14 in the detected camera image 13 is shown in fig. 3.

In contrast to the methods known from the prior art, the embedding of the detection method for the white/black lines 14 in the overall flow of the printing process is different in the method according to the invention and no interaction with the printing press 8 is required anymore. The overall flow of the method in the first preferred embodiment variant is schematically shown in fig. 10:

1. after printing, the printed sheet is digitized by means of a camera system 10, which may be part of an inline image detection system 12. Fig. 2 shows an example of such an image detection system 12 using the method according to the invention. The image detection system comprises at least one image sensor 10 (typically a camera 10) integrated into the inkjet printer 7. At least one camera 10 captures a print image 13 generated by the printing press 7 and transmits the data to the computers 6, 9 for analysis processing. The computers 6, 9 may be separate computers 9 (e.g. one or more dedicated image processing computers 9), or the computers may also be identical to the control computer 6 of the printing press 7. At least the control computer 7 of the printing press 7 has a display 11, on which the result of the image check is displayed to the user 8. The image processing computer 9 is preferably used for the method described below, on which an image processing algorithm is run, which implements the method according to the invention. The camera image 13 thus produced has a lower resolution than the printing. The camera resolution is typically 670dpi and the print resolution is 1200 dpi. The resolution and optics must be chosen such that the white/black lines 14 appear as vertical, illuminated stripes 1-2 camera pixels wide. If the resolution is too high, the image 13 can first be reduced to a matching resolution by means of known image processing methods, in which case it has proved particularly advantageous to represent the pyramid image.

2. The camera image 13 is provided to a white/black line detection algorithm described in more detail below. Additionally, the camera images can also be provided in parallel to further evaluation processes.

3. If a white/black line 13 is recognized by the detection algorithm, the recognized white/black line is reported by the image processing computer 9 to the control computer 6 of the printing press 7, which then, in conjunction with further data of the printing press 7, makes a waste page decision and finally rejects the printed sheet 2 by means of the waste page dispensing mechanism.

4. Alternatively, the found white/black lines 14 may be subjected to more accurate analysis processing in order to identify a malfunctioning nozzle and use the information to compensate for the malfunctioning nozzle.

As can be seen from this flow, it is important for the function of the overall system 12 that the camera image 13 is processed in real time (schrithalten).

Unlike the prior art, the algorithm for white/black line detection mentioned here is only used for the camera image 13. Fig. 3 shows an example of a printed sheet 2 with detected camera images 13, wherein one of the camera images has a white/black line error 14. Alternatively, however, in a further embodiment variant, the subsequent filtering can also be carried out with the aid of the reference image. This is set forth in more detail in the following description.

The detection algorithm proposed here is based on the division of the detected camera image 13 into horizontal strips 15, 15a, 15 b. Here, the algorithm comprises the steps of:

1. the RGB separations are separated and, for each separation C, performed separately:

1.1 the camera image 13 is divided into strips 15 of about 1-10mm height, see fig. 4.

1.2 each swath 15 is averaged in the direction of travel of the sheet 2 (i.e., the y-direction). This gives the signal of the s-th strip 15 as I_s(x)

1.3 detecting the white/black lines 14 individually in each strip 15 by: for each x position a true value is calculated:

1.3.1 as an optional step, not considering having a grey value I_C,s(x)>G_maxBecause the white/black lines 14 are not visible in the bright image area.

1.3.2 WLC(x,s)＝(I_C,s(x)-I_C,s(x-1)>L)and((I_C,s(x)-I_C,s(x+1)>L)

or(I_C,s(x)-I_C,s(x+2)>L))or(I_C,s(x)-I_C,s(x-2)>L)and((I_C,s(x)-I_C,s(x+1)>L))

This expression checks whether there is a white/black line 14 of 1 or 2 pixels width with a brightness greater than the grey level L and excludes the edges of the image in an efficient manner fig. 5 and 6 respectively show image strips 15 with recognized white/black lines 14, which each have a corresponding marking 16, fig. 6 here shows an enlarged section 17 in the strip 15 with the marking 16 and the white/black line 14.

1.4 thus results in a black-and-white image W L CC (x, y) in which all white/black line candidates 14 are marked.

2. The individual color separated images W L Cc (x, y) are "or correlated" and a unique candidate image W L C (x, y)21 is then obtained, which is schematically illustrated in fig. 7, showing an image 21 consisting of image strips 15 with marked white/black line candidates 14.

3. The image W L C (x, y)21 can now also be filtered by means of morphological operations, so that very short white/black lines 14 can be filtered out by means of corrosion in the form of the structural element SE (erioderen):

0 1 0

0 1 0

the height of the SE can be variably adjusted, so that the minimum length of the white/black line 14 to be detected can be adjusted in advance.

4. In another embodiment, the same analysis process as described in steps 1-3 may be used in parallel for a possibly present reference image, which is generated directly by the RIP as an RGB image_{Reference to}(x, y)14 marks the areas of the printed image where white/black line detection is likely to produce false positives by the customer's material (Kundermotif.) these areas should be removed from W L C (x, y)21 in the camera image 13. to do so, first the W L C is analytically processed by morphological dilation (interpolation)_{Reference to}Region in (x, y.) this corresponds to W L C_{Reference to}(x, y) smoothing followed by W L C_{Reference to}(x, y) to filter W L C (x, y) 21:

WLC(x,y)←WLC(x,y)and(not WLC_{reference to}(x,y))

5. Next, all columns C containing white/black lines 14 in W L C (x, y)21 are probed_WLThis can be achieved by the limit value minW L PerColumn for the sum of the columns, to be precise in coded form, in W L C (x, y)21, no white/black line is 0 and 1, i.e. the term marked white/black line 14 in W L C (x, y)21 is counted:

CWL＝{x|Σ_yWLC(x,y)>minWLPerColumn}

furthermore, the method according to the invention can be adapted in further preferred embodiments:

for example, subsequent filters may be changed.

The number of white/black line candidates 14 must reach a minimum number per column in order to be reported as white/black lines 14

Define a maximum value for the luminance value of a pixel so that a very bright pixel does not distort the average value. Since the white/black lines 14 do not have very bright pixels in the camera image 13 at 670 dpi. I.e. all >50 grey values in the image are limited to 50;

checking: whether or not there is a strong structure at the important relevant position of the reference image, which in the reference image already results in a structure similar to a white/black line, and therefore has to be hidden in the camera image 13 (autosblenden). For this reason, the reference image does not have to be at full resolution, since it only needs to be roughly estimated whether a partial region of the reference image is structured or uniform; see step 4.

To implement a real-time application (schritthaltenen Anwendung), the above method can be implemented on a graphics card (GPU) as a computation accelerator.

The described detection algorithm can be implemented as part of the image detection system 12, which performs image inspection, then data for reporting to the operator or user 8 can also be derived from the W L C (x, y) image 21 by identifying continuous regions (Blobs: binary large objects) in the image 13 and marking these continuous regions in the overview image for the operator 8 for subsequent analysis processing fig. 8 shows an example of a camera image 13 with marked white/black line regions 20 as part of such a report.

In said other preferred embodiments, however, a reference image is generally required, which, in addition to the mentioned disadvantages, impairs the processing speed. However, the use of a reference image may further improve the quality of the method, since detected white/black lines 14 of false positives may be avoided.

The method according to the invention therefore has a number of advantages over the prior art. For example, in situations where the color deviation of the desired image from the camera image 13 is large (e.g., in a workflow for miscalibration of the camera 10), white balance, paper type, short white/black lines 14 are often overlaid in image noise/signal noise (unceghen). This disadvantage is eliminated by means of the method according to the invention. Furthermore, in the methods known in the prior art, the reference image must also be transmitted to the computer 9 at full resolution (e.g. 670 dpi). This can only be achieved with a high cost outlay by means of the existing technical measures. These costs can be saved since the algorithm proposed here works properly even without a reference image (or at least without a high resolution reference image). Finally, no reference image is in principle needed for detection, even though the reference image can be used to avoid false positive white/black line detection due to structures contained in the customer material. In particular, for detecting the white/black line candidates 14, it is no longer necessary to compare the reference image directly with the camera image 13.

There is also another particularly preferred embodiment of the method according to the invention, which further improves the method. For this purpose, the following two-stage algorithm is proposed, based on the previous embodiment variants:

in phase 1, white/black line candidates 14 are searched for in a targeted manner:

for this purpose, the algorithm proposed in the preceding embodiments is called multiple times with different parameters. The results of the algorithm runs are then logically associated. In addition, the algorithm is further improved in its operation.

The method comprises the following steps:

the algorithm is applied several times to the camera image 13 of the sheet 2. For different applications, the parameters are matched in the following way:

1. the camera image 13 is compressed in terms of its grey value/color channel value. The compression is performed such that it will be above a threshold S_maxIs limited to a threshold value S_max. This effectively suppresses all ratios S in the image 13_maxA brighter structure. In this step, therefore, it is very easy to find the white/black lines 14 in the dark areas (in the uniform and non-uniform areas). This compression is performed before the first step of the previous embodiment.

2. Here, the camera image 13 is also compressed in terms of gray values or color channel values. However, here the compression is carried out such that it will be above the threshold K_max(K_max>S_max) Is limited to a threshold value K_max. This compression is performed before the third step of the previous embodiment. Additionally, the local homogeneity of the image 13 is calculated by: when taking the average in the second step of the previous embodiment, the standard deviation of the column section is also calculated with respect to the average. Only a relatively uniform area (i.e., standard deviation)<σ_max) The white/black lines 14 in (a) are recorded into the candidate list. This filtration can be performed in the third step of the previous embodiment. In this way, the white/black lines 14 can be very easily found in bright, uniform areas. The human eye hardly sees the white/black lines 14 in the bright non-uniform areas anyway and is therefore disregarded.

The two results are correlated by means of a logical or and summarized to a white/black line candidate list. Alternatively, complex associations with other information may also be considered.

Furthermore, in the second step of the previous embodiment, in addition to simply taking the average, a different (with advantageous properties) averaging mean square method may be used for the generated image signal. This is for example:

replacing the mean by a median; the advantages are that: the method reacts very robustly to outliers.

Only when the brightness does not exceed the maximum brightness value G_max,meanIs averaged over the pixels of (a). The advantages are that: bright outliers or paper white areas that distort the average can be filtered out. This is exemplarily shown in fig. 9. Here, it is easy to recognize interference with the column average values in the upper and lower portions of fig. 9 due to a bright paper white area or a single bright pixel. However, the problem here is: the pseudo white/black lines 14b occur, and the contrast of the detected print image 13 is insufficient. Here, a printed image 13 with white/black line errors 14 detected by the camera 10 is imaged in the middle. From this printed image 13, a strip 15a with text and a strip 15b at the edge of the image are respectively cut out, from which strips image signals 18, 19 are then respectively generated. In the image signal 18 with the strip of text 15a, the effect of the white/black line error 14 (in the form of the corresponding peak 14a in the signal 18) is easily identified in the signal. In addition, a peak value of the pseudo white line/pseudo black line error 14b derived from the text display is shown. It can be seen that: in the image signal 18, it is difficult to distinguish between the peak value of the true white/black line error 14a and the peak value of the pseudo white/black line error 14b because both peak values 14a, 14b exceed the minimum detection height 19. In the lower part two image signals 18a, 18b are shown, which are for the case of image edge generation signals. Here, the minimum detection height 19 is exceeded only in the signal with the enhanced contrast 18a, and the white/black line 14 is reliably recognized. The minimum detection height 19 is not exceeded in the second signal with the lower contrast 18b, and accordingly no white/black line 14 is recognized.

In the third step of the previous embodiment, the white/black line 14 is detected by a threshold L for which two additional improvements are found in another embodiment:

1. depending on whether a 1-pixel wide or a 2-pixel wide white/black line 14 should be detected, two thresholds are used, depending on the camera resolution, it may also make sense to find a 3-pixel wide, 4-pixel wide, … N-pixel wide white/black line 14.

WLC(x,s)＝((I_C,s(x)-I_C,s(x-1)>L1)and(I_C,s(x)-I_C,s(x+1)>L1))

or((I_C,s(x)-I_C,s(x-1)>L2)and(I_C,s(x)-I_C,s(x+2)>L2))

or((I_C,s(x)-I_C,s(x-2)>L2)and(I_C,s(x)-I_C,s(x+1)>L2))

2. The threshold may be made dependent on the local environment of each pixel x, so that a higher threshold is used for the white/black line 14 in bright image areas than for image areas with lower brightness. As a measure of the local brightness, the gray values can be averaged in a narrow environment around the x position, excluding white/black lines 14 that may be present. Alternatively, a smooth median filter may be used for I_C,s(x)。

As a further advantageous development of the preceding embodiment, the algorithm may not be used for the RGB image 13, but the RGB image 13 may be converted beforehand by means of a suitable method into a grayscale image which has as good a contrast as possible for the white/black lines 14. The conversion operation applied here is:

computing the luminance channel from L ab color space

Computing a luminance or saturation value from the HSB color space

Averaging the appropriately weighted RGB color channels matched to the human eye

In phase 2, the pseudo white/black lines 14b are filtered out from the white/black line candidates 14 found in phase 1 by using one or more filters. To this end, the following improvements are provided over the previous examples:

by using a column filter for the white/black line candidate list, all of the following white/black line candidates 14 are removed from the white/black line candidate list: the white/black line candidates do not have at least a ratio number N in the same image column_col,minMore white/black line candidates 14. The idea behind this filter is to exclude very short and individual occurrencesAnd (4) error detection. Since the white/black lines 14 affect a column of areas in most actual printed material, false detections occur only locally scattered.

All the modifications described above are also carried out here in the preceding exemplary embodiment with the aid of the filter described in step four with reference to the image. Here, as an improvement, the size of the reference image is matched in advance. Also significant are: the reference images of different resolutions are processed multiple times and the results of this stage are aggregated before filtering out. This simulates, by the camera system 10, a loss of quality on a "perfect" reference image and is therefore able to effectively detect various structures that may result in white/black line type structures in the camera image 13.

The further particularly preferred embodiment has the following additional advantages compared to the previous embodiment: has a higher recognition performance of the white/black lines 14 while having a lower number of the pseudo white/black lines 14 b. However, reference image analysis processing is required for this purpose, the additional processing steps of which may lead to longer computation times of the computers 6, 9 used. Therefore, the decision as to which preferred embodiment to use should be based on the requirements of the particular application. The first proposed embodiment should be used more for the following print jobs: in this print job, white/black line detection is time critical or performance critical; the second proposed embodiment should be used even more for the following print jobs: it is particularly important to detect the white/black lines 14 thoroughly in the printed image of the print job and/or there is a high risk of the occurrence of false white/black lines 14b in the print job.

List of reference numerals

1 feeder

2 printing substrate

3 material collector

4 ink-jet printing mechanism

5 ink jet print head

Control computer of 6 ink-jet printer

7 ink jet printer

8 user

9 image processing computer

10 image sensor/camera

11 display

12 image detection system

13 detected print image

14 white/black line typographical errors

14a in the generated image signal, the peak value of the white/black line

14b in the generated image signal, and a peak value of a pseudo white line/pseudo black line

15 detected swath of the printed image

15a detected banding of the printed image with text content

15b detected banding of the printed image at the edges of the image

16 identified and marked white/black lines

17 detected enlarged sections in swaths of the printed image

18 image signal generated from detected bands of printed image with text content

18a image signal generated from a detected swath of the printed image at an image edge

18b image signals generated by detected swaths of the printed image at image edges

19 minimum detection height of white/black lines in image signal generated

20 white/black line areas marked

21 candidate image composed of stripes