阅读说明：本技术 用于陶瓷载体，特别是瓷砖的图形适配方法和系统 (Method and system for adapting a pattern for a ceramic support, in particular a ceramic tile ) 是由 马泰奥·鲁比亚尼 达维德·邦维奇尼 佛兰科·斯特凡尼 于 2018-11-22 设计创作，主要内容包括：本发明描述一种用于陶瓷载体(PC_i)的图形适配方法,特别是瓷砖,适用于接收印刷品,该方法包括以下步骤：(f0)加载原始图像文件(F_PRINT)；(f1)从原始图像文件(F_PRINT)开始在样本陶瓷载体(PCS_j)上印刷多个图像；(f2)获取印刷在样本陶瓷载体(PCS_j)上的多个图像中的样本图像(IM_PCS_j)；(f3)选择表示样本图像(IM_PCS_j)的样本点(P2),并选择原始图像文件(F_PRINT)中存在的原始点(P1)；(f4.1)依据在表示样本图像(IM_PCS_j)的样本点(P2)与原始图像文件(F_PRINT)中存在的原始点(P1)之间的点(M1)进行匹配；(f5)从依据点(M1)进行的匹配开始修改原始图像文件(F_PRINT),确定适配图像文件(F_ADAPT),从而使原始图像文件(F_PRINT)的图形适配样本图像(IM_PCS_j)的图形。本发明进一步描述用于实现所描述的方法的图形适配系统(100)。(The invention describes a method for graphic adaptation of a ceramic carrier (PC _ i), in particular a tile, suitable for receiving a print, comprising the following steps: (f0) loading an original image file (F _ PRINT); (f1) printing a plurality of images on a sample ceramic carrier (PCS _ j) starting from an original image file (F _ PRINT); (f2) acquiring a sample image (IM _ PCS _ j) of a plurality of images printed on a sample ceramic carrier (PCS _ j); (f3) selecting a sample point (P2) representing the sample image (IM _ PCS _ j) and selecting an original point (P1) present in the original image file (F _ PRINT); (F4.1) matching by means of a point (M1) between a sample point (P2) representing the sample image (IM _ PCS _ j) and an original point (P1) present in the original image file (F _ PRINT); (f5) the original image file (F _ PRINT) is modified starting from the matching made according to the point (M1), and an adaptation image file (F _ ADAPT) is determined so that the pattern of the original image file (F _ PRINT) ADAPTs to the pattern of the sample image (IM _ PCS _ j). The invention further describes a graphics adaptation system (100) for implementing the described method.)

1. A method for graphic adaptation of a sample ceramic carrier (PCS j) adapted to receive printed matter, wherein said method comprises the steps of:

- (F0) loading an original image file (F _ PRINT) representing an image (I _ PRINT) for printing, said image (I _ PRINT) comprising graphics to be printed on a plurality of ceramic supports (PC _ I),

wherein the original image file (F _ PRINT) is graphically defined in terms of first identification graphic features (GRAF _1) comprising one or more of:

-the original value of the color rendering (RC _ 1);

-a resolution (RS _ 1);

-a reference perspective view (PR _ 1);

- (F1) printing a plurality of images on the sample ceramic carrier (PCS _ j) starting from the original image file (F PRINT);

- (f2) acquiring, by means of an image acquisition device (20), a sample image (IM _ PCS _ j) of said plurality of images printed on said sample ceramic carrier (PCS _ j), wherein said sample image (IM _ PCS _ j) is graphically defined according to a second identification graphic feature (GRAF _2) according to said image acquisition device (20);

wherein the second recognition pattern feature (GRAF _2) comprises one or more of:

-sample values of the color rendering (RC _ 2);

-sample resolution RS _ 2;

-a second reference perspective PR _ 2;

- (F3) selecting a sample point (P2) representative of said sample image (IM _ PCS _ j) and selecting an original point (P1) present in said original image file (F _ PRINT),

(F4.1) matching the point (M1) between the sample point (P2) representing the sample image (IM _ PCS _ j) and the original point (P1) present in the original image file (F _ PRINT);

- (F5) modifying the original image file (F _ PRINT) starting from the matching made according to the point (M1), determining an adapted image file (F _ ADAPT) so as to ADAPT the graphics of the original image file (F _ PRINT) to the graphics of the sample image (IM _ PCS _ j).

2. The graphics adaptation method according to claim 1, wherein the step (f4.1) of matching by points (M1) comprises: finding a double unique match between the sample point (P2) representing the sample image (IM _ PCS _ j) and the original point (P1) present in the original image file F _ PRINT.

3. The graphics adaptation method according to claim 1 or 2, wherein the step (f4.1) of matching by points (M1) comprises: finding a spatial and color match as a function of the point between the sample point P2 representing the sample image IM _ PCS _ j and the original point P1 present in the original image file F _ PRINT.

4. A graphics adaptation method according to claim 1, 2 or 3, further comprising the steps of:

- (F4.2) extracting, in said original image file (F _ PRINT), at least one original image (IM _ F _ PRINT) corresponding to said sample image (IM _ PCS _ j), according to said matching according to the point (M1) sought.

5. The graphics adaptation method according to claim 4, further comprising the steps of:

- (F4.3) graphically aligning the sample image (IM _ PCS _ j) and the at least one original image (IM _ F _ PRINT) according to one or more of a respective resolution (RS _ 2; RS _1) and a respective perspective view (PR _2) (PR _ 1).

6. The graphics adaptation method according to claim 5, further comprising the steps of:

- (f.4.4) calculating, starting from the matching of the point (M1) between the sample point (P2) and the origin point (P1), a converted value (V1i → V2i) between the sample value (RC 2) of the color rendering and the original value (RC 1) of the color rendering, with reference to the sample image (IM _ PCS _ j) and the at least one original image (IM _ F _ PRINT) aligned and having the same resolution, thus adapting, depending on the color rendering, the graphics of the original image file (F _ PRINT) to the graphics of the sample image (IM _ PCS _ j).

7. The graphics adaptation method according to claim 6, wherein the step (f4.4) of calculating a conversion value (V1i → V2i) comprises the steps of:

- (f4.4.1) identifying a first sliding window (SW _1) in the original image (IM _ F _ PRINT) and a second sliding window (SW _2) in the sample image (IM _ PCS _ j), wherein the sliding windows (SW _1, SW _2) slide in the respective images in a similar manner;

- (f4.4.2) calculating in each of said second sliding window (SW _2) and said first sliding window (SW _1), respectively, a sample representation value (V2i) of the sample color Space (SCC) of said sample image (IM _ PCS _ j) and an original representation value (V1i) of the original color Space (SCO) of said original image (IM _ F _ PRINT);

- (f4.4.3) mapping the sample representation value (V2i) and the original representation value (V1i) to effect a conversion from the original value of color rendering (RC _1) to the sample value of color rendering (RC _ 2).

8. The graphics adaptation method according to claim 7, wherein the step (f4.4) of calculating a conversion value (V1 → V2i) further comprises the steps of:

(f4.4.4) identifying a color conversion (TR _ OK) between the sample representation value (V2i) and the original representation value (V1i), the color conversion (TR _ OK) representing a complete mapping (M _ OK) between the sample color Space (SCC) and the original color Space (SCO).

9. The graphics adaptation method according to claim 8, wherein the full mapping (M _ OK) is obtained by one of:

-a linear regression technique (RL) in which a matrix of coefficients K (1 … n) is calculated, the matrix being a solution of a system of linear equations formed by the samples (SCO/SCC) obtained in a mapping step (f.4.4.3) from the sample representation values (V2i) to the original representation values (V1i),

-an Artificial Neural Network (ANN) technique, wherein the matching between the original color Space (SCO) and the sample color Space (SCC) obtained in a previous mapping step (f.4.4.3) of the sample representation values (V2i) and the original representation values (V1i) is used for training an artificial neural network.

10. A graphics adaptation method according to any of the preceding claims, wherein the sample representation values (V1i, V2i) are values of one or more RGB triplets.

11. The graphics adaptation method according to claim 10, wherein the complete mapping (M _ OK) is obtained by means of a look-up table (LUT), wherein each triplet (RGB) of the original color Space (SCO) is associated with one of the mean, mode, median of the samples (RGB) mapped by the sample color Space (SCC) obtained in the previous step (f.4.4.3) of the mapping from the sample representation value (V2i) to the original representation value (V1 i).

12. The graphics adaptation method according to claim 11, wherein the full mapping (M _ OK) further comprises the steps of:

-verifying said mapping of said look-up table (LUT); if the look-up table (LUT) does not completely map the RGB triplet of the original color Space (SCO):

-obtaining a further sample image (IM _ PCS _ J) for mapping a not yet mapped portion of said original color Space (SCO);

-using 1: 1-matching parameters that are the linear regression system (RL) or alternatively train an Artificial Neural Network (ANN) to estimate missing RGB triplets.

13. The graphics adaptation method according to any of the claims 8 to 12, wherein the modification step F5 of the original image file (F PRINT) comprises: -modifying said original image file F _ PRINT according to said calculated color conversion (TR _ OK), determining said adapted image file (F _ ADAPT).

14. The graphics adaptation method according to any of claims 5 to 13, wherein said step (F4.3) of graphics alignment is achieved by changing the original resolution (RS _1) of the original image (IM _ F _ PRINT) to a value of the sample resolution (RS _2) of the sample image (IM _ PCS _ j).

15. The method according to any of the preceding claims, wherein the step (f1) of printing a plurality of images on the sample ceramic carrier (PCS _ i) is realized in a printing pattern between a random pattern and a fixed-plane pattern.

16. The graphic adaptation method according to claim 15, wherein the printing step (f1) comprises:

-printing, in said random mode, a limited plurality of images on said sample ceramic carrier (PCS _ j) starting from said original image file (F _ PRINT);

and the obtaining step (f2) includes:

-randomly acquiring the sample image (IM _ PCS _ j) of the plurality of images printed on the sample ceramic carrier (PCS _ j) printed in a random pattern.

17. The graphics adaptation method according to claim 16, wherein said step (F4.1) of matching according to points (M1) is realized by printing in a random pattern for finding said positions where said sample images (IM _ PCS _ j) are located in said original image file (F PRINT).

18. The graphic adaptation method according to claim 15, wherein the printing step (f1) comprises:

-printing, in said fixed-plane mode, a limited plurality of images on said sample ceramic carrier (PCS _ j) starting from said original image file (F _ PRINT);

and the obtaining step (f2) includes:

-acquiring said one sample image (IM _ PCS _ j) of said plurality of images printed on said sample ceramic carrier (PCS _ j) printed in a fixed-plane mode.

19. A method of graphics adaptation, as claimed in any preceding claim, wherein the method is a computer-implemented method.

20. A method for quality control of a ceramic carrier (PC _ i), comprising the steps of:

-generating an adapted image file (F ADAPT) starting from the original image file (F PRINT) according to any one of claims 1 to 19;

-receiving a printed ceramic carrier (PC _ i);

-identifying a portion (P _ F _ ADAPT) of the adapted image file (F _ ADAPT) corresponding to the graphics reproduced on the printed ceramic carrier (PC _ i);

-detecting a graphic difference (Dd) between the graphic reproduced on the printed ceramic carrier (PC i) and the portion (P _ F _ ADAPT) of the adapted image file (F _ ADAPT);

-classifying (Ci) the printed ceramic carriers (PC _ i) according to the detected differences (Dd), thereby enabling quality control of the printed ceramic carriers (PC _ i).

21. A graphic adaptation system (100) for a sample ceramic carrier (PCS j) adapted to receive a printed product, wherein the system comprises:

-a printer (1) configured to PRINT a limited plurality of images on said sample ceramic carrier (PCS _ j) starting from an original image file (F _ PRINT) defined graphically in terms of first identification graphic features (GRAF _1) comprising one or more of the following;

-the original value of the color rendering (RC _ 1);

-a resolution (RS _ 1);

-a reference perspective view (PR _ 1);

-image acquisition means (20) configured to acquire a sample image (IM _ PCS _ j) of said plurality of images printed on said sample ceramic carrier (PCS j),

wherein the sample image (IM _ PCS _ j) is graphically defined according to a second recognition pattern feature (GRAF _2) according to the image acquisition means (20) and comprises one or more of:

-sample values of the color rendering (RC _ 2);

-sample resolution RS _ 2;

-a second reference perspective PR _ 2;

a first treatment station (10) comprising:

-a loading module (101) configured (F0) to load said original image file (F _ PRINT) representing an image (I _ PRINT) for printing, said image (I _ PRINT) comprising graphics to be printed on said plurality of said ceramic carriers (PC _ I);

a first selection module (102) configured for (F3) selecting a sample point (P2) representing the sample image (IM _ PCS _ j) and selecting an original point (P1) present in the original image file (F _ PRINT),

-a comparison module (103) configured for (F4.1) matching (F4.1) a point (M1) between said sample point (P2) and said original point (P1) present in said original image file (F _ PRINT);

-an adaptation module (107) configured for (F5) modifying the original image file (F _ PRINT) starting from the matching according to point (M1), determining an adapted image file (F _ ADAPT) such that the pattern of the original image file (F _ PRINT) is adapted to the pattern of the sample image (IM _ PCS _ j).

22. The graphics adaptation system according to claim 21, wherein one comparison module (103) is configured for (F4.1) making said match by finding a bi-unique match between the sample point (P2) representing the sample image (IM _ PCS _ j) and the origin point (P1) present in the original image file F _ PRINT, the point (M1) between the sample point (P2) and the origin point (P1) present in the original image file (F _ PRINT).

23. Graphics adaptation system according to claim 21 or 22, wherein the comparison module (103) is configured for matching per point (M1) by finding a spatial and color match depending on the point between the sample point P2 representing the sample image IM _ PCS _ j and the original point P1 present in the original image file F _ PRINT.

24. System according to claim 21, wherein said first processing station (10) comprises an extraction module (104), said extraction module (104) being configured for (F4.2) extracting, in said original image file (F _ PRINT), at least one original image (IM _ F _ PRINT) corresponding to said sample image (IM _ PCS _ j) according to said matching made as a function of the detected point (M1).

25. The system of claim 24, wherein the first processing station (10) comprises a graphic alignment module (105), the graphic alignment module (105) being configured for (F4.3) graphically aligning the sample image (IM _ PCS _ j) and the at least one original image (IM _ F _ PRINT) according to one or more of a respective resolution (RS _ 1; RS _2) and a respective perspective view (PR _2) (PR _ 1).

26. System according to claim 25, wherein the first processing station (10) comprises a calculation module (106), the calculation module (106) being configured for (f.4.4.4) calculating, starting from the matching of a point (M1) between the sample point (P2) and the origin point (P1), a conversion value (V1i → V2i) between the color-rendered sample value (RC _2) and the color-rendered original value (RC _1), adapting, according to the color rendering, the graphics of the original image file (F _ PRINT) to the graphics of the printed ceramic carrier (PC _ i), with reference to the sample image (IM _ PCS _ j) and to the at least one original image (IM _ F _ PRINT) aligned and having the same resolution.

27. The system of claim 26, wherein the first processing station (10) further comprises:

-a first identification module (106a) configured for (f4.4.1) identifying a first sliding window (SW _1) in the original image (IM _ F _ PRINT) and a second sliding window (SW _2) in the sample image (IM _ PCS _ j), wherein the sliding windows (SW _1, SW _2) slide in the respective images in a similar manner;

-a second calculation module (106b) configured for (f4.4.2) calculating in each of said second sliding window (SW _2) and said first sliding window (SW _1), respectively, a sample representation value (V2i) of a sample color Space (SCC) of said sample image (IM _ PCS _ j) and an original representation value (V1i) of an original color Space (SCO) of said original image (IM _ F _ PRINT);

-a mapping module (106c) configured for (f4.4.3) mapping the sample representation values (V2i) and the original representation values (V1i) to enable a conversion from the color rendered original values (RC _1) to the color rendered sample values (RC _ 2).

28. The system according to claim 27, wherein the first processing station (10) further comprises:

-a second identifying module 106d configured for (f4.4.4) identifying a color conversion (TR _ OK) between the sample representation value (V2i) and the original representation value (V1i), the color conversion (TR _ OK) representing a complete mapping (M _ OK) between the sample color Space (SCC) and the original color Space (SCO).

29. The system as recited in claim 28, wherein the second identification module (106d) includes one or more of:

-a first conversion submodule (106d1) configured to identify the color conversion (TR OK) by means of a linear regression technique;

-a second conversion submodule (106d2) configured to identify the color conversion (TR OK) by means of artificial neural network techniques.

30. The system of claim 28 or 29, wherein the second identification module (106d) comprises:

a third conversion submodule (106d3) configured to identify the color conversion (TR _ OK) by means of a technique utilizing a look-up table (LUT), wherein the second identification module (106d) is configured to:

-verifying the mapping of the look-up table (LUT);

-if said look-up table (LUT) is not able to fully map said RGB triplet of said original color Space (SCO):

-obtaining a further sample image (IM _ PCS _ J) to map a portion of said original color Space (SCO) that has not been mapped;

-using 1: 1-matching parameters that are the linear regression system (RL) or alternatively training an Artificial Neural Network (ANN) to estimate missing triples.

31. The system according to any one of claims 21 to 30, wherein the printer (1) is configured to print a plurality of images on the ceramic support in one of the following modes:

-random;

-a fixation surface.

32. A quality control system (200) for a ceramic carrier (PC _ i), comprising:

the graphics adaptation system (100) according to any of the claims 21 to 31, the graphics adaptation system (100) being configured to generate an adapted image file (F ADAPT) starting from an original image file (F PRINT);

a receiving device (400) for printing a ceramic carrier (PC _ i);

-a sorting system (300) coupled to the graphics adaptation system (100) and to the receiving device (400) of printed ceramic carriers (PC _ i), the sorting system (300) comprising a second processing station (110), the second processing station (110) comprising:

-a study module (111) configured to identify a portion (P _ F _ ADAPT) of the adapted image file (F _ ADAPT) corresponding to the graphics reproduced on the printed ceramic carrier (PC _ i);

-a detection module (112) configured to identify a graphic difference (Dd) between the graphic reproduced on the printed ceramic carrier (PC _ i) and the portion (P _ F _ ADAPT) of the adapted image file (F _ ADAPT);

-a sorting module (113) configured to sort (Ci) the printed ceramic carriers (PC _ i) according to the detected differences (Dd), thereby enabling quality control of the printed ceramic carriers (PC _ i).

33. A computer program configured to perform one or more steps of the method of claim 19 when run on a computer.

34. A production system for producing ceramic supports, comprising a quality control system for ceramic supports (PC _ i) according to claim 32, interposed between an image printing system on a ceramic support (PC _ i) and a kiln for firing the ceramic support (PC _ i).

Technical Field

The invention relates to a method and a system for graphic adaptation of ceramic supports, in particular tiles.

More specifically, the invention relates to a method and a system for graphic adaptation for ceramic supports, in particular tiles, suitable for receiving prints performed starting from an original graphic file.

Background

In recent years, the production of ceramics has undergone a revolution in relation to printing technology. From the traditional methods with ceramic decorative features, up to the end of the last century, i.e. roto-printing by means of silicone cylinders, the methods have evolved into non-contact printing based on inkjet printers, in which a suitable graphic file, processed and divided into one colour channel, is used as a source for printing on tiles of various sizes by means of a printer provided with print heads arranged along a crossbar.

Generally, there are two different printing modes:

fixed-side printing: the graphic studio prepares a discrete, even high, mass of fixing surfaces, which are printed in a predetermined sequence on the ceramic carrier;

random printing, starting from a plate-like larger pattern. The algorithm randomly calculates a print area as large as the size of the tile panel to be printed, without prior preparation (in real time), and can cut it out at the time of printing and print it onto the ceramic carrier. In this case, the number of faces is very large.

In both printing modes, the ceramic products obtained can be inspected directly after decoration, at the outlet of the digital printer or after firing, in preparation for boxing.

In general, in order to be able to successfully inspect ceramic products, it is necessary to have a comparative model.

If printing is performed using a fixed surface, it is necessary to acquire a defect-free print by a television camera and save it in an archive, by a step called "learning" step, in order to remember all the surfaces.

At the end of the learning step, each tile of the work acquired in the same way can be compared with the images stored in the learning step and then, once the homologous model is determined and passed through the image processing algorithm, any differences, i.e. potential production defects, are highlighted.

Therefore, in the fixed-surface printing provided with the tile image, it is necessary to find out one most like it among all possible sample images to be used as a reference image (template) for printing.

If the printing is random (random), there may be a large number of comparison surfaces, and therefore it is not possible to obtain all the comparison surfaces by means of a television camera.

Therefore, taking the material directly does not make it possible to obtain an effective comparative sample.

Therefore, the sample must be obtained by extrapolating the sample from the original graphic file representing the printing source.

In the case of random printing, the acquisition of the sample image determines a number of problems, one of the most important of which is the difference in color rendering between the source graphic file and the acquired image, and the final ceramic product may not match the color of the source graphic file.

"color rendering" is known to refer to the color presented by each point of an image, expressed in a known colorimetric system (e.g., RGB, HSL, etc.).

Other problems encountered are:

the printing resolution of the ceramic product may vary with respect to the acquisition resolution of the television camera;

the perspective distortion that may be caused by the acquisition of the television camera;

possible effects of incomplete equalization of the images due to the geometry of the acquisition system in combination with the geometry of the illumination system;

possible rotation of the part with respect to the figure due to incorrect alignment between the tile and the printing machine;

imperfect matching due to dots between the graphic and the product (blurring effect due to interaction between ink and ceramic support, for example due to spreading and absorption of droplets).

In general, there may be a problem of inconsistency between the identifying graphic features of the original graphic file and the identifying graphic features of the image taken from the sample ceramic carrier.

Therefore, it is impossible to previously evaluate the correctness (quality control) thereof using the original graphic file for printing without processing.

Furthermore, in the solutions of the prior art, any defects caused by digital printing, such as dark bands, stripes, etc., can be found only on the ceramic product, i.e. only after firing and at the exit of the kiln.

It can be understood how defective prints imply a specific risk of rejection/reclassification of products that have been fired and are ready for boxing.

It is an object of the present invention to provide a method and/or a graphic adaptation system for a ceramic carrier adapted to receive printed products that overcomes the drawbacks of the prior art.

Another object of the invention is to guarantee a colour range match between the original graphic file and the acquired ceramic support.

It is also an object of the present invention to provide a method and/or a quality control system for a ceramic carrier adapted to receive printed matter that overcomes the drawbacks of the prior art.

Disclosure of Invention

In a first aspect, the present invention describes a method for graphic adaptation of a sample ceramic carrier suitable for receiving printed matter, wherein the method comprises the steps of:

-loading an original image file representing an image to be printed, the image comprising graphics to be printed on a plurality of ceramic supports,

wherein the original image file is graphically defined in terms of a first identifying graphical feature comprising a color rendering original value;

-printing a plurality of images on the sample ceramic carrier starting from the original image file;

-acquiring a sample image of the plurality of images printed on a sample ceramic carrier, wherein the sample image is graphically defined in terms of second identifying graphical features comprising color-rendered sample values;

-selecting sample points representing the sample image and selecting original points present in an original image file,

-matching in dependence of points between the sample points representing the sample image and original points present in the original image file;

-modifying said original image file starting from said matching by points, determining an adapted image file, such that the image of the original image file is adapted to the image of the sample image.

Preferably, the step of matching in terms of points comprises finding a bi-unique match between a sample point representing the sample image and an original point present in the original image file.

Preferably, the step of point-wise matching envisages seeking a dual unique spatial and colour match, in terms of points, between the sample points representing the sample image and the original points present in the original image file.

Preferably, the method comprises the steps of:

-extracting at least one original image corresponding to said sample image in said original image file, according to said matching made as a function of the points sought.

Preferably, the method comprises the steps of:

-graphically aligning the sample image and the at least one original image according to one or more of the respective resolutions and the respective perspectives.

Preferably, the method comprises the steps of:

-calculating, with reference to the sample image and the at least one original image aligned and having the same resolution, conversion values between the sample values of the color rendering and the original values of the color rendering starting from a matching of points between the sample points and the original points, thus adapting, in terms of color rendering, the graphics of the original image file to the graphics of the sample image.

Preferably, the step of calculating the conversion value includes the steps of:

-identifying a first sliding window in the original image and a second sliding window in the sample image, wherein the sliding windows slide in a similar manner in the respective images;

-calculating in each of said second sliding windows and said first sliding windows, respectively, a sample representation value of a sample color space of said sample image and an original representation value of an original color space of said original image;

-mapping the sample representation values and the original representation values to enable a conversion from the original values of a color rendering to the sample values of a color rendering.

Preferably, the step of calculating the conversion value further comprises the steps of:

-identifying a color conversion between the sample representation values and the original representation values, the color conversion representing a complete mapping between the sample color space and the original color space.

Preferably, the complete mapping is obtained by one of:

-a linear regression technique, wherein a matrix of coefficients is calculated, which matrix is a solution of a system of linear equations formed by the samples obtained in the mapping step from said sample representation values to said original representation values.

-an artificial neural network technique, wherein a match between the original color space and the sample color space obtained from the previous mapping step of the sample representation values and the original representation values is used for training the artificial neural network.

Preferably, the sample representation values are values of one or more RGB triplets.

Preferably, the complete mapping is obtained by a look-up table, wherein each triplet of the original color space is associated with one of the mean, mode, median of the samples mapped by the sample color space obtained in the previous mapping step in the mapping from the sample representation values to the original representation values.

Preferably, the complete mapping further comprises the steps of:

-verifying the mapping of the look-up table; if the "look-up table" does not fully map the RGB triplet of the original color space:

-acquiring further sample images to map a portion of the original color space that has not been mapped;

-using 1: 1-matching parameters that estimate missing RGB triplets as a linear regression system or alternatively train an artificial neural network.

Preferably, said modifying step f5 of said original image file comprises: modifying the original image file in dependence on said calculated color conversion, thereby determining said adapted image file.

Preferably, the pattern alignment step is implemented by changing the original resolution of the original image to a value of the sample resolution of the sample image.

Preferably, said step of printing a plurality of images on the sample ceramic support is carried out in a printing mode between a random mode and a fixed-plane mode.

Preferably, said step of printing a plurality of images on a ceramic support comprises:

-printing a limited plurality of images on said sample ceramic carrier in a random pattern starting from said original image file;

and the acquiring step comprises:

-randomly acquiring the sample images of the plurality of sample ceramic carriers printed in a random pattern.

Preferably, said step of matching by site is carried out by printing in a random pattern to find the location where the sample image is located in the original image file.

Preferably, the printing step comprises:

-printing a limited plurality of images on the sample ceramic carrier starting from the original image file in a fixed-plane mode;

and the acquiring step comprises:

-acquiring said one sample image of said plurality of images printed on said sample ceramic carrier printed in a fixed-plane pattern.

In a second aspect, the present invention discloses a quality control method for a ceramic carrier, the method comprising the steps of:

-generating a suitable image file starting from the original image file, as described in the first aspect of the invention;

-receiving a printed ceramic carrier;

-identifying a portion of the adapted image file corresponding to the graphic reproduced on the printed ceramic support;

-detecting a graphic difference between the graphic reproduced on the printed ceramic support and the portion of the adapted image file;

-classifying the printed ceramic carriers into classes according to the detected differences, thereby enabling quality control of the printed ceramic carriers.

In a third aspect of the invention, the method of the first aspect of the invention is a computer-implemented method.

In a fourth aspect of the invention, the method of the aspects of the invention is a computer-implemented method.

In a fifth aspect, the present disclosure describes a graphic adaptation system for a sample ceramic carrier, the system being adapted to receive printed matter, wherein the system comprises:

-a printer configured to print a limited plurality of images on said sample ceramic carrier starting from an original image file, the original image file being graphically defined in terms of a first identifying graphical feature comprising original values of the color rendering;

-an image acquisition device configured to acquire a sample image of the plurality of images printed on the sample ceramic carrier,

wherein the sample image is graphically defined in accordance with the image capture device in dependence upon a second identifying graphical feature and comprises sample values of a color rendering;

a first processing station comprising:

a loading module configured to load the original image file representing an image to be printed, the image including graphics to be printed on the plurality of the ceramic carriers;

a first selection module configured to select sample points representing the sample image and to select original points present in an original image file;

a comparison module configured to match points between the sample points and original points present in an original image file;

an adaptation module configured to modify the original image file starting from the matching according to the point, determine an adapted image file, thereby adapting the pattern of the original image file to the pattern of the sample image.

Preferably, the comparison module is configured to perform the matching in dependence on a point between the sample point representing a sample image and the original point present in an original image file by finding a double unique match between the sample point and the original point present in the original image file.

Preferably, the comparison module is configured to find the match in dependence on a point between a sample point representing the sample image and an original point present in the original image file between searches for spatial and colour matches.

Preferably, the first processing station comprises an extraction module configured to extract, in dependence on the detected points, in the original image file, at least one original image corresponding to the sample image, according to the matching.

Preferably, the first processing station comprises a graphical alignment module configured to graphically align the sample image and the at least one raw image according to one or more of the respective resolutions and the respective perspectives.

Preferably, the first processing station comprises a calculation module configured to calculate, with reference to the sample image and the at least one original image aligned and having the same resolution, conversion values between the sample values of the color rendering and the original values of the color rendering starting from a matching according to points between the sample points and the original points, adapting, according to the color rendering, the graphics of the original image file to the graphics of the printed ceramic support.

Preferably, the processing station further comprises:

a first identification module configured to identify a first sliding window in the original image and a second sliding window in the sample image, wherein the sliding windows slide in the respective images in a similar manner;

a second calculation module configured to calculate a sample representation value of a sample color space of the sample image and an original representation value of an original color space of the original image in each of the second sliding window and the first sliding window, respectively;

a mapping module configured to map the sample representation values and the original representation values to enable conversion from the color rendered original values to the color rendered sample values.

Preferably, the processing station further comprises:

a second identification module configured to identify a color conversion between the sample representation value and the original representation value, the color conversion representing a complete mapping between the sample color space and the original color space.

Preferably, the second identification module comprises one or more of:

a first conversion submodule configured to identify a color conversion by means of a linear regression technique;

a second conversion submodule configured to identify a color conversion by means of an artificial neural network technique.

Preferably, the second identification module comprises:

a third conversion submodule configured to identify a color conversion by means of a technique utilizing a look-up table, wherein the second identification module is configured to:

verifying the mapping of the lookup table;

if the "look-up table" does not completely map the triplet of the original color space:

acquiring other sample images to map a portion of the original color space that has not been mapped;

using 1 of the lookup table: 1-matching as parameters of a linear regression system or alternatively parameters of a trained artificial neural network to estimate missing RGB triplets.

Preferably, the printer is configured for printing in a random or fixed-plane pattern.

In a sixth aspect, the present invention discloses a quality control system for a ceramic support, the system comprising:

a graphics adaptation system as described in the fifth aspect, configured to generate an adapted image file starting from an original image file;

a receiving device for receiving a printed ceramic carrier;

a sorting system coupled to the graphic adaptation system and the receiving device for receiving the printed ceramic carriers, the sorting system comprising a second processing station comprising:

a research module configured to identify a portion of the adapted image file corresponding to a graphic rendered on the printed ceramic carrier;

a detection module configured to identify a graphic difference between the graphic rendered on the printed ceramic carrier and the portion of the adapted image file;

-a classification module configured to classify the printed ceramic carriers into classes according to the detected differences, thereby enabling quality control of the printed ceramic carriers.

In a seventh aspect, the invention features a production system for producing a ceramic support, the system including a quality control system for a ceramic support, according to the sixth aspect, interposed between an image printing system on a ceramic support and a kiln for firing the ceramic support.

In an eighth aspect, the invention features a program for a calculator configured in use to perform the method of the first aspect of the invention.

In a ninth aspect, the invention describes a program for a calculator configured in use to perform the method of the second aspect of the invention.

The invention achieves the main technical effect of ensuring the color range matching between the original graphic file and the printing ceramic carrier.

The technical effect achieved is that possible production defects are effectively extracted, and false identification is reduced.

The technical effects mentioned, the advantages cited and other technical effects/advantages of the invention will emerge in further detail from the following description of an example of embodiment provided by way of approximate and non-limiting example with reference to the accompanying drawings.

Drawings

Fig. 1 shows a graphic adaptation system of the invention in a schematic view.

Fig. 1a shows a quality control system comprising the graphic adaptation system of fig. 1 in a schematic view.

Fig. 2 shows in a schematic view a first group of functional modules and/or memory modules comprised in a processing station according to the invention.

Fig. 3 shows a second group of functional modules and/or memory modules comprised in the processing station according to the invention in a schematic view.

Fig. 4 shows a graphical comparison between an acquired image and an original image file according to the present invention.

Fig. 5 shows the selection of points representing an acquired image and the selection of images in an original file starting from the representative points according to the invention.

Fig. 6A shows a sliding window in an original file.

Fig. 6B shows a sliding window in the acquired file.

Detailed Description

The graphic adaptation system 100 for sample ceramic carriers according to the present invention generally comprises: a printer configured to print a limited plurality of sample ceramic carriers starting from an original image file; a color or black-and-white image obtaining device configured to obtain a sample image of each sample ceramic product; and a processing station configured to graphically adapt the original image file to the acquired image in terms of color rendering, resolution and perspective.

Referring to fig. 1 and 2, a first processing station 10 is shown.

It should generally be noted that in the present context and in the following claims, the first processing station 10 is presented subdivided into different functional modules (memory modules or operating modules) for the sole purpose of clearly and thoroughly describing its functionality.

Such a first processing station 10 may comprise a single electronic device suitably programmed to perform the described functions, and the different modules may correspond to hardware entities and/or routine software being part of the programmed device.

Alternatively or additionally, the functions may be performed by a plurality of electronic devices, on which the above-described functional modules may be distributed.

The first processing station 10 may also execute instructions contained in the memory module using one or more processors.

The functional modules may also be distributed over different local or remote computers, depending on the architecture of the network in which they reside.

In particular, in fig. 2, the first processing station 10 comprises a loading module 101 configured to perform a loading step F0 (fig. 2) of the original image file F _ PRINT.

The file loading step envisages loading an original image file F _ PRINT representing an image I _ PRINT (fig. 1) for printing, the image I _ PRINT comprising graphics to be printed on a plurality of ceramic supports PC _ I.

In other words, the first processing station 10 is configured for loading an original image file F _ PRINT representing an image I _ PRINT (fig. 1) for printing, the image I _ PRINT comprising graphics to be printed on a plurality of ceramic supports PC _ I.

In one embodiment of the invention, particularly for fixed-side printing, the combination of portions of the pattern coincides with the image for printing, I _ PRINT.

In another embodiment of the invention, particularly for random printing, portions of the graphic represent random samples of the image used to PRINT the I _ PRINT.

The original image file F _ PRINT is graphically defined according to the first recognition pattern feature GRAF _ 1.

Preferably, the first recognition pattern feature GRAF _1 includes one or more of:

-the original value of the color rendering RC _ 1;

-a resolution RS _ 1;

reference is made to the perspective view PR _ 1.

Referring to fig. 1, a printer 1 is shown configured for printing on a ceramic support.

In particular, printing may be performed in a random or fixed-plane pattern.

The invention envisages a printing step F1 (fig. 1) of a limited number of sample ceramic carriers PCS _ j starting from the original image file F PRINT.

According to the prior art in the field of printing on ceramic carriers, printing of a plurality of sample ceramic carriers PCS _1 was performed in one of a random pattern and a fixed-plane pattern.

It will be understood that step F1 is a production step for producing a limited number of sample ceramic carriers PCS _ j with PRINTs carrying the image for printing I _ PRINT contained in the original image file F _ PRINT.

The term "printed on a ceramic support" in technical terms has the meaning specified in the preceding paragraph and, in the course of the description, reference will be made to this interpretation unless otherwise stated.

With reference to the described prior art (page 2), it is evident that a plurality of ceramic carriers are produced on a production line and that, after this, an image is printed onto the plurality of ceramic carriers.

Referring to fig. 1, there is further shown an image acquisition device 20 configured to (f2) acquire sample images IM _ PCS _ j (fig. 3) of a plurality of sample ceramic carriers PCS _ j printed by the printer 1.

In one embodiment relating to a random printing mode, the image acquisition means 20 are configured to acquire sample images IM _ PCS _ j of a limited plurality of sample ceramic carriers PCS _ j printed by the printing press 1 in a random pattern (fig. 3).

In particular, the image acquisition means 20 comprise one or more colour or black and white television cameras provided for acquiring colour or black and white images, respectively.

The sample image IM _ PCS _ j is graphically defined according to the second recognition pattern feature GRAF _2 by the camera 20.

The second recognition pattern feature GRAF _2 includes one or more of the following:

-sample values of the color rendering RC _ 2;

-sample resolution RS _ 2;

a second reference perspective PR _ 2.

It should be understood that the first recognition pattern feature GRAF _1 is different from the second recognition pattern feature GRAF _2 at least in terms of color rendering, the former being related to the original image file and the latter being dependent on the "change" caused by the image acquisition means 20.

The invention also envisages identifying in each sample ceramic carrier image IM _ PCS _ j acquired a series of features characteristic of the pattern reproduced on the ceramic carrier.

According to the invention, with reference to fig. 2 and 5, a step f3 is therefore envisaged for selecting sample points (keypoints) P2 of the acquired ceramic carrier, wherein these points represent the sample images IM _ PCS _ j.

Each sample point P2 is characterized by a descriptor D2, and the descriptor D2 contains a plurality of feature elements representing the local graphic feature of each point, defined as "local features".

It is desirable to select a sufficient number of sample points so that the sample image can be well described.

The first processing station 10 comprises a first selection module 102 (fig. 2) configured to perform the steps described.

According to the invention, with reference to fig. 2 and 5, the point selection step F3 is also configured for selecting an original point P1 present in the original image file F _ PRINT.

Each origin point P1 is characterized by a descriptor D1, the descriptor D1 comprising a plurality of feature elements, representing the local features of each point, defined as "local features".

It is desirable to select a sufficient number of original points P1 to be able to describe the sample image well.

The first selection module 102 (fig. 2) is configured to also perform this selection step of origin P1.

According to the invention, a matching finding step f4.1 is also provided.

In other words, the first processing station 10 is configured to select sample points (keypoints) P2 of the acquired ceramic carrier, wherein these points represent the sample images IM _ PCS _ j, and are used to select the original point P1 present in the original image file F _ PRINT.

Referring to fig. 2 and 5, step F4.1 comprises matching as a function of the point M1 between the sample point P2 representing the sample image IM _ PCS _ j and the original point P1 present in the original image file F _ PRINT.

In other words, step F4.1 comprises finding a double unique match between the sample point P2 representing the sample image IM _ PCS _ j and the original point P1 present in the original image file F _ PRINT.

In particular, the match M1 is a spatial and color match in terms of points between the sample point P2 representing the sample image IM _ PCS _ j and the original point P1 present in the original image file F _ PRINT.

In particular, for printing in a random pattern, a matching is performed to find the position of the sample ceramic carrier PCS _ j in the original image file F _ PRINT.

The first processing station 10 comprises a comparison module 103 (fig. 2) configured to perform the steps described.

According to the invention, the original image file F _ PRINT is modified at this point so that it matches the sample image in terms of points.

Precisely, the invention comprises a modification step F5, which modification step F5 modifies the original image file F _ PRINT starting from a match according to point M1, determining an adapted image file F _ ADAPT, so as to ADAPT the pattern of the original image file F _ PRINT to the pattern of the sample image IM _ PCS _ j.

The adaptation module 107 in the first processing station 10 is configured to perform the described step f 5.

According to the invention, the step F5 of modifying the original image file F _ PRINT, after the step F4.1 of point-by-point matching, envisages a step F4.2 of image extraction for identifying, in the original image F _ PRINT, a region of interest ROI matching the sample image IM _ PCS _ j printed on the ceramic support.

In other words, the first processing station 10 is configured for matching as a function of the point M1 between the sample point P2 and the original point P1 present in the original image file F _ PRINT.

With reference to fig. 2 and 6A, 6B, step F4.2 envisages that, as a function of the detected point M1, at least one original image IM _ F _ PRINT is extracted in the original image file F _ PRINT corresponding to the sample image IM _ PCS _ j according to a match.

The first processing station 10 comprises an extraction module 104 (fig. 2) configured to perform the steps described.

However, the two images retain different recognition graphics characteristics GRAF _1, GRAF _2, in particular different resolutions, color renderings and perspectives.

In other words, the first processing station 10 is configured for extracting, from the original image file F _ PRINT corresponding to the sample image IM _ PCS _ j, at least one original image IM _ F _ PRINT according to a match, depending on the detected point M1.

According to the invention, a further step f4.3 of graphic realignment is also included.

Referring to fig. 2, step F4.3 envisages graphically aligning the sample image IM _ PCS _ j and the at least one original image IM _ F _ PRINT so that they are perfectly aligned.

The first processing station 10 includes a pattern realignment module 105 (fig. 2) configured to perform the described steps.

In other words, the first processing station 10 is configured for graphically aligning the sample image IM _ PCS _ j and the at least one original image IM _ F _ PRINT according to the respective resolution RS _2, RS _1 and/or the respective perspective PR _2, PR _1, so as to perfectly align them.

According to the present invention, the graphic alignment is achieved by utilizing the spatial relationship between the coordinates of the sample point P2 in the sample image IM _ PCS _ j and the coordinates of the original point P1 in the original image IM _ F _ PRINT.

In other words, the spatial relationship between the coordinates of the sample point P2 and the coordinates of the original point P1 allows to identify a perspective transformation or affine transformation (in matrix form) which allows the original image IM _ F _ PRINT to be aligned with the desired graphic portion of the sample image IM _ PCS _ j according to the respective resolution RS _1, RS _2 and/or the respective perspective PR _1, PR _ 2. In particular, the resolution is defined as a function of the resolutions RS _1, RS _2 and preferably of the quality control algorithm used.

According to the invention, step f4.4 also comprises a conversion of the color space for adapting the value of the color rendering RC _1 of the original file to the value of the color rendering RC _2 of the sample.

Referring to fig. 2, step f4.4 comprises: with reference to the sample image IM _ PCS _ j and the at least one original image IM _ F _ PRINT, which are aligned and have the same resolution, values of conversion V1i → V2i from the original value of the color rendering RC _1 to the sample value of the color rendering RC _2 are calculated from the matching of the point M1 between the sample point P2 and the original point P1.

The technical effect achieved is that the graphics of the original image file F _ PRINT is adapted to the graphics of the printed ceramic carrier PC _ i in terms of color rendering.

In other words, starting from two alignable images with the same resolution, the invention consists in suitably processing the original graphic in order to perform a color conversion and to make it as similar as possible to the color of the obtained ceramic support.

The conversion step F4.4 envisages the superimposed calculation, at the same resolution, of the conversion value V1i → V2i between the original color space SCO of the original image IM _ F _ PRINT and the sample color space SCC of the sample image IM _ PCS _ j.

The conversion value V1i → V2i defines a possible conversion between elements of the original color space SCO of the original image IM _ F _ PRINT and the sample color space SCC of the sample image IM _ PCS _ j.

In a preferred embodiment of the invention, the converted value V1i → V2i is the value of one or more RGB triplets.

The first processing station 10 comprises a first calculation module 106 configured to perform the steps described.

In other words, the first processing station 10 is configured to superimpose and calculate, at the same resolution, the conversion value V1i → V2i between the original color space SCO of the original image IM _ F _ PRINT and the sample color space SCC of the sample image IM _ PCS _ j.

As shown in fig. 6A, 6B, according to the present invention, in order to calculate the converted value V1i → V2i, the present invention includes:

- (f4.4.1) identifies a first sliding window SW _1 (fig. 6A) in the original image IM _ F _ PRINT and a second sliding window SW _2 in the sample image IM _ PCS _ j, wherein the windows SW _1, SW _2 slide in the respective images in a similar manner.

In other words, the invention envisages identifying two windows of settable size (for example from 1 × 1 to 11 × 11), one respectively sliding on the region of interest ROI of the original image IM _ F _ PRINT and the other sliding on the acquired sample image IM _ PCS _ J.

The first calculation module 106 of the first processing station 10 comprises a first identification module 106a (fig. 3) configured to perform the described step f 4.4.1.

In other words, the first processing station 10 is configured to recognize a first sliding window SW _1 in the original image IM _ F _ PRINT and a second sliding window SW _2 in the sample image IM _ PCS _ j (fig. 6B), wherein the windows SW _1, SW _2 slide in the respective images in a similar manner.

The invention then comprises a step (f4.4.2) of calculating in each of the two sliding windows determined a sample representation value V2i of the sample color space SCC of the sample image IM _ PCS _ j and an original representation value V1i of the original color space SCO of the original image IM _ F _ PRINT, respectively.

In particular, the invention comprises calculating one or more transformed values V1i → V2i expressed by RGB triplets for a sliding window SW _1 sliding over the region of interest ROI and for a similar sliding window SW _2 sliding over the sample image.

According to the invention, when calculating the converted value V1i → V2i, the sliding windows SW _1, SW _2 are in the same respective positions with respect to the respective images, preferably represented as values of RGB triplets.

The first calculation module 106 of the first processing station 10 comprises a second calculation module 106B (fig. 2B) configured to perform the described step f4.4.2.

In other words, the first processing station 10 is configured to calculate, in each of the two sliding windows determined, respectively a sample representative value V2i of the sample color space SCC of said sample image IM _ PCS _ j and an original representative value V1i of the original color space SCO of the original image IM _ F _ PRINT.

In this regard, the invention includes the step of mapping (f4.4.3) the sample representation value V2i and the original representation value V1i to allow conversion from the original value of the color rendering RC _1 to the value of the sample's color rendering RC _ 2.

In other words, the invention comprises creating a transfer function between each component of the color rendering RC _1 of the original graphic and a plurality of possible values obtained from the acquired image; this is acceptable, for example, because multiple values in the acquired image may correspond to the same average value of the sliding window SW _1 on the ROI.

The first calculation module 106 of the first processing station 10 comprises a mapping module 106c (fig. 3) configured to perform the described step f4.4.3.

The invention further comprises a step f4.4.4 of identifying a color conversion TR _ OK between the sample representation value V2i and the original representation value V1i, which color conversion TR _ OK represents a complete mapping M _ OK between said sample color space SCC and the original color space SCO.

The first calculation module 106 of the first processing station 10 comprises a second identification module 106d (fig. 3) configured to perform the described step f 4.4.4.

In other words, the first processing station 10 is configured for identifying a color conversion TR _ OK between the original representation value V1i and the sample representation value V2i, which color conversion TR _ OK represents a complete mapping M _ OK between the original color space SCO and the sample color space SCC.

At the end of step f.4.4.4, the RGB triplets of the original color space SCO may be associated with a plurality of RGB triplets of the sample color space SCC. Thus, a generalized transformation is computed that approximates this multiple mapping in the best possible way.

According to the invention, in order to obtain the complete mapping M _ OK, three different solutions are implemented starting from the techniques known in the literature:

linear regression RL: a matrix of coefficients K (1 … n) is calculated which is a solution to the system of linear equations formed by the SCO/SCC samples obtained in the previous mapping step f.4.4.3 which mapped the sample representation value V2i to the original representation value V1 i.

In particular, as already said, the sample representation values V1i, V2i are values of one or more RGB triplets.

To calculate the prediction of the transformed values of the RGB triplets of the original color space SCO, it is multiplied by a matrix of coefficients K.

The second identification module 106d comprises a first conversion submodule 106d1 configured to identify the color conversion TR OK by means of the described linear regression technique RL.

In other words, the first processing station 10 is configured to recognize the color conversion TR _ OK by means of the described linear regression technique.

Artificial neural network ANN: the match between the original color space SCO and the sample color space SCC obtained in the previous mapping step f4.4.3 from the sample representation value V2i and the original representation value V1i is used to train the artificial neural network.

In order to obtain a prediction for converting the RGB triplets of the original color space SCO into RGB triplets of the sample color space SCC, the RGB triplets of the original color space SCO are converted by a neural network. Depending on the structure of the network, the triplets undergo one or more transformations produced by the network training step and finally obtain the result of the transformation.

Unlike linear regression, the function is not necessarily linear, and thus the conversion from the original color space SCO to the sample color space SCC can be better described.

The second identification module 106d comprises a second conversion submodule 106d2 configured to identify the color conversion TR _ OK by means of the described artificial neural network technique.

In other words, the first processing station 10 is configured to identify the color conversion TR _ OK by means of the artificial neural network technique ANN described.

Look-up table LUT: each triplet RGB of the original color space SCO is associated with one of the mean, mode, median of the samples (RGB) mapped by the sample color space SCC obtained in the previous step f.4.4.3 of the mapping from said sample representation value V2i to said original representation value V1 i.

This results in a one-to-one correspondence between the triplets of the original color space SCO and the triplets of the sample color space SCC.

To obtain a prediction of the conversion values of the RGB triplets of the original color space, the LUT in the location identified by the RGB triplets of the original color space SCO is accessed.

The second identification module 106d comprises a third conversion submodule 106d3 configured to identify the color conversion TR OK by means of the described look-up table LUT technique.

In other words, the first processing station 10 is configured to identify the color conversion TR _ OK by means of the described look-up table LUT technique.

Unlike two previous solutions that determined mathematical functions to predict each RGB triplet transform, the LUT may not be able to fully map the triplets of the original color space SCO.

If this happens:

the invention comprises obtaining further images of the real tiles not yet mapped for mapping the original color space SCO.

The invention includes 1: 1-matching estimates missing triples as parameters of a linear regression system or alternatively parameters used to train an artificial neural network.

In other words, according to the invention, therefore, in order to obtain the complete mapping M _ OK, the following steps are included:

-verifying the mapping of the look-up table LUT;

if the look-up table LUT can not completely map the RGB triple of the original color space SCO;

-obtaining a further sample image IM _ PCS _ J for mapping a portion of the original color space SCO not yet mapped;

-using 1: 1-matching estimates missing RGB triplets as parameters of a linear regression system RL or alternatively parameters used to train an artificial neural network ANN.

Once the conversion function is obtained, the LUT is populated using linear regression or artificial neural network prediction to the locations identified by the missing RGB triplets of the original color space SCO.

The second identification module 106d (fig. 3) is configured to perform the described steps of obtaining a complete mapping M _ OK.

In other words, the first processing station 10 is configured to check whether the LUT does not completely map a triplet of the original color space SCO, to fetch further sample images of real tiles used to map a not yet mapped portion of the original color space SCO that has not yet been mapped, and to use 1: 1-matching estimates missing triples as parameters of a linear regression system or alternatively parameters used to train an artificial neural network.

Referring to fig. 2, step F5 includes modifying the original image file F _ PRINT according to the previously calculated color conversion TR _ OK, determining an adapted image file F _ ADAPT.

The technical effect achieved is that the original image file F _ PRINT is graphically adapted to the acquired pattern of the sample image IM _ PCs _ j of the ceramic carrier PC _ i in terms of color rendering.

The main technical effect of the invention is to achieve an optimal pattern/colour matching between the appropriately processed original pattern file and the image of the printed ceramic support acquired by means of the acquisition system.

As mentioned above, the present invention also has some technical effects; these two images (which differ only slightly in terms of the chromaticity but are only attributable to the production process) can be used to evaluate the correctness of the graphics/decoration, i.e. in other words to check the quality of the ceramic support produced.

A suitable image processing algorithm having two images that are very similar to each other may extract the differences between them, classify them according to size, shape, type, etc., and thereby determine an acceptability classification.

In an automatic inspection system, such a procedure greatly reduces the time of the setup step and ensures a higher reliability in extracting possible production defects, thus reducing false identifications thereof.

In fact, in order to achieve these technical effects, the invention comprises a quality control method for the ceramic support PC _ i.

The method comprises the following steps:

-generating a suitable image file F _ ADAPT starting from the original image file F _ PRINT, as previously described;

-receiving a printed ceramic carrier PC _ i;

-identifying a portion P _ F _ ADAPT in the adapted image file F _ ADAPT obtained as described previously, which corresponds to the graphic reproduced on the printed ceramic support PC _ i;

-detecting a graphic difference Dd between the graphic reproduced on the printed ceramic support PC _ i and the portion P _ F _ ADAPT of the adapted image file F _ ADAPT;

-classifying the printed ceramic carriers PC _ i into Ci according to the detected differences Dd, thereby achieving quality control of the printed ceramic carriers PC _ i.

To start the method, the invention comprises (fig. 1A) a quality control system 200 for a ceramic support PC _ i, the quality control system 200 comprising:

the graphics adaptation system 100 described previously, which is configured to generate an adapted image file F _ ADAPT starting from an original image file F _ PRINT;

a receiving device 400 for printing the ceramic carrier PC _ i;

a sorting system 300 coupled to the graphic adaptation system 100 and the receiving device 400 of the printed ceramic carrier PC _1, the sorting system 300 comprising the second processing station 110 (fig. 1A).

For the second processing station 110, the same structure/function/module notes indicated for the first processing station 10 are valid on pages 6 and 7.

The second processing station 110 includes:

-a study module 111 configured to identify a portion P _ F _ ADAPT in the adapted image file F _ ADAPT corresponding to the graphics reproduced on the printed ceramic support PC _ i;

a detection module 112 configured to identify a graphic difference (Dd) between the graphic reproduced on the printed ceramic support PC _ i and the portion P _ F _ ADAPT of the adapted image file F _ ADAPT;

a classification module 113 configured to classify the printed ceramic carriers PC _ i into classes Ci according to the detected differences Dd, thereby enabling quality control of the printed ceramic carriers PC _ i.

For example, the classification module 113 classifies tiles into categories where one category C1 represents a first selected tile, one category C2 represents a second selected tile, and one category C3 represents a third selected tile.

The detected differences Dd determining the classification are global differences in terms of color and local defects, such as missing decorations, presence of color drops, presence of color lines, absence of graphic parts, impurities, evaluated in terms of shape, position, size, intensity, etc.

Preferably, the graphic adaptation system 100 and the classification system 300 are provided on the same machine; in other words, the quality control system 200 is integral.

Optionally, the graphic adaptation system 100 and the classification system 300 are provided on different machines; in other words, the quality control system 200 is distributed.

In the second case, the image acquisition means of the two systems are mutually calibrated in the color space so as to be able to clearly use the adapted image file F _ ADAPT.

Furthermore, this technique (designed mainly for the quality control of ceramic supports decorated with random printing techniques) can also be used in the case of fixed-surface printing, avoiding the printing of all the necessary samples for comparison, allowing them to be digitally built starting from the original graphics of each surface.

In another aspect, the invention comprises a production system for producing ceramic supports comprising a quality control system for the ceramic supports PC _ i interposed between the image printing system on the ceramic supports PC _ i and the kiln for firing the ceramic supports themselves.

In particular, the quality control system is the system 100 described above.