Method for obtaining color sub-palettes from emotion board

Abstract

The present invention provides a method of obtaining color sub-palettes from a first mood board containing one or more digital color images, the method comprising the steps of: (1) generating a first color palette having k1 representative colors from the first mood board by a method taking into account insufficient representative colors, (2) providing a second color palette having k2 colors, (3) ranking the colors in the first color palette according to a scoring algorithm that assigns a score to each color in the first color palette based on a color difference of each color in the first color palette and each color in the second color palette, and (4) selecting a subset of ksub colors from the first color palette, the subset having a score corresponding to a minimum color difference from a color in the second color palette, the subset of colors being color sub-palettes. The invention also provides a data processing device and a computer program product.

Description

Technical Field

[0001] The present invention relates to a method for obtaining a color subpalette from a moodboard. The color subpalette is obtained from a color palette containing representative colors (moodboard) of a collection of digital color images. Background Art

[0002] Consumers wishing to select one or more paint colors, for example for decorating a room, are presented with a very large number of color options to choose from. Furthermore, selecting multiple paint colors that coordinate with each other, or, for example, with the color of a piece of furniture, is a highly subjective and can be a daunting task. Paint manufacturers typically provide pre-arranged or curated color palettes that may not reflect the preferences of consumers. Therefore, the present invention is intended to facilitate the selection process by providing consumers with personalized color palettes based on one or more color images (mood boards) selected by the consumer.

[0003] Determining a set of representative colors from a digital color image is useful for many applications. For example, to facilitate paint color selection, the method described herein can provide a color palette with representative colors based on one or more digital color images provided by a user. A user can, for example, collect and input a collection of one or more digital color images that he or she likes, and a color palette is provided to the user based on the colors in the input images. Several color images can be input individually or constitute a single mood board image or collage. Each image in the collection of images can be weighted differently to express certain preferences in the collection of images. For example, some of the images that constitute the mood board can be enlarged to amplify the weight of the colors in that particular image. US2021 / 075329A1 describes a method that transforms a digital image by calculating an image color palette based on the image; mapping the colors of the image color palette to the colors of a conceptual color palette representing a mood or scene, and transforming the colors of the digital image based on the mapping.

[0004] WO 2021 / 209413A1 describes a method for determining a color palette from a digital image. The method includes defining a predetermined number K of cluster centers in a color space; forming clusters by associating pixels to the nearest cluster centers; reducing the number of clusters to k by deleting and / or merging cluster centers according to a predefined threshold; and defining a corresponding representative color for each of the obtained k clusters.

[0005] When generating a color palette from a mood board, a number of representative colors may be identified, especially if the mood board contains many visually distinct images. Typically, in order to provide a good representation of all the important colors found in the images in the mood board, the palette needs to contain more than 8 colors, often more than 12 colors, such as 16 colors or more. However, such a large number of colors in a palette is inconvenient for users who need to select paint colors for decoration. In interior design schemes, usually only a few colors are used, such as 2 or 3, and usually no more than 5. This means that the user needs to select only a few colors from the large color palette created by the user's mood board, which may be difficult for users who are not experienced in color design.

[0006] It is therefore desirable to provide a method of deriving a color palette from a mood board having only a limited number of colors, e.g. 5 or fewer colors. The colors should be representative of the images contained in the mood board and be obtained by a meaningful reduction in the number of colors in the color palette created from the user's mood board. It is also desirable that such a method reflect the user's preferences regarding other aspects that are important in interior design, such as preferred design styles or currently trendy colors. Summary of the invention

[0007] The present invention provides a method for obtaining a color sub-palette from a first mood board comprising one or more digital color images, the method comprising the following steps:

[0008] (1) generating a first color palette having k1 representative colors according to the first mood board,

[0009] (2) providing a second color palette having k2 colors,

[0010] (3) sorting the colors in the first color palette according to a scoring algorithm that assigns a score to each color in the first color palette based on a color difference between each color in the first color palette and each color in the second color palette, and

[0011] (4) Select k from the first color palette sub a subset of colors having a score corresponding to a minimum color difference from a color in the second color palette, the subset of colors being the color sub-palette,

[0012] Wherein, the first color palette is obtained from the first mood board by a method comprising the following steps:

[0013] (a) obtaining at least one digital color image having p pixels, each pixel having a color value in an n-dimensional color space;

[0014] (b) regardless of the obtained image, defining a predetermined number K of cluster centers, the cluster centers being distributed along a plurality of lines in a cubic RGB color space, the plurality of lines intersecting at an intersection point located at the center of the cubic RGB color space, wherein K>k1;

[0015] (c) forming clusters by associating the color value of each pixel with the nearest cluster center; and after forming the clusters,

[0016] c1) For each cluster, re-determine the cluster center;

[0017] (d) reducing the number of clusters to k1 by deleting cluster centers and / or merging clusters, wherein the deletion includes:

[0018] - for each cluster, determine the number of pixels associated with that cluster, and

[0019] - if the number of pixels associated with a cluster center is below a predefined pruning threshold, deleting said cluster center;

[0020] and wherein the merger includes:

[0021] - determine the distance between two clusters, and

[0022] - merge those clusters whose distance is less than a predefined merging threshold;

[0023] wherein steps c), c1) and d) are iterated; and

[0024] (e) Based on each of the k1 clusters obtained, define the corresponding representative color,

[0025] (f) Based on k1 representative colors, a first color palette is formed.

[0026] In a second aspect, the invention provides a method of selecting a paint color, the method comprising the above method and further comprising selecting a paint color corresponding to each color of the color sub-palette.

[0027] In a third aspect, the invention provides a data processing apparatus comprising means for performing the steps of the method of the invention.

[0028] In a fourth aspect, the invention provides a computer program product comprising instructions which, when executed by a computer, cause the computer to perform the steps of the method of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1-3 A schematic flow chart of a method is shown in each case.

[0030] Figure 4 A schematic example of a predetermined K number of predefined cluster centers distributed in a predetermined pattern in a color space is shown. DETAILED DESCRIPTION

[0031] The invention is based on the sensible insight that the number of colors in a large color palette obtained from a mood board can be reduced and color palette subsets (color sub-palettes) created in a way that is meaningful to the user. Meaningful to the user means that the color selection is non-random and takes into account other user preferences, such as a specific theme, such as a preferred design style or currently trending colors.

[0032] This approach results in a color sub-palette containing fewer colors than the color palette originally obtained from the mood board, and containing only the colors that have the best match to the user preferences represented by another (second) color palette.

[0033] In step (1) of the method according to the invention, a first color palette with k1 representative colors is generated based on the digital color images contained in the first mood board. A mood board is a collection of digital color images containing at least one image and preferably at least two images. A mood board can be created by a user by selecting images that appeal to the user.

[0034] A color palette is a set of at least two colors. A color palette subset (color sub-palette) is a color palette with a reduced number of colors compared to an original color palette, where the color palette subset is obtained from the original color palette. The colors in the color palette subset represent a subset of the colors from the original color palette.

[0035] The first color palette contains representations of the colors in the digital color image(s) in the first mood board. The number k1 of colors in the first color palette is preferably at least 8, or at least 12, or more preferably at least 16.

[0036] A color palette is generated from a first mood board using a method that takes into account the underrepresented (emphasized) colors included in the mood board. An underrepresented or emphasized color is a color that occupies only a small area of ​​the digital image relative to other colors but forms a strong contrast with the surrounding environment. Such colors are prominent to human observers, but are easily ignored by traditional algorithms for color palette generation, which only include colors in the color palette that occupy the largest area of ​​the digital image. For example, a mood board can be composed of several images of an interior design, in which dark tones of blue and gray are the most common. However, in some of the images that make up the mood board, there may be small bright yellow features depicted, such as yellow pillows, which stand out in an environment that is mainly dark blue and gray. Overall, this bright yellow only covers a smaller portion of the mood board, that is, only a few pixels are yellow, but its presence can be clearly perceived by the user and may be in the mood board for this reason.

[0037] Such a method is known, for example, from WO 2021 / 209413 A1, the content of which is incorporated herein.

[0038] In particular, the first color palette is obtained by a method comprising the steps of:

[0039] (a) obtaining at least one digital color image having p pixels, each pixel having a color value in an n-dimensional color space;

[0040] (b) regardless of the obtained image, defining a predetermined number K of cluster centers, the K cluster centers being distributed along a plurality of lines in the cubic RGB color space, the plurality of lines intersecting at an intersection point located at the center of the cubic RGB color space, wherein K>k1;

[0041] (c) forming clusters by associating the color value of each pixel with the nearest cluster center; and after forming the clusters,

[0042] c1) For each cluster, re-determine the cluster center;

[0043] (d) reducing the number of clusters to k1 by deleting cluster centers and / or merging clusters, wherein the deletion includes:

[0044] - for each cluster, determine the number of pixels associated with that cluster, and

[0045] - if the number of pixels associated with a cluster center is below a predefined pruning threshold, deleting the cluster center; and wherein the merging comprises:

[0046] - determine the distance between two clusters, and

[0047] - merge those clusters whose distance is less than a predefined merging threshold;

[0048] wherein steps c), c1) and d) are iterated; and

[0049] (e) Based on each of the k1 clusters obtained, define the corresponding representative color,

[0050] (f) Form a first color palette based on k1 representative colors.

[0051] In the method, the number k1 is predefined. The number k1 of representative colors may be defined a priori as any desired number, but is preferably at least 8, or at least 12, or more preferably at least 16.

[0052] The first color palette generation method is as follows Figure 1-3 As shown in the schematic flow chart of the method, the first step a) of the method comprises:

[0053] a) Obtaining at least one digital color image having p pixels, each pixel having a color value in an n-dimensional color space.

[0054] The at least one digital color image can be a single image or a collection of images. A single image can be, for example, a combination of multiple color images, such as constituting a mood board reflecting the user's personal preferences. Typically, a digital color image is represented by red, green and blue channels, wherein each pixel has a red, green and blue color value assigned to it. In a 24-bit digital image, each channel has 8 bits, so that, for example, in the range from 0 to 255, each pixel can have 256 different red, 256 different green and 256 different blue color values. Therefore, the n-dimensional color space can be a three-dimensional RGB space, wherein the three axes of the RGB space define red, green and blue color values. The color value of each pixel can be represented by a vector in a 256×256×256 RGB vector space. It should be understood that the color space can also be another color space, such as a CMYK space, a CIELAB space or a CIEXYZ space.

[0055] The subsequent steps of this method are:

[0056] b) Regardless of the obtained image, a predetermined number K of cluster centers are defined, the cluster centers are distributed along a plurality of lines in the cubic RGB color space, the plurality of lines intersecting at an intersection point located at the center of the cubic RGB color space, wherein K>k1.

[0057] The K initial cluster centers are scattered in the color space. The number of K predefined clusters thus exceeds the number of k1 representative colors. Since the K initial cluster centers are predefined and distributed in the color space in a predetermined pattern, the initialization in step b) is non-random and the method gives reproducible results for a given digital color image. In other words, repeatedly performing the method on the same digital color image results in the same set of resulting representative colors.

[0058] Figure 4 shows a pattern of how the K initial clusters are seeded in the color space. In particular, Figure 4 A specific distribution of K predefined cluster centers is shown, which are distributed along multiple straight lines (e.g., 13) in RGB color space, such as extending between opposite corners, rib midpoints, and centers of planes in RGB space. These lines intersect at center points in RGB space and create a star configuration of predefined cluster centers. Using this star initialization, the K initial cluster centers are well dispersed in color space and effectively span the color space observed by the human eye. Preferably, the cluster centers are uniformly extended in color space (in a regular manner), i.e., at the same distance from each other. In this way, the method is more likely to identify underrepresented colors in at least one digital color image, i.e., small-scale and / or isolated clusters.

[0059] The step c) of the method comprises forming clusters by associating the color value of each pixel with the nearest cluster center. The cluster center nearest to the pixel can be based on Euclidean distance or any other distance metric. Clusters are defined by the set of pixel color values ​​associated with a single cluster center. When the cluster center did not have a pixel color value associated therewith, some clusters could be empty.

[0060] In step c1), after forming the clusters, for each cluster, the cluster center is re-determined. For example, the re-determined cluster center of the cluster can be a color value that minimizes the variance within the cluster, such as the average value of the color values ​​in the cluster. Other options for re-determining the cluster center of the cluster include setting the cluster center to the median, medoid, or other value of the color values ​​in the cluster. The re-determined cluster center can be a constituent color value of the cluster, i.e., a member of the color value of the cluster, but this is not necessary.

[0061] Step (d) of the method includes reducing the number of clusters to k1 by deleting cluster centers and / or merging clusters, wherein deleting includes:

[0062] - for each cluster, determine the number of pixels associated with that cluster, and

[0063] - if the number of pixels associated with a cluster center falls below a predefined pruning threshold, deleting the cluster center;

[0064] and wherein the merger includes:

[0065] - determine the distance between two clusters, and

[0066] -Merge those clusters whose distance is less than a predefined merging threshold.

[0067] In order to obtain a predetermined number of k1 representative colors, it is desirable to reduce the number of clusters and cluster centers from K to k1. This is accomplished by merging clusters and / or deleting cluster centers. Preferably, deletion and merging of clusters are performed. Preferably, the pruning threshold is set to zero so that only the cluster centers of empty clusters without pixels in the cluster are deleted. In some embodiments, when the cluster center of a non-empty cluster is removed, the pixels previously assigned to the cluster may be marked as unassigned or assigned to the cluster with the closest cluster center. If they are marked as unassigned, they may be assigned to different new clusters in the next iteration. Clusters whose centers are close to each other may be merged because these clusters may represent similar colors. The similarity between clusters may be defined by the distance between their corresponding cluster centers. The merging threshold may be defined as a distance, wherein the two clusters are merged when the distance between two clusters is less than the threshold distance. The merging and / or pruning threshold may be adjusted so that the number of clusters is reduced from K to k1.

[0068] The term "distance" as used herein refers to the similarity or dissimilarity between elements in the color space. It should be understood that any distance metric can be used within the scope of the method to determine the similarity or distance between elements in the color space, such as the 1-norm, 2-norm, 3-norm, ∞-norm, etc. For example, the distance between two color values ​​in the color space can be expressed as the Euclidean distance, i.e., the 2-norm distance. In this regard, the cluster center that is "closest" to a particular color value of a pixel is that particular cluster center for which the distance metric between that color value and any cluster center is minimized.

[0069] Also in this regard, the distance between two clusters refers to the similarity or dissimilarity between the clusters. For example, the distance between two clusters can be defined as the distance between their corresponding cluster centers, or the distance between their corresponding cluster boundaries. Preferably, the distance between two clusters is the distance between their corresponding cluster centers.

[0070] Clusters are merged by, for example, removing one or more cluster centers from a group of clusters that are similar to each other. For example, two similar clusters can be merged into a single cluster by deleting any one of the cluster centers, for example, the cluster center with the minimum or maximum number of pixels associated therewith. Optionally, the lowest chromaticity cluster center is deleted to avoid substantially non-color colors (e.g., black, white, and shades of gray) in the color space. The pixel values ​​associated with the deleted cluster centers can be reassociated with any remaining cluster centers. Similar clusters can also be merged by deleting old cluster centers and defining new cluster centers based on the color values ​​of the old cluster centers and / or similar clusters. The new cluster centers, for example, can be set to the average or weighted average of similar clusters and / or their corresponding centers.

[0071] Steps c), c1) and d) can be iterated until the convergence criteria are met. In this embodiment, after step d) has been completed, the method continues with step c) instead of step e). After meeting certain convergence criteria, the method continues with step e after step d). The convergence criteria can be a predefined number of iterations and / or a convergence metric. The convergence criterion can be that the number of clusters is reduced from K to k1. Iteration steps c) and c1) are similar to the steps taken in known clustering techniques, such as k-means clustering and similar clustering, in which the cluster centers converge to a local optimum. Including step d) in the iteration provides convergence of the number of cluster centers from K predefined clusters to k1 clusters. By relocating the cluster centers in step c1), the clusters are iteratively reduced from K to k1. The cluster centers can be relocated from their predefined positions (step b)) so that the distance between some cluster centers becomes smaller at each iteration. For example, some cluster centers converge to the same local optimum, and at some points, the distance between them is defined to be smaller than a predefined merging threshold. In step d) these clusters are merged until k1 clusters are obtained. The method may not converge to k1 clusters. In this case, the pruning threshold and / or the merging threshold may need to be set (eg manually) to different values.

[0072] After reducing the number of clusters in step d), the method comprises the following steps:

[0073] e) Based on each of the obtained k1 clusters, define the corresponding representative color.

[0074] The representative color may be, for example, the corresponding average color value of the cluster or the center of the cluster. The representative color may also be a component color, i.e., the color value of a member of p pixels in the original digital color image. In this case, the pixel color value closest to the average color value of the cluster (or the center of the cluster) may be selected, or alternatively, the median or median of the cluster may define the representative color of the cluster.

[0075] Finally, in step (f), the defined k1 representative colors form a first color palette.

[0076] As a result of step (1), a first color palette is generated. When the described method is implemented on a computer or a mobile device with a screen, the color palette is preferably hidden from the user (not displayed on the screen).

[0077] In step (2), a second color palette having k2 colors is provided. The second color palette may, for example, contain colors representing a particular design style (modern, industrial, romantic, etc.) or a particular word or phrase (brisk, hazy, bright sky). In this case, the second color palette may be obtained from a digital color image representing the corresponding design style, word or phrase. Suitably, the second color palette may be generated from a second mood board comprising one or more digital color images. Preferably, the second color palette is generated from the second mood board using the algorithm described above for the first color palette taking into account the emphasis colors.

[0078] In other embodiments, the second color palette is not obtained from a mood board, but is already an existing color palette, such as curated by a paint manufacturer or compiled by a designer. An example of such a palette is a color palette containing trendy colors of a certain year, such as AkzoNobel's Sikkens ColorFutures TM .

[0079] The number k2 of colors in the second color palette is not essential, and any practical number can be used. It can be the same number as k1, or a lower or higher number. In some embodiments, the number k2 of colors is preferably at least 8, or at least 12, or more preferably at least 16. In other embodiments, the number k2 of colors can be less than 8, for example 5 or less. The number k2 of colors can also be 1, which means that the second color palette consists of only one color.

[0080] In step (3), the colors in the first color palette are ranked according to a scoring algorithm that assigns a score to each color in the first color palette based on the color difference between each color in the first color palette and each color in the second color palette. In this step, the colors in the first color palette are compared to the colors in the second color palette. This is done using a scoring algorithm.

[0081] Any suitable scoring algorithm that allows comparing sets of colors to each other and generating a score assigned to each color may be used. Preferably, the scoring algorithm comprises the steps of: for each color of the first color palette, calculating the color difference with each color in the second color palette, determining the minimum color difference from all obtained color differences, and assigning a score to the color from the first color palette based on the minimum color difference, thereby assigning a score to all colors of the first color palette.

[0082] Color difference is expressed as a value. The color difference between two colors can be based on the distance between these colors in an n-dimensional color space. The integer n can be any possible integer, such as 1, 2, 3, and larger numbers, such as 6, 9, 12, for example, a three-dimensional RGB space or a CIELAB space can be used, preferably a CIELAB space. It should be understood that the color space can also be another color space such as an HSV or HSL space. A known algorithm for calculating color difference can be used, such as the CIEDE2000 color difference formula. Color difference is not limited to a multidimensional color space. Color difference also includes the difference in brightness or chromaticity, which represents the corresponding dimension in a multidimensional color space. For example, chromaticity (C) is a dimension in a 3-dimensional CIE-LCh space. Brightness (L) is a dimension from a 3-dimensional CIE-LAB space (or Y from a CIE-XYZ space).

[0083] When all colors of the first color palette receive scores based on the above criteria, the colors of the first color palette are sorted according to their scores, for example from the lowest score to the highest score or vice versa. Depending on the scoring algorithm used, the highest score may, for example, represent the smallest color difference from the second color palette. The color with the highest score will represent the color with the lowest color difference from the second color palette.

[0084] Therefore, in step (4), k is selected from the first color palette sub The subset of colors that have the smallest color difference (e.g., highest score) with the colors in the second color palette is the color sub-palette. sub corresponds to the number of colors desired in the final color sub-palette and is preferably a predefined number. The number of colors k sub It is usually less than 8, preferably less than 5, more preferably less than 2, 3 or 4. In any case, the number of colors k sub Less than the number k1 of colors in the first color palette.

[0085] For example, a first color palette may contain 16 colors, while a second color palette may contain 9 colors. First, the color difference between the first color of the 16-color palette and all the colors of the second color palette (which has 9 colors) is calculated, resulting in 9 color difference values ​​for the first color. The assigned score for that color is then the lowest color difference, i.e., the color difference between the first color of the 16-color palette and its closest match in the 9-color palette. The process then continues to the second color in the 16-color palette, and the comparison is repeated with all 9 colors in the second color palette, calculating and assigning a score for the second color in the 16-color palette. The process is repeated for all 16 colors, ending with 16 scores, one for each color. The final 4-color palette is determined by selecting the 4 colors with the lowest scores from the 16-color palette.

[0086] In some embodiments, the second color palette may contain only one color, such as blue. In this case, the target sub-palette will contain the color with the lowest color difference score with that particular color, such as the bluest color from the first color palette.

[0087] As a further stop, the colors in the obtained sub-palette can optionally be compared with standard paint colors (e.g. of commercially available paints) and converted into standard paint colors. This can be achieved by replacing the colors in the sub-palette with the most similar standard paint color. The color difference can be calculated in the same way as described above. This has the advantage that the user obtains a color sub-palette which only shows commercially available colors, such as wall paint or wood paint.

[0088] In one embodiment, in a method of selecting a paint color, each color of a color sub-palette is replaced by the most similar paint color within a range of commercially available paint colors.

[0089] When the described method is implemented on a computer or a mobile device having a screen, the color sub-palette obtained in step (4) is preferably displayed to the user on the screen. Any suitable color screen conventionally used with computers and mobile devices can be used.

[0090] After obtaining the color sub-palette in step (4), a paint of any one of the colors contained in the sub-palette can be prepared. This can be performed, for example, using a paint mixing device as is common in the paint industry.

[0091] The present invention also provides a data processing device comprising means for performing the steps of the above method. The method can be performed on a data processing device such as a point of sale computer system or a mobile computing system including a display, such as a smart phone, tablet computer, laptop computer, etc. The display is preferably a color display.

[0092] The present invention also provides a computer program product comprising instructions, which, when executed by a computer, cause the computer to perform the steps of the method of the present invention. The computer program product may be an application or web-based software loaded and executed on a general-purpose computer or mobile computing system.

[0093] In one embodiment, the computer program will cause the computer to output an identifier for each color of the color sub-palette, the identifier being the color of the paint. For example, the identifier may be a code or name identifying the color from a range of paint colors.

[0094] In one embodiment, the computer program will cause the computer to output a link to an ordering platform from which paint of at least one color of the color sub-palette can be ordered. The ordering platform can be, for example, in a mobile computing system application or a web browser. The ordering platform can be, for example, a web store.

[0095] It will be understood that all features and options mentioned in relation to the method are equally applicable to the system and computer program product, and vice versa.It will also be clear that any one or more of the above-mentioned aspects, features and options may be combined.

[0096] In the claims, any reference signs in brackets shall not be construed as limiting the claim. The word "comprising" does not exclude the presence of other features or steps than those listed in a claim. Furthermore, the words "a" and "an" shall not be interpreted as being limited to "only one", but are used to mean "at least one", and do not exclude a plurality. The mere fact that certain measures are recited in mutually different claims does not indicate that a combination of these measures cannot be used to advantage.

Claims

1. A method for obtaining a color sub-palette from a first mood board comprising one or more digital color images, the method comprising the following steps: (1) generating a first color palette having k1 representative colors according to the first mood board, (2) providing a second color palette having k2 colors, (3) sorting the colors in the first color palette according to a scoring algorithm that assigns a score to each color in the first color palette based on a color difference between each color in the first color palette and each color in the second color palette, and (4) Select k from the first color palette sub a subset of colors having a score corresponding to a minimum color difference from a color in the second color palette, the subset of colors being the color sub-palette, Wherein, the first color palette is obtained from the first mood board by a method comprising the following steps: (a) obtaining at least one digital color image having p pixels, each pixel having a color value in an n-dimensional color space; (b) regardless of the obtained image, defining a predetermined number K of cluster centers, the cluster centers being distributed along a plurality of lines in a cubic RGB color space, the plurality of lines intersecting at an intersection point located at the center of the cubic RGB color space, wherein K>k1; (c) forming clusters by associating the color value of each pixel with the nearest cluster center; and c1) after forming the clusters, for each cluster, re-determining the cluster center; (d) reducing the number of clusters to k1 by deleting cluster centers and / or merging clusters, wherein the deletion includes: - for each cluster, determine the number of pixels associated with that cluster, and - if the number of pixels associated with a cluster center is below a predefined pruning threshold, deleting said cluster center; and wherein the merger includes: - determine the distance between two clusters, and - merge those clusters whose distance is less than a predefined merging threshold; wherein steps c), c1) and d) are iterated; and (e) Based on each of the k1 clusters obtained, define the corresponding representative color, (f) Based on k1 representative colors, a first color palette is formed.

2. The method according to claim 1, wherein: The second color palette is generated based on a second mood board comprising one or more digital color images.

3. The method according to claim 2, wherein: The digital color image in the second color palette represents a design style, a word, or a phrase.

4. The method according to claim 1, wherein: The colors in the second color palette are selected by a designer.

5. The method according to any one of claims 1 to 4, wherein: The scoring algorithm comprises the following steps: for each color in the first color palette, calculating the color difference with each color in the second color palette, determining the minimum color difference based on all the obtained color differences, and assigning a score to the color from the first color palette based on the minimum color difference, thereby assigning a score to all colors of the first color palette.

6. The method according to any one of claims 1 to 5, wherein: k1 is at least 12.

7. The method according to any one of claims 1 to 6, wherein: k sub is 5 or less.

8. A method of selecting a paint color, comprising the method according to any one of claims 1 to 7, and further comprising selecting a paint color corresponding to each color of the color sub-palette.

9. The method according to claim 8, wherein: Each color in the color sub-palette is replaced with the most similar paint color available from a range of commercially available paint colors.

10. A data processing device comprising means for performing the steps of the method according to any one of claims 1 to 9.

11. A computer program product comprising instructions which, when said program is executed by a computer, cause said computer to perform the steps of the method according to any one of claims 1 to 9.

12. A computer program product according to claim 11 which will cause the computer to output an identifier for each colour of the colour sub-palette, the identifier being for the colour of the paint.

13. A computer program according to claim 11 or 12, which will cause the computer to output a link to an ordering platform from which paint of at least one colour in the colour sub-palette can be ordered.

Images