Method and system of colour matching

Abstract

A method of colour matching comprising loading an image containing an item of interest such as an article of clothing, creating a sample comprising a region of the item of interest in the image, determining contours within the sample, removing contours from the sample to identify a colour profile of the sample; and matching the colour profile of the sample with at least one colour category from a predetermined colour mapping such as determined by seasonal colour analysis.

Description

METHOD AND SYSTEM OF COLOUR MATCHINGFIELD OF THE INVENTION

[0001] The invention relates to colour matching. In particular, the invention relates, but is not limited, to a method and system of colour matching images with a predefined palette.BACKGROUND TO THE INVENTION

[0002] Reference to background art herein is not to be construed as an admission that such art constitutes common general knowledge.

[0003] Photographs, and in particular website images, are often supplied to provide an indication of a product. In some cases it is desirable to match the colour of the product in that photograph with a predefined palette. However, due to various factors, including contours and shadows, it can difficult to accurately assess the colour of the article.

[0004] For example, in the clothing industry, some consumers have a preferred colour palette for their clothing. The colour palette may be determined by various factors including their skin tone and hair colour of the person. Such palettes are determined to compliment the natural features of the person, with categories of colour palettes being known among well informed consumers.

[0005] For example, one colour system has twelve main categories of colour profiles that are considered to be well suited to individuals having different physical parameters. Some consumers spend consideration time, effort, and money engaging a professional to determine colours that suit their physiological features. Once this has been done, they are then compelled to find suitably matching clothing, or the like, that fits within their predetermined colour profile. When looking for suitable products online it can, however, bedifficult to: (a) accurately determine the colour of a product from a photograph that may include, for example, natural shading and other contours, and (b) find products that fall within the predefined colour palettes of the individual.OBJECT OF THE INVENTION

[0006] It is an aim of this invention to provide a method and system of colour matching which overcomes alleviates one or more of the disadvantages or problems described above, or which at least provides a useful alternative and / or commercial choice.

[0007] Other preferred objects of the present invention will become apparent from the following description.SUMMARY OF INVENTION

[0008] In one form, although it need not be the only or indeed the broadest form, there is provided a method of colour matching comprising: loading an image containing an item of interest; creating a sample comprising a region of the item of interest in the image; determining contours within the sample; removing contours from the sample to identify a colour profile of the sample; and matching the colour profile of the sample with at least one colour category from a predetermined colour mapping.

[0009] The step of loading an image containing an item of interest may comprise downloading an image from a source. The source may be a webpage. The image may be one of a plurality of images on the webpage. The item of interest may comprise a consumer item. The consumer item maycomprise an item of clothing. The consumer item may comprise a personal accessory.

[0010] The step of creating a sample may comprise selecting a portion of the image around a sampling point. The sampling point may be located within a central region of the image. The sampling point may be derived from a determination of an item in the image. The sampling point may be located within a region determined to be the item in the image. The determination of an item in the image may comprise using a machine learned model. The sampling point may be located in a central region of the item in the image. Determination of a location of the sampling point within the image may comprise selecting a sampling point in the centre of the image. The portion of the image may comprise a cropped portion of the image. The cropped portion of the image may be cropped by between approximately 25-75%, preferably by 30-70%, and even more preferably between 40 and 60%. In a preferred form, the cropped portion of the image is cropped by approximately 50%.

[0011] The step of determining contours within the sample may comprise converting the sample to greyscale. The step of determining contours within the sample may comprise converting the sample to a binary representation of at least the sample of the image. The step of converting the sample to binary representation of at least the sample may comprise thresholding each pixel of the sample to one of two colours or shades of colour. The two colours or shades of colour may consist of black and white. The step of thresholding may comprise binning each pixel relative to a threshold value. The threshold value may be predetermined. The threshold value may comprise an average shade of grey in the greyscale image. The method may comprise the further step of determining an average shade of grey in the greyscale sample.

[0012] The step of determining contours within the sample may comprise determining segments in the binary representation of at least the sample.Determining segments in the binary representation of at least the sample may comprise finding contours. Finding contours find may comprise locating boundaries between the colours or shades of colour in the binary representation of at least the sample.

[0013] The step of removing contours from the sample to identify a colour profile of the sample may comprise masking the sample with areas determined to be contours. The areas determined to be contours may be masked and / or removed from the sample. The areas determined to be contours may be required to be of a minimum size. The minimum size may be a predetermined value. The minimum size may be a predetermined ratio of the sample size.

[0014] The step of removing contours from the sample to identify a colour profile of the sample may comprise filtering the sample. The step of removing contours from the sample to identify a colour profile of the sample may comprise determining average RGB (Red, Green, and Blue) values on the filtered sample. The step of filtering the sample may comprise determining an average pixel intensity in the sample. The step of filtering the sample may comprise removing outlier histogram frequency intensities. Removing outlier histogram frequency intensities may comprise removing histogram frequencies more than a standard deviation from an average intensity.

[0015] The step of matching the colour profile of the sample with at least one colour category from a predetermined colour mapping may comprise determining an indicative colour value of the sample. The indicative colour value of the sample may comprise an average colour value of the sample after the contours have been removed. The indicative colour value may comprise RGB values. The step of matching the colour profile of the sample may comprise converting the indicative colour value from a first colour model to a second colour model. The first colour model may comprise RGB values. The second colour model may comprise HSL (Hue, Saturation, and Lightness)values. The step of matching the colour profile of the sample may comprise converting RGB values of the indicative colour value to HSL values.

[0016] The predetermined colour mapping may comprise colour categories in the form of ranges of colour values within a colour model. The colour categories of the predetermined colour mapping may comprise ranges of colour values as HSL values. The ranges of colours within a colour model may match predefined styles.

[0017] The ranges of colour values may comprise a plurality of selected colour values each defining the centre of a range. Each range may be defined by one or more percentage value shifts from the selected colour value. The one or more percentage value shifts may comprise a hue value shift, a saturation value shift, and a lightness value shift. The hue value shift may be a percentage between 1 % and 10%, preferably between 2% and 5%. The saturation value shift may be a percentage between 1 and 10%, preferably between 2% and 5%. The lightness value shift may be a percentage between 1 and 10%, preferably between 2% and 5%.

[0018] The step of matching the colour profile of the sample with at least one colour category from a predetermined colour mapping may comprise identifying one or more colour categories from the predetermined colour mapping that the colour profile values fall within.

[0019] Further features and advantages of the present invention will become apparent from the following detailed description.BRIEF DESCRIPTION OF THE DRAWINGS

[0020] By way of example only, preferred embodiments of the invention will be described more fully hereinafter with reference to the accompanying figures, wherein:

[0021] Figure 1 illustrates a flow chart of a method of colour matching; and

[0022] Figure 2 illustrates a flow chart of a method of colour matching with sub-steps.DETAILED DESCRIPTION OF THE DRAWINGS

[0023] Figures 1 and 2 illustrate a flow chart of a method of colour matching that comprises loading an image 100, taking a sample of the image 200, determining contours 300, removing contours 400, and matching 500. Each of these steps will now be described in further detail.

[0024] At step 100, an image 10 is loaded. The image 10 is typically an image of interest to be categorised. In the present example, the image 10 is an image of an item of interest in the form of an item of clothing 12 being worn by a model 14. By way of example, the invention will be described with respect to categorising the item of clothing 12 into at least one colour category of a predetermined colour mapping. Such colour categories may be predetermined to match features of different people based upon, for example, skin tone, hair colour, and eye colour. In use, the image 10 may be loaded from a website, or the like. It may be an image of a product for sale that a consumer may be interested in purchasing. Rather than the consumer looking at all items for sale on a plurality of webpages to try to find items that match their preferred colour category from the predetermined colour mapping, they may prefer to only view items determined to be within their preferred colour category from the predetermined colour mapping to save time and effort.

[0025] Although the invention is described with respect to an item of clothing, it should be appreciated that it could be applied to other items such as, for example, a personal accessory. Such personal accessories may be fashion accessories such as shoes, belts, wallets, hats, watches, and the like. It shouldalso be appreciated that the method of colour matching may be applied to other industries such as, for example, colour matching vehicle colours.

[0026] At step 200, a sample 20 from the image 10 is taken. The sample 20 is preferably located centrally within the item of interest 12. The sample 20 may be taken from a central sampling point of the image 10. The size of the sample 20 may be selected according to predetermined dimensions or, more preferably, as a ratio of the image 10 size.

[0027] At step 300, contours in the sample are determined. The sample 20 may be processed to determine and outline boundaries in the sample 20. As shown in figure 2, in a preferred form step 300 may be broken down into substeps as follows. First, the at least the sample 20 of the image 20 is converted to greyscale 310. Secondly, the greyscale sample is converted to a binary representation by thresholding pixels 320 to either of the binary values. In preferred forms the thresholding converts the greyscale image to a binary black and white image, but it should be appreciated that other binary values (e.g. other than black and white) may achieve the same desired outcome. The binary representation is then processed to identify boundaries between the binary values and determine segments 330, preferably of a predetermined minimum size, which represent areas of contour 32. The predetermined minimum size may be a fixed minimum size (e.g. 500 pixels) or, more preferably, a percentage of the size of the sample and / or image.At step 400, the identified areas of contours are removed. This can be performed by masking the sample with areas determined to be contours 32. The remaining areas of the sample 20 can then be averaged to a colour profile of the sample 40. This may be performed by calculating an average RGB value on the remaining areas of the sample 20 using pixel intensity. A histogram of pixel intensities can be used, where outlier pixel intensities, such as at a maximum frequency intensity, can be removed. Such outlier pixel intensitiescan be identified by calculating an average pixel intensity and identifying pixel intensities more than one standard deviation from that average.

[0028] At step 500 the colour profile of the sample 40 is compared to at least one colour category 52 from a predetermined colour mapping 50. As shown in figure 2, in a preferred form step 500 may be broken down into sub-steps as follows. RGB values from the profile of the sample 40 are determined 510 and then converted to HSL values 520. The HSL values can then be compared 530 against colour categories 52 from a predetermined colour mapping 50 and any matched colour categories 52 are returned 540. It should be appreciated that although RGB and HSL colour models are preferred, that other colour models may be utilised instead. Furthermore, it may be possible to perform a colour model conversion much earlier in the process such as, for example, when loading the image. In this way, images with varying colour models can be loaded and processed in the same colour space.

[0029] In the present example, the colour categories 52 of the colour mapping 50 may be centred around predetermined colours considered to suit particular features of different people. The illustrated colour mapping 50 has 12 colour categories 52, each comprising a plurality of different colour ranges. For the purpose of matching each colour in each colour mapping may represent a range of colour model values.

[0030] In the HSL colour model suitable ranges may defined by taking HSL values from each colour and allowing for a percentage value shift from the selected colour values. In a preferred form a ±9 absolute hue value shift, ±3% saturation value shift, and ±3% lightness value shift may be applied to each predetermined colour to provide the range. For example, a predetermined HSL colour of interest may have a value of (38, 85%, 64%) which may then result in a colour range of (31 -47, 82-88%, 61 -67%) for one of the colour categories 52. It should be appreciated, however, that larger and smaller value shiftpercentages could be utilised to provide broader or narrower ranges for each colour category.

[0031] Advantageously, the present invention allows products in images to be matched to colour categories from a predetermined colour mapping accurately. In the clothing example, this allows consumers to readily identify items of clothing that match their preferred colour category quickly and efficiently without having to open and assess each image individually and / or have to sort through numerous items of non-matching colour compatibility. By following the process of the invention colours of an article of interest can be accurately determined taking into account contours, and the like. Without this, colour indications of the article can be heavily influenced by shadows and reflections for example. The method of the present invention is able to take such influences into account to provide a genuine and accurate indication of colour of the item of interest, allowing it to be matched as required by the end user.

[0032] In this specification, adjectives such as first and second, left and right, top and bottom, and the like may be used solely to distinguish one element or action from another element or action without necessarily requiring or implying any actual such relationship or order. Where the context permits, reference to an integer or a component or step (or the like) is not to be interpreted as being limited to only one of that integer, component, or step, but rather could be one or more of that integer, component, or step etc.

[0033] The above description of various embodiments of the present invention is provided for purposes of description to one of ordinary skill in the related art. It is not intended to be exhaustive or to limit the invention to a single disclosed embodiment. As mentioned above, numerous alternatives and variations to the present invention will be apparent to those skilled in the art of the above teaching. Accordingly, while some alternative embodiments havebeen discussed specifically, other embodiments will be apparent or relatively easily developed by those of ordinary skill in the art. The invention is intended to embrace all alternatives, modifications, and variations of the present invention that have been discussed herein, and other embodiments that fall within the spirit and scope of the above described invention.

[0034] As used herein, an element or operation recited in the singular and proceeded with the word “a” or “an” should be understood as not excluding plural elements or operations, unless such exclusion is explicitly recited. Furthermore, references to “one embodiment” of the present disclosure are not intended to be interpreted as excluding the existence of additional embodiments that also incorporate the recited features.

[0035] In this specification, the terms ‘comprises’, ‘comprising’, ‘includes’, ‘including’, or similar terms are intended to mean a non-exclusive inclusion, such that a method, system or apparatus that comprises a list of elements does not include those elements solely, but may well include other elements not listed.

Claims

CLAIMS:

1. A method of colour matching comprising: loading an image containing an item of interest; creating a sample comprising a region of the item of interest in the image; determining contours within the sample; removing contours from the sample to identify a colour profile of the sample; and matching the colour profile of the sample with at least one colour category from a predetermined colour mapping.

2. The method of claim 1, wherein the step of creating a sample comprises selecting a portion of the image around a sampling point.

3. The method of claim 2, wherein the sampling point is derived from a determination of an item in the image.

4. The method of claim 2 or 3, wherein the portion of the image comprises a cropped portion of the image whereby the image has been cropped by between approximately 25-75%.

5. The method of any one of claims 1 to 4, wherein the step of determining contours within the sample comprises converting the sample to greyscale.

6. The method of any one of claims 1 to 5, wherein the step of determining contours within the sample comprises converting the sample to a binary representation of at least the sample of the image.

7. The method of claim 6, wherein the step of converting the sample to binary representation of at least the sample comprises thresholding each pixel of the sample to one of two colours or shades of colour.

8. The method of claim 7, wherein the two colours or shades of colour consist of black and white.

9. The method of claim 7 or 8, wherein the step of thresholding comprises binning each pixel relative to a threshold value10. The method of any one of claims 7 to 9, wherein the step of determining contours within the sample comprises determining segments in the binary representation of at least the sample.

11. The method of claim 10, wherein determining segments in the binary representation of at least the sample comprises finding contours by locating boundaries between the colours or shades of colour in the binary representation of at least the sample.

12. The method of any one of claims 1 to 11 , wherein the step of removing contours from the sample to identify a colour profile of the sample comprises masking the sample with areas determined to be contours.

13. The method of claim 12, wherein the areas determined to be contours are masked and / or removed from the sample.

14. The method of claim 12 or 13, wherein the areas determined to be contours are of a minimum size that is a predetermined ratio of the sample size.

15. The method of any one of claims 1 to 14, wherein the step of removing contours from the sample to identify a colour profile of the sample comprises filtering the sample and determining average RGB (Red, Green, and Blue) values on the filtered sample.

16. The method of claim 15, wherein the step of filtering the sample comprises determining an average pixel intensity in the sample and removing outlier histogram frequency intensities.

17. The method of any one of claims 1 to 16, wherein the step of matching the colour profile of the sample with at least one colour category from a predetermined colour mapping comprises determining an indicative colour value of the sample, the indicative colour value of the sample comprising anaverage colour value of the sample after the contours have been removed.

18. The method of claim 17, the step of matching the colour profile of the sample comprises converting the indicative colour value from a first colour model to a second colour model.

19. The method of claim 18, wherein the first colour model comprises RGB values and the second colour model comprises HSL (Hue, Saturation, and Lightness) values.

20. The method of claim 19, wherein the step of matching the colour profile of the sample comprises converting RGB values of the indicative colour value to HSL values.

21. The method of any one of claims 1 to 20, wherein the predetermined colour mapping comprises colour categories in the form of ranges of colour values within a colour model.

22. The method of claim 21 , wherein the ranges of colour values comprise a plurality of selected colour values each defining the centre of a range and wherein each range is defined by one or more percentage value shifts from the selected colour value.

23. The method of claim 22, wherein the one or more percentage value shifts comprises a hue value shift, a saturation value shift, and a lightness value shift.

24. The method of any one of claims 1 to 23, wherein the step of matching the colour profile of the sample with at least one colour category from a predetermined colour mapping comprises identifying one or more colour categories from the predetermined colour mapping that the colour profile values fall within.

25. The method of any one of claims 1 to 24, wherein the step of loading an image containing an item of interest comprises downloading an image from a source.

26. The method of claim 25, wherein the source is a webpage and the image is one of a plurality of images on the webpage.

27. The method of any one of claims 1 to 26, wherein the item of interest comprises an item of clothing.

Images