TEXTILE MATCHING USING COLOR AND PATTERN RECOGNITION AND METHODS OF USE

• US 20180046882A1
• Filed: 11/03/2014
• Published: 02/15/2018
• Est. Priority Date: 11/03/2014
• Status: Active Grant

• Current Assignee: myDiModa, Inc.
• Original Assignee: myDiModa, Inc.
• Inventors: CARROLL, Terrence A., HUSSAMI, Peter, KALECZ-SIMON, Andrs, JIN, Bin
• Granted Patent: US 10,460,199 B2
• Time in Patent Office: Days
• CPC Class Codes: G06F 16/58 Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
• CPC Class Codes: G06F 16/583 using metadata automatically derived from the content
• CPC Class Codes: G06F 16/5838 using colour
• CPC Class Codes: G06F 16/60 of audio data
• CPC Class Codes: G06F 16/63 Querying
• CPC Class Codes: G06F 18/24155 Bayesian classification
• CPC Class Codes: G06Q 30/0631 Item recommendations
• CPC Class Codes: G06Q 30/0643 Graphical representation of items or shoppers
• CPC Class Codes: G06T 2200/24 involving graphical user interfaces [GUIs]
• CPC Class Codes: G06T 2207/10024 Color image
• CPC Class Codes: G06T 2207/20021 Dividing image into blocks, subimages or windows
• CPC Class Codes: G06T 2207/30196 Human being; Person
• CPC Class Codes: G06T 2210/16 Cloth
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06T 7/90 Determination of colour characteristics

Abstract

Textile matching using color and pattern recognition and methods of use are provided herein. An example method includes analyzing an image (305) of a first article of clothing to obtain color information and pattern information (205-230), comparing the color information and pattern information of the first article of clothing to color information and pattern information for a plurality of other articles of clothing (using Bayesian probability analysis to determine matched pairs, and providing a user with wardrobe suggestions using the matched pairs (705-740).

Claims

1. A method, comprising:

analyzing an image of a first article of clothing to obtain color information and pattern information;

locating a second article of clothing for pairing with the first article of clothing by;

obtaining a plurality of possible matching articles of clothing from a database, each for the plurality of possible matching articles comprising color and pattern information;

calculating a set of probability values, the set of probability values corresponding to possible matches between the color and pattern information of the first article of clothing and the color and pattern information of each of the plurality of possible matching articles, the set of probability values each having a unique range of values measured in increments from a first of the set of probability values to a last of the set of probability values;

calculating a sum of the probability values;

selecting a random number between zero and the sum of the probability values;

selecting one of the plurality of possible matching articles of clothing based on a match between the random number and one of the unique range of values; and

displaying the one of the plurality of possible matching articles of clothing to the user.

2. The method according to claim 1, wherein the plurality of possible matching articles of clothing are selected based on an initial rule based analysis between the color and pattern information of the first article of clothing and the color and pattern information for the plurality of possible matching articles of clothing.

3. The method according to claim 2, wherein the initial rule based analysis comprises fashion rules that apply to color combinations, event criteria, calendar date, or combinations thereof.

4. The method according to claim 1, further comprising:

receiving feedback from a user in response to the selection of the a possible matching article of the plurality of possible matching articles of clothing;

if the feedback is positive, calculating an updated probability using;

(P_PRIOR*P_GOOD)/(P_PRIOR*P_GOOD+(1−

P_PRIOR)*(1−

P_BAD)

where P_PRIOR is a current probability for a possible matching article, and P_GOOD and P_BAD are both selected between 0.5 and 1;

the updated probability indicating that the possible matching article is more likely to be selected in a future pairing than the possible matching article with the current probability.

5. The method according to claim 1, further comprising:

receiving feedback from a user in response to the selection of the a possible matching article of the plurality of possible matching articles of clothing;

if the feedback is positive, calculating an updated probability using;

1−

((1−

P_PRIOR*P_BAD)*P_BAD)/(1−

P_PRIOR)*P_BAD+P_PRIOR*(1−

P_GOOD))

where P_PRIOR is a current probability for a possible matching article, and P_GOOD and P_BAD are both selected between 0.5 and 1;

selecting a second or more of the plurality of possible matching articles of clothing based on a match between the random number and one of the unique range of values; and

providing the second or more of the plurality of possible matching articles of clothing to the user.

6. The method according to claim 1, further comprising selecting an outfit for a user comprising the first article of clothing and a plurality of additional articles of clothing, each of the first article of clothing and the plurality of additional articles of clothing having a different type so as to comprise the outfit.

7. The method according to claim 1, further comprising repeating steps of claim 1 to identify a plurality of articles of clothing for a user that are matched to the first article of clothing so as to create a virtual wardrobe.

8. The method according to claim 1, wherein the first article of clothing is for sale, further wherein the first article of clothing is selectable for purchase only if one of the plurality of possible matching articles of clothing are successfully matched to the first article of clothing.

9. The method according to claim 1, further comprising receiving event criteria from a user of the first article of clothing, the event criteria defining an initial set of rules used in obtaining a plurality of possible matching articles of clothing from the database.

10. The method according to claim 1, further comprising obtaining skin color information for a user who will wear the first article of clothing and the second article of clothing.

11. The method according to claim 1, wherein color information is obtained by:

performing a color space conversion of the image of the first article of clothing;

performing color recognition of the color space converted image to obtain color information;

extracting image segmentation information for the image that identifies image segments;

for each identified image segment, obtaining dominant color information by performing a focusing region selection of the color information of an image segment; and

calculating an average of the dominant color information for the identified image segments.

12. The method according to claim 11, wherein pattern information is obtained by:

calculating an edge orientation field using a Sobel operator in a dominant color region of the image;

calculating a split of an edge orientation of the image;

calculating a mean orientation in each sub-region of the dominant region;

obtaining an output of a strip and square image using the mean orientation; and

obtaining a output of a dot image and empty image using the dominant color information and the split of the edge orientation of the image.

13. The method according to claim 11, further comprising detecting skin color information for a wearer of the first article of clothing, wherein the skin color information is used in the calculation of the dominant color information.

14. The method according to claim 11, wherein the color recognition uses any of:

(1) k-nearest neighbors analysis, a red, green, blue analysis, a hue, saturation, and value analysis, and a cyan, magenta, yellow, key analysis.

15. The method according to claim 1, further comprising a clothing type criteria from a user of the first article of clothing, the clothing type criteria defining a kind of the first article of clothing, the user also selecting a clothing type criteria for the second article of clothing.

16. The method according to claim 1, further comprising:

selecting for purchase, the one of the plurality of possible matching articles, if selected as acceptable by the user.

17. A method, comprising:

providing a virtual wardrobe comprising articles of clothing, each of the articles of clothing comprising color information and pattern information;

analyzing an image of an article of clothing for purchase, to obtain color information and pattern information for the article of clothing for purchase;

identifying at least one color and pattern match between the color information and pattern information for the article of clothing for purchase and the color information and pattern information of each of the articles of clothing, so as to create a matched pair of articles of clothing;

receiving an indication that the article of clothing for purchase has been purchased; and

adding the article of clothing for purchase to the virtual wardrobe.

18. A method, comprising:

analyzing a representation of a first article of clothing to obtain color information, the color information comprising a dominant color region in the image, and pattern information, the pattern information defining artwork or aesthetic pattern of the first article of clothing;

comparing the color information and pattern information of the first article of clothing to color information and pattern information for a plurality of other articles of clothing using Bayesian probability analysis to determine matched pairs; and

providing a user with wardrobe suggestions using the matched pairs.

19. The method according to claim 18, further comprising:

receiving feedback from the user in response to the matched pairs;

updating the Bayesian probability analysis with the feedback; and

determining new matched pairs using the updated Bayesian probability analysis.

20. The method according to claim 18, further comprising receiving an event criteria and performing an initial rule based analysis of the plurality of other articles of clothing so as to exclude at least a portion of the plurality of other articles of clothing that are not appropriate for the event criteria, prior to performing the Bayesian probability analysis.