Automated decomposition of graphic designs

1. A computing device for decomposition of graphic designs into constituent parts comprising:

one or more processors;

a decomposition module coupled with the one or more processors, wherein operation of the decomposition module by the one or more processors causes the computing device to:

partition pixels of a graphic design image into superpixels based on an analysis of respective colors of individual pixels of the graphic design image;

group superpixels into components such that each component includes superpixels that are within a level of similarity from one another;

categorize a component as text of the graphic design based on at least one text identification feature indicating consistency of a plurality of characters within the superpixels of the text component, wherein the text component is comprised of grouped superpixels based on the level of similarity of the superpixels rather than a contiguous connection between the superpixels; and

a graphic design engine coupled with the one or more processors, wherein operation of the graphic design engine by the one or more processors enables the computing device to fulfill requests for suggested graphic designs from the database of decomposed graphic designs.

2. The computing device of claim 1 , wherein the text identification features include one or more of:

stroke width consistency within each component;

average stroke width within each component;

length of brush stroke within each component;

orientation of text within each component;

orientation consistency of each component;

contrast of each component with respect to a background of the graphic design image;

foreground color consistency within each component;

background color consistency within each component;

consistency of character sizes within each component;

depth of each component in the graphic design image; and

aspect ratio of each component.

3. The one or computing device of claim 2 , wherein to categorize of the component based on the text identification features, further comprises application of a plurality of rules to the text identification features, wherein each of the plurality of rules is a product of binary criteria applied to the text identification features.

4. The computing device of claim 3 , wherein individual rules of the plurality of rules are designed to identify one or more of:

text that is smaller than a first predefined threshold of size;

text that is larger than a second predefined threshold of size;

text that is below a predefined threshold of contrast with a color of the background of the graphic design image; and

connected text.

5. The computing device of claim 4 , wherein a weight for each rule is determined using a support vector machine (SVM).

6. The computing device of claim 4 , wherein the binary criteria is defined using covariance matrix adaptation (CMA).

7. The computing device of claim 4 , further comprising:

segmenting the text component by identifying individual pixels of the text component, the individual pixels precisely defining the plurality of characters within the superpixels of the text component.

8. A computer-implemented method for utilizing decomposed graphic designs comprising:

partitioning pixels of a graphic design image into superpixels based on an analysis of respective colors of individual pixels of the graphic design image;

grouping superpixels into components such that each component includes superpixels that are within a level of similarity from one another, wherein each component is grouped based on the level of similarity of the superpixels rather than a contiguous connection between the superpixels;

categorizing a component as text of the graphic design based on at least one text identification feature indicating consistency of a plurality of characters within the superpixels of the text component;

segmenting the text component by identifying individual pixels of the text component, the individual pixels precisely defining the plurality of characters within the superpixels of the text component; and

utilizing a graphic design engine coupled with one or more processors, to enable a computing device to fulfill requests for suggested graphic designs from a database of decomposed graphic designs, including the segmented text component.

9. One or more computer-readable storage media having executable instructions stored thereon, which, in response to execution by a processor of a computing device cause the computing device to:

partition pixels of a graphic design image into superpixels based on an analysis of respective colors of individual pixels of the graphic design image;

group superpixels into components such that each component includes superpixels that are within a level of similarity from one another, wherein each component is grouped based on the level of similarity of the superpixels rather than a contiguous connection between the superpixels;

categorize a component as text of the graphic design based on at least one text identification feature indicating consistency of a plurality of characters within the superpixels of the text component, wherein the text component is comprised of grouped superpixels based on the level of similarity of the superpixels; and

operate a graphic design engine coupled with the one or more processors, wherein the operation of the graphic design engine by the one or more processors enables the computing device to fulfill requests for suggested graphic designs from a database of decomposed graphic designs.