Siding Identification Systems and Methods

• US 20140023280A1
• Filed: 07/23/2012
• Published: 01/23/2014
• Est. Priority Date: 07/23/2012
• Status: Active Grant

• Current Assignee: State Farm Mutual Automobile Insurance Company (State Farm)
• Original Assignee: State Farm Insurance (State Farm)
• Inventors: Lin, Kai-Hsiang, Grant, Rosemarie Geier, Mast, Joshua M., Dewey, Douglas L., Dagli, Charlie K.
• Granted Patent: US 8,718,385 B2
• Time in Patent Office: Days
• US Class Current: 382/218
• CPC Class Codes: G06F 16/5862 using texture
• CPC Class Codes: G06F 18/00 Pattern recognition
• CPC Class Codes: G06V 10/54 relating to texture

Abstract

A method for identifying siding includes receiving particular texture data associated with a physical texture of a particular siding sample, accessing a reference database including reference texture data associated with physical texture of multiple different reference siding samples, performing an automated texture data analysis by analyzing the particular texture data associated with the particular siding sample and the reference texture data associated with different reference siding samples, and automatically identifying at least one reference siding sample that matches the particular siding sample based at least on the automated texture data analysis. The particular texture data associated with the physical texture of the particular siding sample may comprise a digital image, such as a photographic or scanned image of the particular siding sample, or a photographic or scanned image of an ink image physically transferred from the particular siding sample onto a substrate.

Claims

1. A method for identifying siding, comprising:

receiving particular texture data associated with a physical texture of a particular siding sample;

receiving via user input siding characteristic data of the particular siding sample;

cropping the particular texture data into a plurality of patches;

extracting quantified visual features from one of the plurality of patches;

accessing a reference database including, for each of multiple different reference siding samples, both (a) reference quantified visual features associated with a physical texture of the respective reference siding sample and (b) reference siding characteristic data for the respective reference siding sample;

performing an automated texture data analysis by analyzing the quantified visual features associated with the particular siding sample and the reference visual features associated with different reference siding samples;

performing an automated comparison of the user-input siding characteristic data of the particular siding sample with the reference siding characteristic data associated with different reference siding samples; and

automatically identifying at least one reference siding sample that matches the particular siding sample based at least on (a) the automated texture data analysis and (b) the automated siding characteristic data comparison.

3. A method according to claim 1, wherein the siding characteristic data of the particular siding sample includes at least one of siding material, siding style, a siding face size characteristic, a siding profile characteristic, and siding manufacturer.

4. A method according to claim 1, wherein analyzing the particular texture data associated with the particular siding sample and the reference texture data associated with at least a subset of the reference siding samples comprises:

automatically analyzing the particular texture data associated with the particular siding sample to quantify one or more textural features of the particular siding sample; and

automatically comparing the one or more quantified visual features of the particular siding sample with quantified visual features of different reference siding samples.

5. A method according to claim 1, further comprising displaying on a display device an indication of the at least one matching reference siding sample identified as matching the particular siding sample.

6. A method according to claim 1, wherein the particular texture data associated with the physical texture of the particular siding sample comprises a digital image.

7. A method according to claim 1, wherein the particular texture data associated with the physical texture of the particular siding sample comprises digital data representative of the physical texture of the particular siding sample.

8. A method according to claim 6, wherein the digital image comprises a photographic image of the particular siding sample.

9. A method according to claim 6, wherein the digital image comprises a scanned image of the particular siding sample.

10. A method according to claim 6, wherein the digital image comprises a digital image of a transferred ink image of the particular siding sample.

11. A method according to claim 10, wherein the transferred ink image of the particular siding sample is produced on an image substrate by applying ink to the particular siding sample and applying the image substrate to the inked particular siding sample to transfer an ink image from the particular siding sample to the image substrate.

12. A method according to claim 6, wherein the digital image comprises a photographic or scanned image of the particular siding sample having ink applied thereto.

13. A method according to claim 1, further comprising:

generating a transferred ink image of the particular siding sample by applying ink to the particular siding sample and applying an image substrate to the inked particular siding sample to transfer an ink image from the particular siding sample to the image substrate; and

generating a digital image of transferred ink image.

17. A method according to claim 1, wherein automatically identifying at least one reference siding sample that matches the particular siding sample comprises:

determining a similarity score for each of a plurality of reference siding samples, wherein the similarity score indicates a level of similarity; and

identifying at least one reference siding sample that matches the particular siding sample based on the respective similarity scores of the plurality of reference siding samples.

18. A method according to claim 17, wherein identifying at least one reference siding sample that matches the particular siding sample comprises identifying each reference siding sample having a respective similarity score that exceeds a predefined threshold score.

19. A method according to claim 17, wherein identifying at least one reference siding sample that matches the particular siding sample comprises identifying N reference siding samples having the highest similarity scores, where N is a predefined number.

2. (canceled)

14. A method for identifying siding, comprising:

generating a transferred ink image of a siding sample by applying ink to the siding sample and applying an image substrate to the inked siding sample to transfer the ink from the siding sample to the image substrate;

digitizing the transferred ink image;

cropping the digitized transferred ink image into a plurality of patches;

extracting quantified visual features from one of the plurality of patches;

receiving via user input siding characteristic data of the siding sample;

performing an automated comparison of the user-input siding characteristic data of the siding sample with the reference siding characteristic data including reference quantified visual features associated with different reference siding samples; and

automatically identifying from a plurality of reference siding samples at least one reference siding sample that matches the siding sample based at least on (a) the analysis of the digitized transferred ink image and (b) the automated siding characteristic data comparison.

15. A system for identifying siding, comprising:

a processor;

a non-transitory storage medium;

a display device; and

a set of computer readable instructions stored in the non-transitory storage medium and when executed by the processor configured to;

receive particular texture data associated with a physical texture of a particular siding sample;

crop the particular texture data into a plurality of patches;

extract quantified visual features from one of the plurality of patches;

access a reference database including, for each of multiple different reference siding samples, reference quantified visual features associated with a physical texture of the respective reference siding sample;

perform an automated texture data analysis by analyzing the particular extracted quantified visual features associated with the particular siding sample and the reference quantified visual features associated with different reference siding samples;

receive via user input siding characteristic data of the particular siding sample;

performing an automated comparison of the user-input siding characteristic data of the siding sample with stored reference siding characteristic data associated with different reference siding samples; and

based at least on (a) the automated texture data analysis and (b) the automated siding characteristic data comparison, identify at least one of the reference siding samples as a matching reference siding sample that matches the particular siding sample; and

cause an indication of the at least one matching reference siding samples to be displayed on the display device.

16. A system for identifying siding, comprising:

a digital image generating device programmed to digitize a transferred ink image of a particular siding sample, the transferred ink image being generated by applying ink to the particular siding sample and applying an image substrate to the inked particular siding sample to transfer an ink image from the particular siding sample to the image substrate; and

a siding analysis engine comprising a processor and a set of computer readable instructions stored in non-transitory storage media and executable by the processor to;

crop the particular texture data into a plurality of patches;

extract quantified visual features from one of the plurality of patches;

identify from a plurality of reference siding samples at least one reference siding sample that matches the particular siding sample based at least on the analysis of the extracted quantified visual features;

receive via user input siding characteristic data of the siding sample;

perform an automated comparison of the user-input siding characteristic data of the siding sample with stored reference siding characteristic data associated with different reference siding samples; and

identify from a plurality of reference siding samples at least one reference siding sample that matches the siding sample based at least on (a) the analysis of the extracted quantified visual features and (b) the automated siding characteristic data comparison.