Composite image quality assurance

• US 9,898,812 B1
• Filed: 12/09/2015
• Issued: 02/20/2018
• Est. Priority Date: 12/09/2015
• Status: Active Grant

• Current Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Original Assignee: Amazon Technologies, Inc. (Amazon.com, Inc.)
• Inventors: Padfield, Dirk Ryan
• Primary Examiner(s): Yang, Weiwen
• Application Number:
  US14/964,284
• Time in Patent Office: 804 Days
• Field of Search: None
• CPC Class Codes: G06T 2207/10024 Color image
• CPC Class Codes: G06T 2207/20221 Image fusion; Image merging
• CPC Class Codes: G06T 2207/30124 Fabrics; Textile; Paper
• CPC Class Codes: G06T 5/50 using two or more images, e.g. averaging or subtraction
• CPC Class Codes: G06T 7/0004 Industrial image inspection

Abstract

Features are disclosed for processing composite images. Composite images may be received that include a common item such as a t-shirt with different graphics overlaid on the item. Features for determining the quality of composite images based on processing the image data are provided. Detection of a region that the overlaid graphic covers provides a targeted location for analyzing the underlying image. A quality metric may be determined based on whether, which, and how many features of the item shown in the underlying image are obscured or otherwise modified by the overlaid image.

Claims

1. A system comprising:

a data store that stores a first composite image previously transferred to the system and a second composite image previously transferred to the system, the first composite image comprising;

a first underlying image region showing a physical item, and

a first overlaid image region showing a first graphic overlay obscuring at least a first portion of the physical item,

the second composite image comprising;

a second underlying region showing the physical item, and

a second overlaid image region showing a second graphic overlay obscuring at least a second portion of the physical item; and

a computing device in communication with the data store, the computing device having a processor configured with computer-executable instructions, that when executed by the processor, cause the computing device to at least;

receive an input image, the input image comprising;

a third underlying image region showing the physical item, and

a third overlaid image region showing a third graphic overlay obscuring at least a third portion of the physical item;

generate, on a pixel-by-pixel basis, for each pixel location, a mode color value of pixels at a respective pixel location in the input image, the first composite image, and the second composite image; and

generate a reconstructed underlying image using the mode color values for respective pixel locations in the input image, the first composite image, and the second composite image, wherein the reconstructed underlying image does not include the first graphic overlay, the second graphic overlay, or the third graphic overlay, and wherein the reconstructed underlying image shows the at least first portion of the physical item, the at least second portion of the physical item, and the at least third portion of the physical item;

identify a region of the input image obscured by the third overlaid image region; and

identify a set of features of the physical item shown in the region of the reconstructed underlying image using an image feature detection model that detects image features included in an image; and

generate a feature detection result using the set of features of the physical item, wherein the detection result indicates a type of image feature identified and a quantity of features identified for the type.

2. The system of claim 1, wherein the set of image features is identified from a change in color values between a first pixel of the reconstructed underlying image and a second pixel of the reconstructed underlying image,

wherein the change satisfies a threshold and the first pixel and the second pixel are separated by at least a threshold distance, and

wherein an image feature related to the change is included in the set of image features.

3. The system of claim 1, wherein the processor is configured with computer-executable instructions, that when executed by the processor, further cause the computing device to at least:

receive an image authorization rule identifying;

a target feature detection result including a target type of image feature identified and a threshold quantity of features identified for the target type, and

an authorization determination for composite images having the target feature detection result;

compare the target feature detection result to the feature detection result to determine that the feature detection result includes the target type of image feature and at least the threshold quantity of features for the target type; and

provide the authorization determination for the input image.

4. A device comprising:

a computer-readable memory that stores specific computer-executable instructions; and

a processor in communication with the computer-readable memory, wherein the processor executes the specific computer-executable instructions to cause the device to at least;

receive a composite image, the composite image comprising;

an underlying image region showing a physical item, and

an overlaid image region showing a graphic overlay obscuring at least a portion of the physical item;

receive an image showing the physical item, wherein the image shows the at least the portion of the physical item obscured by the graphic overlay;

identify a region of the composite image obscured by the overlaid image region;

detect an image feature of the physical item shown in the image within the region; and

generate a composite image quality metric based at least partly on the image feature detected.

5. The device of claim 4, wherein the region of the composite image covered by the overlaid image region is identified from a shape in the overlaid image region indicated by a binary mask, wherein the region of the composite image covered by the overlaid image region includes the shape.

6. The device of claim 4, wherein the processor executes the specific computer-executable instructions to further cause the device to at least:

receive an image feature recognition model that identifies an image feature of the physical item shown in an image using pixel information from the image, wherein the image feature is detected using the image feature recognition model and pixel information from the underlying image within the region.

7. The device of claim 4, wherein the processor executes the specific computer-executable instructions to further cause the device to at least:

generate a detection result for the image features, the detection result indicating a type of image feature detected and a confidence value for the image feature detected, wherein generating the quality metric includes weighting an image feature using the confidence value for the image feature.

8. The device of claim 4, wherein the image features includes an edge image feature,

wherein the edge image feature is identified from a change in color values between a first pixel within the region of the underlying image and a second pixel within the region of the underlying image, and

wherein the change satisfies a threshold and the first pixel and the second pixel are separated by at least a threshold distance.

9. The device of claim 8, wherein the threshold is identified from a characteristic of the physical item shown in the underlying image, and wherein the threshold distance is identified from the characteristic of the physical item shown in the underlying image.

10. The device of claim 4, wherein the processor executes the specific computer-executable instructions to further cause the device to at least:

receive an image authorization rule identifying;

a target feature detection result including a target type of image feature identified and a threshold quantity of features identified for the target type, and

an authorization determination for composite images having the target feature detection result;

compare the target feature detection result to the feature detection result to determine that the feature detection result includes the target type of image feature and at least the threshold quantity of features for the target type; and

provide the authorization determination for the composite image.

11. The device of claim 4, wherein the processor further executes the specific computer-executable instructions to further cause the device to at least:

identify a set of images in which each image of the set of images has an underlying image region showing the physical item; and

generate the underlying image using pixel color values from the set of images and the composite image, wherein the underlying image does not include the graphic overlay shown in the overlaid image region, and wherein the underlying image shows the at least the portion of the physical item.

12. A computer-implemented method comprising:

under control of a computing device executing specific computer-executable instructions,

receiving a composite image, the composite image comprising;

an underlying image region showing a physical item, and

an overlaid image region showing a graphic overlay obscuring at least a portion of the physical item;

receiving an image showing the physical item, wherein the image shows the at least the portion of the physical item obscured by the graphic overlay;

identifying a region of the composite image obscured by the overlaid image region;

detecting an image feature of the physical item shown in the image within the region; and

generating a composite image quality metric based at least in part on the image feature detected, wherein the composite image quality metric indicates how well the composite image shows the physical item.

13. The computer-implemented method of claim 12, further comprising identifying a shape in the overlaid image, the shape including an anchor point and at least one dimension of the shape, wherein the anchor point indicates a pixel location within the composite image where the shape begins, and wherein the region of the composite image covered by the overlaid image region includes the shape.

14. The computer-implemented method of claim 12, further comprising receiving an image feature recognition model that identifies an image feature of the physical item shown in an image using pixel information from the image, wherein the image feature of the physical item shown in the image is detected using the image feature recognition model and pixel information from the underlying image within the region.

15. The computer-implemented method of claim 12, further comprising generating a detection result for the image features, the detection result indicating a type of image feature detected and a confidence value for the image feature detected, wherein generating the quality metric includes weighting an image feature using the confidence value for the image feature.

16. The computer-implemented method of claim 12, wherein the image feature includes an edge image feature, and

wherein the computer-implemented method further comprises detecting the edge image feature from a change in color values between a first pixel within the region of the underlying image and a second pixel within the region of the underlying image, wherein the change satisfies a threshold and the first pixel and the second pixel are separated by at least a threshold distance.

17. The computer-implemented method of claim 16, further comprising:

identifying the threshold using a characteristic of the physical item shown in the underlying image; and

identifying the threshold distance using the characteristic of the physical item shown in the underlying image.

18. The computer-implemented method of claim 12, further comprising:

receiving an image authorization rule identifying;

a target feature detection result including a target type of image feature identified and a threshold quantity of features identified for the target type, and

an authorization determination for composite images having the target feature detection result;

comparing the target feature detection result to the feature detection result to determine that the feature detection result includes the target type of image feature and at least the threshold quantity of features for the target type; and

providing the authorization determination for the composite image.

19. The computer-implemented method of claim 12, further comprising:

identifying a set of images in which each image of the set of images has an underlying image region showing the physical item; and

generating the underlying image using pixel color values from the set of images and the composite image, wherein the underlying image does not include the graphic overlay shown in the overlaid image region.

20. The computer-implemented method of claim 12, further comprising transmitting a message to a communication device associated with an entity that generated the composite image, where the transmitting of the message is based at least partly upon the composite image quality metric, wherein the message activates an image processing application to cause the message to display composite image content via the communication device.