Granted Patent B2

US 12,639,942 · App. 18/653,592 · Granted May 26, 2026

Inventors: Xuchang Huangfu (Beijing, CN); Yuanyi Xue (Alameda, CA); Wenhao Zhang (Beijing, CN); Yang Zhang (Dübendorf, CH); Chen Liu (Beijing, CN); Xuewei Meng (Beijing, CN)

Assignees: Disney Enterprises, Inc.; Beijing YoJaJa Software Technology Development Co., Ltd.

G06V10/993G06T11/001G06V10/22G06V10/273G06V10/36G06V10/54G06V10/761G06V10/764

Abstract

Claims (87)

1 . A method comprising:

receiving an image to analyze for artifacts;

determining texture information that characterizes texture in the image;

merging the texture information with the image, wherein the texture information is used to focus an analysis of artifacts in regions of the image, wherein merging the texture information with the image comprises:

segmenting the image into a plurality of regions based on object detection of objects in respective regions;

determining a difference in a characteristic of pixels in respective regions in the plurality of regions; and

using the difference of the characteristic in respective regions to classify respective regions with a classification in a plurality of classifications, wherein using the difference of the characteristic to classify respective regions comprises:

determining a category for pixels in a region;

counting a first number of pixels in a first classification and a second number of pixels in a second classification; and

determining whether the region is associated with the first classification or the second classification based on the first number of pixels and the second number of pixels, wherein:

the region is associated with the first classification when the first number of pixels is less than a first threshold and the second number of pixels is greater than a second threshold, and

the first threshold is based on a number of pixels that are classified as artifact pixels and the second threshold is based on a number of pixels that are categorized as texture pixels; and

outputting a score based on the processing of the image that assesses the artifacts in the image.

2 . The method of claim 1 , wherein determining texture information comprises:

analyzing the image to extract characteristics of texture in the image; and

generating a texture map using the extracted characteristics of texture.

3 . The method of claim 1 , wherein determining texture information comprises:

determining entropy of pixels in the image, gradients of pixels in the image, or a local pixel value difference for pixels in the image as the texture information.

4 . The method of claim 1 , wherein merging the texture information with the image comprises:

analyzing the texture information to determine the plurality of regions in the image; and

classifying regions in the plurality of regions into the plurality of classifications based on respective texture information for the respective regions.

5 . The method of claim 4 , wherein merging the texture information with the image comprises:

masking one or more regions in the image based on a classification of the one or more regions, wherein masking filters the one or more regions in the image.

6 . The method of claim 5 , wherein filtering the one or more regions comprises adjusting pixel values of the one or more regions.

7 . The method of claim 4 , wherein merging the texture information with the image comprises:

inputting the plurality of classifications into a prediction network, wherein the prediction network uses respective classifications of regions in the plurality of regions to filter regions in the image.

8 . The method of claim 1 , wherein:

a first classification indicates a region does not have perceptible artifacts, and

a second classification indicates a region does have perceptible artifacts.

9 . The method of claim 1 , wherein merging the texture information with the image comprises:

determining an entropy of pixels in the image; and

analyzing the entropy of pixels based on a threshold to classify the pixels in a first classification or a second classification, wherein pixels values are adjusted when classified in the second classification.

10 . The method of claim 1 , wherein merging the texture information with the image comprises:

inputting the image into one or more channels of a prediction network;

inputting the texture information as an auxiliary channel into the prediction network; and

processing the one or more channels and the auxiliary channel to generate the score.

11 . The method of claim 10 , wherein:

the auxiliary channel and the one or more channels are combined to generate combined channels, and

the combined channels are analyzed to generate the score.

12 . The method of claim 10 , wherein:

the texture information is combined with the image and input into the auxiliary channel.

13 . The method of claim 10 , further comprising:

applying attention or weighting based on the texture information to pixels of the image; and

generating the score based on the attention or weighting that is applied.

14 . A non-transitory computer-readable storage medium having stored thereon computer executable instructions, which when executed by a computing device, cause the computing device to be operable for:

receiving an image to analyze for artifacts;

determining texture information that characterizes texture in the image;

merging the texture information with the image, wherein the texture information is used to focus an analysis of artifacts in regions of the image, wherein merging the texture information with the image comprises:

segmenting the image into a plurality of regions based on object detection of objects in respective regions;

determining a difference in a characteristic of pixels in respective regions in the plurality of regions; and

using the difference of the characteristic in respective regions to classify respective regions with a classification in a plurality of classifications, wherein using the difference of the characteristic to classify respective regions comprises;

determining a category for pixels in a region;

counting a first number of pixels in a first classification and a second number of pixels in a second classification; and

determining whether the region is associated with the first classification or the second classification based on the first number of pixels and the second number of pixels, wherein:

the region is associated with the first classification when the first number of pixels is less than a first threshold and the second number of pixels is greater than a second threshold, and

the first threshold is based on a number of pixels that are classified as artifact pixels and the second threshold is based on a number of pixels that are categorized as texture pixels; and

outputting a score based on the processing of the image that assesses the artifacts in the image.

15 . The non-transitory computer-readable storage medium of claim 14 , wherein merging the texture information with the image comprises:

analyzing the texture information to determine the plurality of regions in the image; and

classifying regions in the plurality of regions into the plurality of classifications based on respective texture information for the respective regions.

16 . The non-transitory computer-readable storage medium of claim 14 , wherein merging the texture information with the image comprises:

inputting the image into one or more channels of a prediction network;

inputting the texture information as an auxiliary channel into the prediction network; and

processing the one or more channels and the auxiliary channel to generate the score.

17 . An apparatus comprising:

one or more computer processors; and

a computer-readable storage medium comprising instructions for controlling the one or more computer processors to be operable for:

receiving an image to analyze for artifacts;

determining texture information that characterizes texture in the image;

merging the texture information with the image, wherein the texture information is used to focus an analysis of artifacts in regions of the image, wherein merging the texture information with the image comprises:

segmenting the image into a plurality of regions based on object detection of objects in respective regions;

determining a difference in a characteristic of pixels in respective regions in the plurality of regions; and

using the difference of the characteristic in respective regions to classify respective regions with a classification in a plurality of classifications, wherein using the difference of the characteristic to classify respective regions comprises:

determining a category for pixels in a region;

counting a first number of pixels in a first classification and a second number of pixels in a second classification; and

determining whether the region is associated with the first classification or the second classification based on the first number of pixels and the second number of pixels, wherein:

the region is associated with the first classification when the first number of pixels is less than a first threshold and the second number of pixels is greater than a second threshold, and

the first threshold is based on a number of pixels that are classified as artifact pixels and the second threshold is based on a number of pixels that are categorized as texture pixels; and

outputting a score based on the processing of the image that assesses the artifacts in the image.

18 . The non-transitory computer-readable storage medium of claim 16 , wherein:

the auxiliary channel and the one or more channels are combined to generate combined channels, and

the combined channels are analyzed to generate the score.

19 . The non-transitory computer-readable storage medium of claim 16 , wherein:

the texture information is combined with the image and input into the auxiliary channel.

20 . The non-transitory computer-readable storage medium of claim 16 , further operable for:

applying attention or weighting based on the texture information to pixels of the image; and

generating the score based on the attention or weighting that is applied.

Assignments (5)

Continuity (1)

References Cited (50)