Granted Patent B2

US 12,579,734 · App. 18/641,421 · Granted Mar 17, 2026

Inventor: Xuan Sun (Beijing, CN)

Assignee: BOE Technology Group Co., Ltd.

G06T15/20G06T5/50G06T5/60G06T2207/20084G06T2207/20221

Abstract

Claims (63)

1 . A method for rendering viewpoints, comprising:

acquiring a first initial feature of a two-dimensional image and a second initial feature of a depth image corresponding to the two-dimensional image by performing initial feature extraction on the two-dimensional image and the depth image;

acquiring a first initial dimension reduction feature by splicing the first initial feature and the second initial feature in a channel dimension and performing channel dimension reduction;

acquiring a fusion feature by performing image distortion and restoration on the first initial dimension reduction feature for multiple times; and

generating a plurality of viewpoint images by performing fusion and channel dimension reduction on the fusion feature;

wherein after generating the plurality of viewpoint images by performing fusion and channel dimension reduction on the fusion feature, the method further comprises;

acquiring a third initial feature and a fourth initial feature by performing initial feature extraction on the viewpoint image and a corresponding label image;

acquiring a second initial dimension reduction feature and a third initial dimension reduction feature by performing width and height dimension down-sampling on the third initial feature and the fourth initial feature, reducing the width and height dimensions to 1, and reducing the channel dimension to 1; and

discriminating the second initial dimension reduction feature and the third initial dimension reduction feature based on the label image.

2 . The method for rendering viewpoints according to claim 1 , wherein after generating the plurality of viewpoint images by performing fusion and channel dimension reduction on the fusion feature, the method further comprises:

acquiring a fifth initial feature and a sixth initial feature by performing initial feature extraction on the viewpoint image and a corresponding label image;

acquiring a fourth initial dimension reduction feature and a fifth initial dimension reduction feature by performing width and height dimension down-sampling on the fifth initial feature and the sixth initial feature and reducing the channel dimension to 1; and

discriminating the fourth initial dimension reduction feature and the fifth initial dimension reduction feature based on the label image.

3 . The method for rendering viewpoints according to claim 2 , wherein a number of groups of the fourth initial dimension reduction feature and the fifth initial dimension reduction feature is a plurality of groups; wherein

each group of the fourth initial dimension reduction feature and the fifth initial dimension reduction feature corresponds to discriminated pixel blocks from large to small.

4 . The method for rendering viewpoints according to claim 1 , wherein before acquiring the first initial feature of the two-dimensional image and the second initial feature of the depth image corresponding to the two-dimensional image by performing initial feature extraction on the two-dimensional image and the depth image, the method further comprises:

generating the depth image corresponding to the two-dimensional image by inputting the two-dimensional image into a monocular depth estimation model.

5 . The method for rendering viewpoints according to claim 1 , wherein after generating the plurality of viewpoint images by performing fusion and channel dimension reduction on the fusion feature, the method further comprises:

acquiring a synthesized viewpoint image by splicing the plurality of the viewpoint images in a width dimension; and

generating a three-dimensional image by interleaving the synthesized viewpoint image.

6 . An electronic device, comprising:

at least one processor; and

a memory stores one or more computer programs executable by the at least one processor,

wherein the at least one processor, when loading and executing the one or more computer programs, is caused to perform:

acquiring a first initial feature of a two-dimensional image and a second initial feature of a depth image corresponding to the two-dimensional image by performing initial feature extraction on the two-dimensional image and the depth image;

acquiring a first initial dimension reduction feature by splicing the first initial feature and the second initial feature in a channel dimension and performing channel dimension reduction;

acquiring a fusion feature by performing image distortion and restoration on the first initial dimension reduction feature for multiple times; and

generating a plurality of viewpoint images by performing fusion and channel dimension reduction on the fusion feature;

wherein the at least one processor, when loading and executing the one or more computer programs, is caused to perform:

acquiring a third initial feature and a fourth initial feature by performing initial feature extraction on the viewpoint image and a corresponding label image;

acquiring a second initial dimension reduction feature and a third initial dimension reduction feature by performing width and height dimension down-sampling on the third initial feature and the fourth initial feature, reducing the width and height dimensions to 1, and reducing the channel dimension to 1; and

discriminating the second initial dimension reduction feature and the third initial dimension reduction feature based on the label image.

7 . The electronic device according to claim 6 , wherein the at least one processor, when loading and executing the one or more computer programs, is caused to perform:

acquiring a fifth initial feature and a sixth initial feature by performing initial feature extraction on the viewpoint image and a corresponding label image;

acquiring a fourth initial dimension reduction feature and a fifth initial dimension reduction feature by performing width and height dimension down-sampling on the fifth initial feature and the sixth initial feature and reducing the channel dimension to 1; and

discriminating the fourth initial dimension reduction feature and the fifth initial dimension reduction feature based on the label image.

8 . The electronic device according to claim 7 , wherein a number of groups of the fourth initial dimension reduction feature and the fifth initial dimension reduction feature is a plurality of groups; wherein

each group of the fourth initial dimension reduction feature and the fifth initial dimension reduction feature corresponds to discriminated pixel blocks from large to small.

9 . The electronic device according to claim 6 , wherein the at least one processor, when loading and executing the one or more computer programs, is caused to perform:

generating the depth image corresponding to the two-dimensional image by inputting the two-dimensional image into a monocular depth estimation model.

10 . The electronic device according to claim 6 , wherein the at least one processor, when loading and executing the one or more computer programs, is caused to perform:

acquiring a synthesized viewpoint image by splicing the plurality of the viewpoint images in a width dimension; and

generating a three-dimensional image by interleaving the synthesized viewpoint image.

11 . A non-transitory computer-readable storage medium storing at least one computer program therein, wherein the at least one computer program, when executed by a processor, causes the processor to perform:

acquiring a first initial feature of a two-dimensional image and a second initial feature of a depth image corresponding to the two-dimensional image by performing initial feature extraction on the two-dimensional image and the depth image;

acquiring a first initial dimension reduction feature by splicing the first initial feature and the second initial feature in a channel dimension and performing channel dimension reduction;

acquiring a fusion feature by performing image distortion and restoration on the first initial dimension reduction feature for multiple times; and

generating a plurality of viewpoint images by performing fusion and channel dimension reduction on the fusion feature;

wherein the at least one computer program, when executed by a processor, causes the processor to perform:

acquiring a third initial feature and a fourth initial feature by performing initial feature extraction on the viewpoint image and a corresponding label image;

acquiring a second initial dimension reduction feature and a third initial dimension reduction feature by performing width and height dimension down-sampling on the third initial feature and the fourth initial feature, reducing the width and height dimensions to 1, and reducing the channel dimension to 1; and

discriminating the second initial dimension reduction feature and the third initial dimension reduction feature based on the label image.

12 . The non-transitory computer-readable storage medium according to claim 11 , wherein the at least one computer program, when executed by a processor, causes the processor to perform:

acquiring a fifth initial feature and a sixth initial feature by performing initial feature extraction on the viewpoint image and a corresponding label image;

acquiring a fourth initial dimension reduction feature and a fifth initial dimension reduction feature by performing width and height dimension down-sampling on the fifth initial feature and the sixth initial feature and reducing the channel dimension to 1; and

discriminating the fourth initial dimension reduction feature and the fifth initial dimension reduction feature based on the label image.

13 . The non-transitory computer-readable storage medium according to claim 12 , wherein a number of groups of the fourth initial dimension reduction feature and the fifth initial dimension reduction feature is a plurality of groups; wherein

each group of the fourth initial dimension reduction feature and the fifth initial dimension reduction feature corresponds to discriminated pixel blocks from large to small.

14 . The non-transitory computer-readable storage medium according to claim 11 , wherein the at least one computer program, when executed by a processor, causes the processor to perform:

generating the depth image corresponding to the two-dimensional image by inputting the two-dimensional image into a monocular depth estimation model.

15 . The non-transitory computer-readable storage medium according to claim 11 , wherein the at least one computer program, when executed by a processor, causes the processor to perform:

acquiring a synthesized viewpoint image by splicing the plurality of the viewpoint images in a width dimension; and

generating a three-dimensional image by interleaving the synthesized viewpoint image.

Assignments (1)

Continuity (2)

References Cited (26)