Method and device, and storage medium for processing an image

Patent metadata

Field	Value
Publication number	US-11216918-B2
Application number	US-202017036606-A
Country	US
Kind code	B2
Filing date	Sep 29, 2020
Priority date	Oct 11, 2019
Publication date	Jan 4, 2022
Grant date	Jan 4, 2022

Abstract

Disclosed are a method and device for processing an image, and a storage medium. The method includes: determining first position information of key points of first type of an object in the image based on the image and a trained model, where the first position information of each key point of first type indicates where each key point of first type is in the image; determining second position information of key points of second type of the object based on the first position information and a preset algorithm, where the second position information of each second type indicates where each key point of second type is in the image; liquefying the object based on the first position information and the second position information by using a liquefying level corresponding to a portion to be processed of the object.

First claim

What is claimed is:

1. A method for processing an image, comprising:
determining first position information of key points of first type of an object in the image based on the image and a trained model, wherein the first position information of each key point of first type indicates where each key point of first type is in the image;
determining second position information of key points of second type of the object based on the first position information and a preset algorithm, wherein the second position information of each second type indicates where each key point of second type is in the image; and
liquefying the object based on the first position information and the second position information by using a liquefying level corresponding to a portion to be processed of the object;
wherein said that determining second position information comprises:
determining each detail portion of the object;
determining first position information of key points of first type corresponding to each detail portion; and
determining second position information of key points of second type corresponding to each detail portion, based on the first position information of key points of first type corresponding to each detail portion and a preset algorithm corresponding to each detail portion.

2. The method according to claim 1 , wherein said that liquefying the object comprises:
determining position information of the portion to be processed, wherein the position information of the portion to be processed comprises first position information corresponding to the portion to be processed, or second position information corresponding to the portion to be processed;
determining the liquefying level corresponding to the portion to be processed, wherein the liquefying level is preset; and
liquefying pixels corresponding to the position information of the portion to be processed based on the liquefying level and a preset liquefying type.

3. The method according to claim 2 , wherein said that determining the liquefying level comprises:
determining, in the object, an associating portion of the portion to be processed and position information of the associating portion;
determining, in the object, an influencing portion of the portion to be processed and the associating portion;
determining position information of a first part of the influencing portion corresponding to the portion to be processed, and position information of a second part of the influencing portion corresponding to the associating portion;
determining a first distance between the first part and the portion to be processed based on the position information of the portion to be processed and the position information of the first part;
determining a second distance between the second part and the associating portion based on the position information of the associating portion and the position information of the second part;
acquiring a weight of the liquefying level corresponding to the portion to be processed by normalizing the first distance and the second distance;
determining the liquefying level corresponding to the portion to be processed by weighting a preset liquefying level corresponding to the portion to be processed based on the weight.

4. The method according to claim 1 , wherein the method further comprises:
weighing a sum of position information of a processing point in the image and position information of a same processing point in a previous image, wherein the previous image and the image are successive frames in a video, based on a preset weight, wherein the processing point is a key point of first type or a key point of second type; and
determining a weighted result as position information of the processing point in the image.

5. The method according to claim 1 , wherein the method further comprises:
selecting a preset quantity of successive frames following the image, in a case that a processing point exists in a previous image of the image and does not exist in the image, during the processing point being liquefied based on a preset liquefying level; and
liquefying images of the preset quantity of successive frames based on gradually diminished liquefying levels.

6. The method according to claim 1 , wherein said that liquefying the object comprises:
acquiring a resolution decreased image by decreasing a resolution of the image based on a preset factor;
liquefying the object in the resolution decreased image;
acquiring a processed image by restoring the resolution of the resolution decreased image after liquefaction.

7. An electronic device, comprising:
a processor; and
a memory for storing executable instructions of the processor,
wherein the processor is configured to execute the instructions to perform:
determining first position information of key points of first type of an object in the image based on the image and a trained model, wherein the first position information of each key point of first type indicates where each key point of first type is in the image;
determining second position information of key points of second type of the object based on the first position information and a preset algorithm, wherein the second position information of each second type indicates where each key point of second type is in the image; and
liquefying the object based on the first position information and the second position information by using a liquefying level corresponding to a portion to be processed of the object;
wherein said that determining second position information comprises:
determining each detail portion of the object;
determining first position information of key points of first type corresponding to each detail portion; and
determining second position information of key points of second type corresponding to each detail portion, based on the first position information of key points of first type corresponding to each detail portion and a preset algorithm corresponding to each detail portion.

8. The electronic device according to claim 7 , wherein said that liquefying the object comprises:
determining position information of the portion to be processed, wherein the position information of the portion to be processed comprises first position information corresponding to the portion to be processed, or second position information corresponding to the portion to be processed;
determining the liquefying level corresponding to the portion to be processed, wherein the liquefying level is preset; and
liquefying pixels corresponding to the position information of the portion to be processed based on the liquefying level and a preset liquefying type.

9. The electronic device according to claim 8 , wherein said that determining a liquefying level comprises:
determining, in the object, an associating portion of the portion to be processed and position information of the associating portion;
determining, in the object, an influencing portion of the portion to be processed and the associating portion;
determining position information of a first part of the influencing portion corresponding to the portion to be processed, and position information of a second part of the influencing portion corresponding to the associating portion;
determining a first distance between the first part and the portion to be processed based on the position information of the portion to be processed and the position information of the first part;
determining a second distance between the second part and the associating portion based on the position information of the associating portion and the position information of the second part;
acquiring a weight of the liquefying level corresponding to the portion to be processed by normalizing the first dis

Assignees

• Beijing Dajia Internet Information Tech Co Ltd

Inventors

• Yan Xin

Classifications

• G06T7/73

  using feature-based methods · CPC title

• G06T2207/30196

  Human being; Person · CPC title

• G06T7/75 Primary

  involving models · CPC title

• G06T2207/20081

  Training; Learning · CPC title

• G06T3/40

  Scaling of whole images or parts thereof, e.g. expanding or contracting · CPC title