Method for obtaining multi-dimensional information by picture-based integration and related device

Patent metadata

Field	Value
Publication number	US-2022084314-A1
Application number	US-202117536774-A
Country	US
Kind code	A1
Filing date	Nov 29, 2021
Priority date	Dec 30, 2019
Publication date	Mar 17, 2022
Grant date	—

Abstract

The disclosure provides a method for obtaining multi-dimensional information by picture-based integration and a related device. The method includes the following operations. A to-be-detected picture is acquired. The to-be-detected picture is detected and multiple pieces of feature information are extracted from the to-be-detected picture. Target feature information and associated feature information are selected from the multiple pieces of feature information, and the target feature information is associated to the associated feature information to generate multi-dimensional information. The multi-dimensional information includes multiple pieces of feature information associated with each other.

First claim

What is claimed is:

1 . A method for obtaining multi-dimensional information by picture-based integration, comprising:
acquiring a to-be-detected picture; detecting the to-be-detected picture and extracting a plurality of pieces of feature information from the to-be-detected picture; and selecting, from the plurality of pieces of feature information, target feature information and associated feature information, and associating the target feature information to the associated feature information to generate multi-dimensional information, wherein the multi-dimensional information comprises a plurality of pieces of feature information associated with each other.

2 . The method of claim 1 , wherein the plurality of pieces of feature information extracted comprise at least two different types of feature information of the following:
human face feature information, human body feature information, or vehicle feature information.

3 . The method of claim 2 , further comprising after selecting, from the plurality of pieces of feature information, the target feature information and the associated feature information, and associating the target feature information to the associated feature information to generate the multi-dimensional information:
retrieving a first target image from a first database based on the target feature information in the multi-dimensional information; retrieving a second target image from a second database based on the associated feature information in the multi-dimensional information; and determining the first target image and the second target image as retrieval result pictures of the to-be-detected picture.

4 . The method of claim 1 , wherein the to-be-detected picture comprises one or more to-be-detected pictures; and
detecting the to-be-detected picture and extracting the plurality of pieces of feature information from the to-be-detected picture comprises: detecting the one or more to-be-detected pictures, and extracting the plurality of pieces of feature information from the one or more to-be-detected pictures.

5 . The method of claim 2 , wherein selecting, from the plurality of pieces of feature information, the target feature information and the associated feature information, and associating the target feature information to the associated feature information to generate the multi-dimensional information comprises:
selecting target human face feature information corresponding to a target human face with a highest quality score in the to-be-detected picture as the target feature information, and selecting at least one of the following as the associated feature information:
target human body feature information corresponding to the target human face feature information, or
target vehicle feature information corresponding to a vehicle closest to a center point of the target human face; and
associating the target feature information to the associated feature information to generate the multi-dimensional information, wherein the multi-dimensional information comprises at least two different types of feature information of the following:
the target human face feature information,
the target human body feature information, or
the target vehicle feature information.

6 . The method of claim 2 , wherein selecting, from the plurality of pieces of feature information, the target feature information and the associated feature information, and associating the target feature information to the associated feature information to generate the multi-dimensional information comprises:
receiving a control instruction, and selecting, based on the control instruction, the target feature information from the plurality of pieces of feature information, wherein the target feature information is one of: the human face feature information, the human body feature information, or the vehicle feature information; selecting, according to the selected target feature information, associated feature information matching with the target feature information, wherein the associated feature information matching with the target feature information comprises at least one type of feature information of the following other than a type of the target feature information: the human face feature information, the human body feature information, or the vehicle feature information; and associating the target feature information to the associated feature information matching with the target feature information to generate the multi-dimensional information.

7 . The method of claim 6 , wherein
in response to that the selected target feature information is target human face feature information in the human face feature information, the selecting, according to the selected target feature information, the associated feature information matching with the target feature information comprises: automatically selecting, according to the selected target human face feature information, at least one of the following as the associated feature information: human body feature information corresponding to the target human face feature information, or vehicle feature information corresponding to a vehicle closest to a center point of a human face associated with the target human face feature information; or in response to that the selected target feature information is target human body feature information in the human body feature information, the selecting, according to the selected target feature information, the associated feature information matching with the target feature information comprises: automatically selecting, according to the selected target human body feature information, at least one of the following as the associated feature information: human face feature information corresponding to the target human body feature information, or vehicle feature information corresponding to a vehicle closest to a center point of a human body associated with the target human body feature information; or in response to that the selected target feature information is target vehicle feature information in the vehicle feature information, the selecting, according to the selected target feature information, the associated feature information matching with the target feature information comprises: automatically selecting, according to the selected target vehicle feature information, at least one of the following as the associated feature information: human face feature information corresponding to a human face closest to a center point of a vehicle associated with the target vehicle feature information, or human body feature information corresponding to the target vehicle feature information.

8 . The method of claim 2 , wherein selecting, from the plurality of pieces of feature information, the target feature information and the associated feature information, and associating the target feature information to the associated feature information to generate the multi-dimensional information comprises:
receiving a control instruction, and selecting, based on the control instruction, the target feature information and the associated feature information from the plurality of pieces of feature information, wherein the selected target feature information and the selected associated feature information comprise at least two different types of feature information of the following: the human face feature information, the human body feature information, or the vehicle feature information; and associating the target feature information to the associated feature information to generate the multi-dimensional information.

9 . The method of claim 3 , further comprising after determining the first target image and the second target image as the retr

Assignees

• Shenzhen Sensetime Technology Co Ltd

Inventors

• Huang Xiaoying
• Fu Hao
• Zhang Guiming
• Zhang Min
• Gao Huixuan

Classifications

• G06V40/168

  Feature extraction; Face representation · CPC title

• G06V40/10

  Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands · CPC title

• G06F16/583 Primary

  using metadata automatically derived from the content · CPC title

• G06V20/52 Primary

  Surveillance or monitoring of activities, e.g. for recognising suspicious objects (recognising microscopic objects G06V20/69) · CPC title

• G06V10/74

  Image or video pattern matching; Proximity measures in feature spaces · CPC title