Method and apparatus for security inspection

Patent metadata

Field	Value
Publication number	US-10509985-B2
Application number	US-201715819559-A
Country	US
Kind code	B2
Filing date	Nov 21, 2017
Priority date	Dec 23, 2016
Publication date	Dec 17, 2019
Grant date	Dec 17, 2019

Abstract

Disclosed in the present disclosure are a method and an apparatus for security inspection. The method comprises: acquiring identity related information of a person being inspected, the identity related information comprising a certificate photo; acquiring a real-time facial image of the person being inspected; by comparing the real-time facial image and the certificate photo, acquiring a facial comparison similarity; dynamically determining the threshold of the facial comparison similarity; and performing a human-certificate verification on the person being inspected according to the facial comparison similarity and the threshold. The method for security inspection of the present disclosure enables the fast and accurate facial comparison of the person being inspected during security inspection.

First claim

The invention claimed is:

1. A method for security inspection, comprising:
acquiring identity related information of a person being inspected, the identity related information comprising a certificate photo;
acquiring a real-time facial image of the person being inspected;
by comparing the real-time facial image and the certificate photo, acquiring a facial comparison similarity;
generating a dynamic threshold model from data of historical human-certificate verification, the dynamic threshold model being configured to acquire a threshold of the facial comparison similarity;
dynamically determining the threshold of the facial comparison similarity; and
performing a human-certificate verification on the person being inspected according to the facial comparison similarity and the threshold,
wherein the generating a dynamic threshold model from data of historical human-certificate verification comprises:
acquiring the data of the historical human-certificate verification;
marking the data of the historical human-certificate verification according to the result of the actual human-certificate verification; and
storing the data of the historical human-certificate verification and the marked entry in the data of the historical human-certificate verification into a sample library.

2. The method according to claim 1 , wherein the identity related information further comprises certificate number, gender, nationality, date of birth, residential address, and length of time of certificate handling.

3. The method according to claim 1 , wherein dynamically determining the threshold of the facial comparison similarity comprises:
dynamically determining the threshold of the facial comparison similarity through the dynamic threshold model.

4. The method according to claim 1 , further comprising:
generating a first data set from the sample library by means of data cleaning;
mining the first data set through a large data visualization analysis technology, and acquiring facial comparison associated features; and
extracting, from the sample library, data corresponding to the facial comparison associated features, and generating a facial comparison associated feature library.

5. The method according to claim 4 , further comprising:
mining the facial comparison associated feature library through a machine learning algorithm, and generating a facial comparison threshold model.

6. The method according to claim 5 , wherein the machine learning algorithm is implemented based on Spark Mllib.

7. The method according to claim 4 , wherein the large data visualization analysis technology is based on an ElasticSearch server and processes the data of the historical human-certificate verification.

8. The method according to claim 4 , wherein the large data visualization analysis technology comprises a Kibana visualization interface framework.

9. An apparatus for security inspection, comprising:
a processor; and
a memory for storing instructions executable by the processor;

wherein the processor is configured to:
acquire identity related information of a person being inspected, the identity related information comprising a certificate photo;
acquire a real-time facial image of the person being inspected;
acquire, by comparing the real-time facial image and the certificate photo, a facial comparison similarity;
generate a dynamic threshold model from data of historical human-certificate verification, the dynamic threshold model being configured to acquire the threshold of the facial comparison similarity;
dynamically determine the threshold of the facial comparison similarity; and
perform a human-certificate verification on the person being inspected according to the facial comparison similarity and the threshold,
wherein the processor configured to generate a dynamic threshold model from data of historical human-certificate verification is configured to:
acquire the data of the historical human-certificate verification;
mark the data of the historical human-certificate verification according to the result of the actual human-certificate verification; and
store the data of the historical human-certificate verification and the marked entry in the data of the historical human-certificate verification into a sample library.

10. A non-transitory computer-readable storage medium storing instructions which, when executed by a processor, cause the processor to perform a method comprising:
acquiring identity related information of a person to be checked, the identity related information comprising a certificate photo;
acquiring a real-time facial image of the checked person;
by comparing the real-time facial image and the certificate photo, acquiring a facial comparison similarity;
generating a dynamic threshold model from data of historical human-certificate verification the dynamic threshold model being configured to acquire a threshold of the facial comparison similarity;
dynamically determining the threshold of the facial comparison similarity; and
performing a human-certificate verification on the checked person according to the facial comparison similarity and the threshold,
wherein the generating a dynamic threshold model from data of historical human-certificate verification comprises:
acquiring the data of the historical human-certificate verification;
marking the data of the historical human-certificate verification according to the result of the actual human-certificate verification; and

storing the data of the historical human-certificate verification and the marked entry in the data of the historical human-certificate verification into a sample library.

Assignees

• Nuctech Co Ltd

Inventors

• Cui Jin
• Tan Huabin
• Zhou Shenglian
• Peng Zhi

Classifications

• G06V10/776

  Validation; Performance evaluation · CPC title

• G06V10/764

  using classification, e.g. of video objects · CPC title

• G06F18/22

  Matching criteria, e.g. proximity measures · CPC title

• G06N20/00 Primary

  Machine learning · CPC title

• G06F18/217

  Validation; Performance evaluation; Active pattern learning techniques · CPC title