Identity authentication method, storage medium, processor and system

Abstract

The invention provides an identity authentication method, a storage medium, a processor and a system. The method comprises the steps of receiving GPS information, mobile terminal information carried by a to-be-identified body and image information of the to-be-identified body; colliding the GPS information and the mobile terminal information, and scheduling the feature library data collided by theGPS information and the mobile terminal by a server to update the data to a local dynamic database at regular time; and extracting biological characteristic information from the to-be-identified bodyimage information, matching the biological characteristic information with the local dynamic database, and if matching succeeds, determining that the to-be-identified body passes authentication. Theoperation capability can be improved, the comparison time can be shortened, the portrait library can be reduced, the personnel identity recognition accuracy can be improved, the portrait library accurate search capability can be improved, and the hardware server cost can be reduced.

Description

【Technical field】 [0001] The technical field of identity authentication of the present invention relates in particular to an identity authentication method, storage medium, processor and system. 【Background technique】 [0002] At present, the violent comparison method mainly used in portrait comparison in the existing technology needs to consume a lot of computing resources and hardware costs under the premise of satisfying QPS and TPS. The accuracy rate of face recognition will also decrease with the increase of the face database, which will also bring corresponding risks in identity authentication. The main flaws are as follows: [0003] Computing resource consumption is high. Existing AI technology will use high-performance servers to perform calculations on the premise of satisfying QPS and TPS. However, in the case of massive data, performing a full database comparison search will occupy a large amount of machine computing resources and will not release them, resultin...

AI Technical Summary

Problems solved by technology

[0003] Computing resource consumption is high

Existing AI technology will use high-performance servers to perform calculations on the premise of satisfying QPS and TPS. However, in the case of massive data, performing a full database comparison search will occupy a large amount of machine computing resources and will not release them, resulting in a backlog of tasks. Unable to output results in real time

[0004] The huge portrait library leads to a decrease in recognition accuracy

At present, the LWF face recognition accuracy rate reaches 99.83%, but it is only limited to the results of fixed competition sets. In actual scenes, due to various factors in the image quality of the face database, when the portrait database is too large, the recognition accuracy will decrease when comparing and searching the target person. Dropped significantly

It leads to an increase in the misidentification rate, which leads to identification errors. If it is used in the security and financial fields, it will cause certain losses.

[0005] The portrait database cannot be accurately positioned and searched

The traditional portrait library does not know when and where the target person appears in a certain portrait database. As a result, when building the portrait database, the target person and the non-target person will share the same portrait database, which cannot accurately locate and compare the database. When comparing the identity of the controlled person , will search and compare the whole library, increase the target search range, and reduce the QPS of the server

[0006] High cost of server hardware

In the industry, because it is impossible to accurately locate when the target person will appear, the portrait database exists as a global database. When comparing 1:N and M:N, a violent comparison method is used. When the data volume of the portrait database is huge, the consumed The computing resources are large, and the corresponding server hardware cost will also increase under the condition of satisfying QPS

Images