CN111126244A

CN111126244A - Security authentication system and method based on facial expressions

Info

Publication number: CN111126244A
Application number: CN201911321797.4A
Authority: CN
Inventors: 李华康; 涂敏; 程科; 徐梦婷; 孙国梓; 余瀚
Original assignee: Journal Of Jiangxi Public Security College; Nanjing University of Posts and Telecommunications
Current assignee: Journal Of Jiangxi Public Security College; Nanjing University of Posts and Telecommunications
Priority date: 2019-12-20
Filing date: 2019-12-20
Publication date: 2020-05-08

Abstract

The invention discloses a security authentication system and a security authentication method based on facial expressions, wherein the security authentication system comprises an input module, a classification module and an authentication module; the input module is used for inputting images or videos of the positive and negative facial expressions and transmitting data to the classification module; the classification module is used for carrying out feature extraction and model training on the acquired data and generating a classification model; the authentication module authenticates the identified face according to the classification result. The authentication of the facial expression of the invention can avoid the problem of forced face brushing, thereby improving the security of the authentication.

Description

Security authentication system and method based on facial expressions

Technical Field

The present invention relates to a system and a method for authenticating facial security, and more particularly, to a system and a method for authenticating facial security based on facial expressions.

Background

The development of artificial intelligence and its technical development in the field of image processing have promoted the development of face recognition technology. The micro expression recognition technology can recognize facial action units of people and can recognize more than 90% expression changes. At present, systems based on the face recognition technology in the market bloom all the time, and in daily life, people also often use the technology, such as paying treasures, brushing faces and opening doors, and the like.

The face recognition technology brings convenience to people, and meanwhile, many potential safety hazards are caused, such as stealing brushing, for example, brushing faces is forced under the condition that a party is reluctant to do so.

Disclosure of Invention

The purpose of the invention is as follows: the invention aims to provide a safe and efficient safety authentication system and method based on facial expressions.

The technical scheme is as follows: the invention relates to a safety authentication system based on facial expressions, which comprises an entry module, a classification module and an authentication module.

The input module is used for inputting videos of the positive and negative facial expressions and transmitting data to the classification module.

The positive and negative expressions include at least two expressions, respectively.

The expression images are two images with certain difference.

The classification module is used for carrying out feature extraction and model training on the collected data and generating a classification model.

The authentication module authenticates the identified face according to the classification result.

And if the identification result is not in the classification result, the authentication module rejects the authentication request.

A safety authentication method based on facial expressions comprises the following steps:

(1) recording videos of positive and negative facial expressions;

(2) extracting features and carrying out model training to generate a classification model;

(3) and carrying out face recognition and verification.

The generation of the classification model in the step (2) comprises the following steps: extracting features through a convolutional neural network identification, and storing a feature matrix; and classifying the acquired features by using a classification algorithm.

Has the advantages that: compared with the prior art, the invention has the following remarkable advantages: the authentication of the facial expression of the invention can avoid the problem of forced face brushing, thereby improving the security of the authentication.

Drawings

FIG. 1 is a schematic diagram of the system of the present invention;

FIG. 2 is a schematic flow diagram of an entry module according to the present invention;

FIG. 3 is a schematic flow diagram of a classification module according to the present invention;

FIG. 4 is a schematic flow diagram of an authentication module according to the present invention.

Detailed Description

The technical scheme of the invention is further explained by combining the attached drawings.

As shown in fig. 1, a security authentication system based on facial expressions of the present invention includes: and the positive and negative expression recording module 100 is used for recording facial expression videos including positive expression videos and negative expression videos.

The facial expression and emotion classification module 200 is mainly used for performing feature extraction and model training on expression data acquired by the module 100 to generate a classification model. In the recording process, two face photos or a video with certain difference are obtained.

The authentication scene verification module 300 checks face recognition using the model obtained in the facial expression emotion classification module 200, determines whether the input expression is a trusted expression, and determines whether the input expression can pass authentication.

Positive expressions refer to expressions such as happy or smiling, and negative expressions are expressions such as anger or embarrassment. By inputting the positive and negative expressions and classifying, whether the person is in a voluntary state or not can be judged during face authentication, the problem of face brushing is avoided, and the authentication safety is improved. The user can also input any expression as a negative expression to improve the safety in the authentication process.

In the positive and negative expression entry module 100, the positive and negative expression entry specifically includes the following steps:

step 101: starting facial expression input, and preparing to receive an expression video uploaded by a user; step 102: inputting a first positive expression video; step 103: inputting a second positive expression video; step 104: inputting a third positive expression video, and step 105: inputting a first negative expression video; step 106: inputting a second negative expression video; step 107: recording a third negative expression video; step 108: and finishing the facial expression input, and storing all expression videos uploaded by the user.

In the facial expression emotion classification module 200, the following steps are specifically included:

step 201: starting expression classification training, and reading the video expression uploaded by a user; step 202: extracting the face features of the video, namely extracting the features of the face of each frame of picture in the video through a convolutional neural network and storing a feature matrix; step 203: extracting expression video characteristics, namely extracting the characteristics of the expression of each frame of picture in the video through a convolutional neural network identification, and storing a characteristic matrix; step 204: training an expression video classification model, classifying the expression video classification model by using a classification algorithm according to the obtained characteristics, and finally forming a classification model; step 205: and finishing the expression classification training.

In the authentication scenario verification module 300, the following steps are specifically included:

step 301: starting expression verification to obtain an expression video uploaded by a user; step 302: extracting the facial features of the video, namely extracting the facial features of the video according to the method in the step 202; step 303: extracting expression video characteristics, and extracting expression characteristics according to the method in the step 203; step 304: classifying the expression video features by using a pre-trained expression classification model to obtain a classification result, and step 305: judging positive and negative expressions, judging whether the positive and negative expressions are classified according to the classification result, if so, performing step 306, otherwise, performing step 307; step 306: rejecting the negative expression authentication request; step 307: the positive expression authentication request is completed.

Claims

1. A safety authentication system based on facial expressions is characterized by comprising an input module, a classification module and an authentication module; the input module is used for inputting images or videos of positive and negative facial expressions and transmitting data to the classification module; the classification module is used for carrying out feature extraction and model training on the acquired data and generating a classification model; the authentication module authenticates the identified face according to the classification result.

2. The facial expression-based security authentication system of claim 1, wherein the positive and negative expressions each comprise at least two expressions.

3. The system of claim 1, wherein the expression image is two images with a certain difference.

4. The system of claim 1, wherein the recognition result is not in the classification result, and the authentication module rejects the authentication request.

5. A safety authentication method based on facial expressions is characterized by comprising the following steps:

(1) recording videos of positive and negative facial expressions;

(3) and carrying out face recognition and verification.

6. The method for security authentication based on facial expressions according to claim 5, wherein the classification model generation in the step (2) comprises the steps of: extracting features through a convolutional neural network identification, and storing a feature matrix; and classifying the acquired features by using a classification algorithm.