Silent in-vivo detection methodTechnical Field
The invention relates to the technical field of face recognition, in particular to a silence living body detection method.
Background
In a biometric system, in order to prevent a malicious person from forging and stealing the biometric characteristics of another person for identity authentication, the biometric system needs to have a liveness detection function, i.e., to determine whether the submitted biometric characteristics are from a living individual.
The living body detection technology of the general biological characteristics utilizes the physiological characteristics of people, for example, living body fingerprint detection can be based on the information of temperature, perspiration, conductivity and the like of fingers, living body face detection can be based on the information of head movement, respiration, red-eye effect and the like, and living body iris detection can be based on iris flutter characteristics, motion information of eyelashes and eyelids, contraction and expansion response characteristics of pupils to the intensity of a visible light source and the like.
As the face recognition technology becomes mature, the commercial application becomes wider, but the face is easily copied by using the modes of photos, videos and the like, so that the counterfeit of the face of a legal user is an important threat to the safety of the face recognition and authentication system. At present, a silent in-vivo detection method based on a single photo or video frame has made certain progress.
At present, living body detection is mainly dynamic living body detection, which is used for identity authentication to prevent a malicious person from forging and stealing biological characteristics of other people, and a biological identification system needs to have a living body detection function, namely, to judge whether submitted biological characteristics come from a living individual.
The living body detection technology generally utilizes physiological characteristics of people, for example, living body fingerprint detection can be based on information such as temperature, perspiration and electric conductivity of fingers, living body face detection can be based on information such as head movement, respiration and red-eye effect, and living body iris detection can be based on iris flutter characteristics, motion information of eyelashes and eyelids, contraction and expansion response characteristics of pupils to visible light source intensity, and the like.
Compared with other biological feature recognition technologies, the dynamic face detection recognition technology has natural and unique advantages in practical application: the camera is used for directly obtaining the information, so that the identification process can be completed in a non-contact mode, and convenience and rapidness are realized. At present, the dynamic face detection and recognition technology is applied to the fields of finance, education, scenic spots, travel, social security and the like. The realization process is mainly through the combination action cooperation such as blink, open mouth, shake head, nod, ensure that what operate is real live body people face, and the defect lies in that the instruction action is loaded down with trivial details, needs artificial initiative cooperation.
Disclosure of Invention
In view of the above, the present invention is directed to the defects in the prior art, and the main objective of the present invention is to provide a silent biopsy method, which can effectively solve the problems of complicated instruction and operation and the need of artificial active coordination in the conventional dynamic biopsy method.
In order to achieve the purpose, the invention adopts the following technical scheme:
a silent liveness detection method comprising the steps of:
(1) extracting a face frame: positioning and extracting the face position by using a face frame extraction module, wherein the face frame extracted by the face frame extraction module is used as the input of feature modeling;
(2) modeling the face part characteristics: extracting the features of the face image by using a feature modeling module to characterize, wherein the output of the feature modeling module is used as the input of a classification module;
(3) and (4) classification: and (5) sending the characteristics into a classifier to realize classification judgment of living bodies and non-living bodies.
Preferably, in the step (2), SURF is used, feature descriptors are calculated and generated, PCA is used to project the feature descriptors to principal components, GMM is used to perform principal component coding to obtain a feature code with length of 76800, and SVM binary classification is finally performed.
Preferably, the PCA principal component analysis is introduced with the addition of LBP features.
Compared with the prior art, the invention has obvious advantages and beneficial effects, and specifically, the technical scheme includes that:
by adopting the method, the living body judgment can be carried out only by dynamically grabbing the face image by the camera, the defect of complicated dynamic instruction living body verification is effectively overcome, the artificial participation amount is reduced, the judgment on whether the living body exists or not can be realized, the action instruction coordination is not needed, the result can be obtained in a shorter time, the speed is high, the detection result is not influenced no matter under the conscious and unconscious conditions, different gait identification can be consciously imitated and changed in the future, the carrying of a card is not needed, the identification speed is high, the operation is simple and convenient, the contact with equipment is not needed, the contact infection of viruses is not needed to be worried, and the method is sanitary and safe.
Drawings
FIG. 1 is a flow chart illustrating a preferred embodiment of the present invention.
Detailed Description
The invention discloses a silent in vivo detection method, as shown in figure 1, comprising the following steps:
(1) extracting a face frame: the face position is positioned and extracted by using a face frame extraction module, and the face frame extracted by the face frame extraction module is used as the input of feature modeling.
(2) Modeling the face part characteristics: and extracting the features of the face image by using a feature modeling module to perform characterization, wherein the output of the feature modeling module is used as the input of a classification module.
(3) And (4) classification: and (5) sending the characteristics into a classifier to realize classification judgment of living bodies and non-living bodies.
In this embodiment, SURF is used in step (2), feature descriptors are calculated and generated, PCA is used to project the feature descriptors to principal components, GMM is used to perform principal component coding to obtain a feature code with length of 76800, and SVM binary classification is finally performed. And moreover, the LBP characteristic is added while PCA principal component analysis is introduced, so that double-characteristic recheck is realized, and the robust performance of the algorithm is improved.
The main program codes of the invention:
the main mathematical formula of the invention is as follows:
1、PCA：
algorithm flow
Suppose we need to reduce the feature dimension from the n dimension to the k dimension. The PCA is executed as follows:
1. feature normalization, balancing individual feature scales:
μjis the mean value of the features j, sjIs the standard deviation of characteristic j.
2. Calculating the covariance matrix Σ:
3. by Singular Value Decomposition (SVD), the eigenvectors (egenerctors) of Σ are found:
(U，S，VT)＝SVD(∑)
4. and taking out the first k left singular vectors from the U to form an approximation matrix reduce:
Ureduce＝(u(1)，u(2)，…，u(k))
5. calculating a new feature vector: z is a radical of(i)
2、LBP
Wherein,
the design key points of the invention are as follows: by adopting the method, the living body judgment can be carried out only by dynamically grabbing the face image by the camera, the defect of complicated dynamic instruction living body verification is effectively overcome, the artificial participation amount is reduced, the judgment on whether the living body exists or not can be realized, the action instruction coordination is not needed, the result can be obtained in a shorter time, the speed is high, the detection result is not influenced no matter under the conscious and unconscious conditions, different gait identification can be consciously imitated and changed in the future, the carrying of a card is not needed, the identification speed is high, the operation is simple and convenient, the contact with equipment is not needed, the contact infection of viruses is not needed to be worried, and the method is sanitary and safe.
The technical principle of the present invention is described above in connection with specific embodiments. The description is made for the purpose of illustrating the principles of the invention and should not be construed in any way as limiting the scope of the invention. Based on the explanations herein, those skilled in the art will be able to conceive of other embodiments of the present invention without inventive effort, which would fall within the scope of the present invention.