Face recognition systems need to have in vivo detection, facial feature to determine whether the submitted individuals from real life.
The basic principle of face detection in vivo
The basic functions of the face authentication access of a human face (Face Verification), and in vivo detection of anti-counterfeiting technology belonging to the face (Face Anti-Spoofing). Face Verification and face security, the two technologies have different emphases.
Face Verification: two face authentication is judged whether a face of the same person in FIG algorithm, i.e., alignment of two face features obtained similarity, and then compared with a preset threshold value by the face, the similarity is larger than the threshold value, compared with the same person, otherwise different. It is very popular in recent years, a research also produced a large number of models and algorithms loss function.
Face security: User brush face when screening algorithm to this face is not a real person living face, and for photos, videos and prosthetic masks *** behavior, the algorithm should be rejected.
1, in vivo detection and photo ***
*** photo is the easiest way to use social media, such as micro-channel circle of friends or Twitter, you can easily get photos related personnel. But after all photos are static and can not make a blink of an eye, mouth, turned to other activities. Using this feature, the detection system may be released in vivo several operation command by determining the conformity of operation of testers, the interactive operation can be achieved in vivo detection.
To combat action in vivo testing, there were improvements *** photo camouflage print another person's photos by life-size, in the eyes and mouth parts of the hollowed photographs, attached to the face, her eyes and mouth. In vivo detection system according to the instruction, the execution blinking, mouth, head and other movements. However, this effect is forged with the actual movement of real people face a far cry from detection algorithm can easily be identified.
Common Access Method face detected in vivo
Vivo detection operation mode with high security, but it requires the user to do with the designated action, the actual poor user experience. In order to achieve the effect of the passage without a sense, face access response command is rarely used in vivo detection operation, generally carried out in vivo screening based on the difference image and the optical effect
Vivo detection ordinary camera, infrared camera detecting the living body, the living body is detected 3D camera
In vivo detection technology is an essential human face access control system of non-thinking pills, the accuracy rate of up to 99%, widely praised by the public. In no sense of identity and access increasingly popular today, in vivo detection to provide more security and convenience for the public life of work and life.