When face recognition meets 3D technology, who will be the final winner? Face recognition industry has long been plagued by the latest solution to the security problem. Recently, Ali security’s face recognition security technology solution was authorized by the U.S. patent office. According to this new patent scheme, only through the change of light and shadow and mobile phone position, can effectively prevent the attack of 3D face mask which is difficult to solve previously.
Photo caption: Ali security obtained the patent authorization notice of new technology of face recognition from US patent office
With the wide application of smart phones, online identity authentication and verification services such as remote account opening and card opening are becoming more and more popular. Through face recognition technology, people can verify their identity without leaving home, which greatly reduces the time cost of on-site verification, but also faces the threat of being attacked by criminals.
In order to prevent the system from being cheated by criminals by means of photos, videos and 3D software, there are two kinds of living detection schemes commonly used in the security industry. One is based on special hardware devices, such as binocular lens and depth camera for 3D detection of objects. This method can usually prevent attacks such as photos and videos, but it is limited to devices and cannot be used by smart phones.
The other is to recognize the specified actions, such as nodding, shaking, blinking, mouth opening and other actions. This method can be used in smart phones, but if attackers record all the verification actions in the video or use 3D face model, they can still cheat the system to break the security line.
For face recognition technology, 3D face model of entity was once an almost unsolvable security threat. If the attacker controls the 3D face model and makes the specified verification action in front of the camera, whether based on hardware detection or recognition based on specified action, it is difficult to prevent such attack. Prior to this, industry experts generally believe that there is no recognized mature, effective, convenient and easy-to-use safety testing scheme.
The two problems of effectiveness and ease of use are well solved in the face recognition security patent scheme proposed by Ali security: the security problem of preventing face recognition is comprehensively considered from three dimensions of equipment, algorithm and person. Jue’ao, a senior algorithm expert of Ali security, said that the scheme is an integrated judgment of people, equipment and algorithm, and the detection is mainly based on the changes of light, shadow and location.
“The mobile phone screen can take the initiative to light and change the color of the light to obtain the color and brightness changes of different areas of the face. When the user moves the mobile phone, the sensor can get the spatial position change of the device. ” Jue’ao explained that the authenticity of the face in front of the camera can be judged by combining the changes of color, brightness, face posture and equipment space position. This security technology scheme is not only convenient and easy to use, but also can effectively identify 3D model, video action and other attack means.
Since this year, digital infrastructure has become the focus of attention of all sectors of society, and the concept of security infrastructure has also become a consensus in the field of network security. How to keep digital infrastructure on a high security baseline from the beginning requires strong security technology as support. “The core technology of face recognition security technology solution comes from the important algorithm technology at the bottom of Ali’s new generation security architecture.” Jue’ao said that the wide application of this security technology scheme can improve the anti spoofing attack ability of face recognition system, solve the common problems faced by the industry, and make face recognition technology have broader and safer application scenarios in the wave of digital infrastructure.