Glory plug-in Face ID iPhone X look after marvel

Not long ago glory V10 conference, the end of the release of a very interesting gadget: point cloud depth camera. This camera is not similar to other external cameras used to enhance the shooting level, but one can be used for 3D face recognition , 3D face modeling external camera module.

According to the conference data, the point cloud depth camera has a depth resolution of 1280 * 800, with sub-millimeter modeling accuracy, rebuild of 10 million cloud 10s, payment security identification, and 400ms speed unlocking. Support for 3D small object modeling, the same can reach sub-millimeter level, support smart to go to the background, to achieve gorgeous texture reproduction Interestingly, the point cloud depth camera also supports the same as the iPhone X animated emoticons, and more details.

Speaking of face recognition is no stranger to us who have entered the era of full-screen smartphones, and facial recognition has become a less than intuitive alternative to post-fingerprint recognition due to the backward transfer of fingerprint recognition modules. Yes, there are a lot of phones that do not support facial recognition. After updating the system, they started to support this biometric identification function.

There are purely through the software to achieve, of course, also achieved through the hardware principle, for example, today we say the point cloud depth camera.In fact, the point cloud depth camera is not the first mobile phone-related 3D face recognition camera module, iPhone X Face ID has enabled us to recognize the high-level face recognition feature style.In fact, they achieve the same principle, are the result of 3D structured light stereo recognition products. Based on the principle, deep-click about 3D structure of light is how to achieve three-dimensional graphics.

This article belongs to the original article, Ruiluo reprint, please indicate the source: Glory plug-in Face ID iPhone X after see yourself http://mobile.zol.com.cn/669/6690100.html

Face recognition is how to achieve

Biometrics is actually not a particularly complex process, simply to get through the biometric information and stored information compared to achieve the biometric function on the phone through the front camera in a non-contact way to obtain biological information images After the image is acquired, the data is processed and compared with the existing biological information in the database. When the information features and the recorded information reach a limited similarity, facial identification can be unlocked.

So, only through the software level facial recognition and 3D facial recognition What is the difference? Android camp now, the basic use of mobile phones are only through the software level facial recognition, and precisely because of this we can in some Facial recognition can be experienced on the phone only through a system upgrade, which is quite fast from a practical experience, but it is not good at security.

The reason why this method is unsafe is that facial recognition is a non-contact biometric method, which is also one of the salient features that are different from fingerprinting, so fingerprinting can be avoided by increasing the biological function of acquiring living beings Fake fingerprints are identified, but facial recognition is difficult to exclude from photo fraudulent use of biological characteristics.

No matter what kind of recognition method, the feature is obtained by recognizing the feature points.Generally speaking, the more feature points are entered, the higher the security of the biological recognition is, and if the algorithm allows, a certain area of ​​the occlusion Or changes in recognition accuracy of the less affected.Face recognition is also true, so in some Android mobile facial recognition propaganda we can see the description of the identification of feature points.But the problem is that as long as it is through the plane Photos get information points, no matter how many feature points, have the possibility of being unlocked photos.

Of course, we can enhance the security for the characteristics of the photo.Compared with the living body, there are three characteristics of the photo, one is that there is no living body characteristics, and this is difficult to achieve has been ruled out; the other is that there is only one angle of a photo.Other previous facial recognition devices , There is a need to increase the security by turning the face to obtain the multi-angle feature of the face, but the problem with this solution is that the recognition efficiency is reduced. In a speed-sensitive smart phone, such face recognition Fingerprint identification on the back has no advantage.

And the third characteristic is that the photos are flat, and the people are three-dimensional, so obtaining the three-dimensional information of the living face becomes the best direction for facial recognition. So what we call 3D facial recognition is such a solution. How does the program obtain biological information? This involves the 3D structure of the light principle.

This article belongs to the original article, Ruiluo reprint, please indicate the source: Glory plug-in Face ID iPhone X after see yourself http://mobile.zol.com.cn/669/6690100.html

Chop a grilled 3D structure of light is a what

How to get 3D facial features information on the phone? Some people say that the double camera, double camera can not measure the depth of field? In fact, so think no problem, our eyes see the world, see the world is three-dimensional, because our eyes There is a certain distance, to see a goal is the existence of a certain angle of view, so the eyes see the picture is different, the brain in the synthesis of the picture is three-dimensional and now there binocular stereoscopic imaging equipment, is based on the principle of parallax Three-dimensional image.

Our commonly used facial recognition, facial occlusion can also be achieved a certain area can be unlocked, this is because the same data as long as the same rate to reach the set conditions can be. Although mobile phone dual camera can take different perspective photos, but the difference is particularly small , Can not have a sufficient effect on the comparative data of facial recognition.

One of the best ways to get facial stereoscopic information is the 3D structured light we are talking about, for example, it works like a sonar system that can draw a shallow sea topographic map and reflect information to determine depth. Determines the depth of information by reflecting the light through the face.

The whole system of 3D structured light consists of structured light projection equipment, video camera, image acquisition and processing system, the process is that the projection equipment emits light to the measured object, the camera photographs the three-dimensional light pattern formed on the measured object, the captured image is collected After the processing system obtains the surface data of the measured object, the degree of light distortion projected on the measured object depends on the depth of the surface of the object when the relative positions of the camera and the projection device are relatively constant. Therefore, in the captured image, A deep light image.

3D Structured Light Mode A pattern of points, lines, and planes, which refers to the type of light that is projected. In general, a network structure consisting of multiple vertical bi-directional lines is most commonly used because it does not require scanning for a three-dimensional Contour measurement, and fast speed.With the 3D structure of light on the phone is a light system composed of multiple points, the choice of infrared can be unlocked by the light of a face embarrassment.

Write in the end

See here, I believe we have a 3D structure of the facial recognition theory of light have a certain understanding.Of course, these are only a brief introduction to the identification technology, the details are still more complex.For the current status quo, iPhone X Face recognition is not as fast as it is directly at the software level, but 3D facial recognition of light will be the mainstream biometric solution if face detection does not allow in-vivo detection of facial recognition.

At the beginning of the text we also mentioned the animated facial expressions and 3D modeling of point cloud camera.It is interesting to note that the animated emoticons supported by the iPhone X do not need the structured light reflection system to recognize as long as the front camera is exposed and the depth of the point cloud Camera details richer animated emotions, it is estimated that thanks to the use of 3D structured light system and better software adaptation.

3D modeling is not the exclusive feature of the point cloud depth camera. The recently released Sony Xperia XZ1 supports 3D cameras for face scan, head scan, food scan and free scan, but want to achieve perfect modeling The operation is very difficult, but in the conference, the point cloud depth camera can very easily scan the doll's model, but also useful excellent hand and background effect.This may be attributed to the point cloud depth camera 3D structured light technology and excellent algorithm applied.

This article belongs to the original article, Ruiluo reprint, please indicate the source: Glory plug-in Face ID iPhone X after see yourself http://mobile.zol.com.cn/669/6690100.html