fingerprintmodule

[fingerprintmodule]

The application of face recognition module in New retail industries

[fingerprintmodule]

The application of face recognition module in New retail industries

The application of face recognition module in New retail industries

 (1) The advantages of applying face recognition

 Portrait of key customers: Help sellers obtain more accurate information about customers and potential customers, and build user portraits.

It can be installed at the entrance of supermarkets, shopping malls, stores, etc., to count the number of people entering stores every day, approximate age and gender, etc.

The other can be mounted on a shelf to analyze customers' concerns and spending habits.

Through big data analysis to find repeat customers, improve customer bag carrying rate and VIP conversion rate;

Cost reduction gain for retailers: intelligent system to replace the manual, with face recognition system to connect the payment terminal to replace the cashier, can quickly realize the retail store diversion and commodity flow analysis.

Reduce the occurrence of emergencies: In the event of commodity theft, the store can also put bad customers into the "black list" or reduce their credit level by analyzing the acquired data.

Perfect connection between online and offline: the user preferences obtained by the identification system can feed back to the online, and the data obtained can be fed back to the manufacturer through the online, helping the manufacturer to have a more comprehensive understanding of consumer needs, and then accurately develop products and design marketing strategies.

These are perfect ways to achieve the new retail "get through online and offline" inherent requirements.

(2) The security risks of face recognition

 Facial features are easy to copy: It is well known that the most common way to crack passwords is to copy them, and there have been numerous cases in which digital passwords and fingerprints have been stolen to decrypt them.

Exposed faces are easier to copy than digital codes recorded in the brain or on other media.

By taking photos, a person's facial features can be obtained and copied, and fraudulent methods such as plastic surgery technology or photo recognition can be used to fool the face payment system.

Personal information leakage problem: In today's advanced technology, people seem to easily find a variety of consumer information through all kinds of channels.

For face payment, such as facial features of the human body password once handed over to others, how to ensure the security factor of personal information?

Is access to users' facial features a matter of personal privacy?

Will widespread use of facial scanner-based payments lead to the disclosure of personal whereabouts through location-based services?

 From: www.zyjjhome.com

Gouku Technology, focus on human biometric identification products research and development and manufacturing, committed to the development of the best fingerprint module and face recognition module.

e-mail: mark@zyjjhome.com

[fingerprintmodule]

The application of face recognition module in New retail industries

The application of face recognition module in New retail industries (1) The advantages of applying face recognition Portrait of key customers: Help sellers obtain more accurate information about customers and potential customers, and build user portraits. It can be installed at the entrance of supermarkets, shopping malls, stores, etc., to count the number of people entering stores every day,…

[fingerprintmodule]

The application of face recognition module in Medical industry

(1) Key application We will crack down on medical crimes and ensure the safety of medical treatment. A targeted distribution and control database for medical offenders has been established, and real-time distribution and control has been carried out in cooperation with local public security departments. Control duty crimes and control improper competition. To control the medical…

[fingerprintmodule]

The application of face recognition module in Medical industry

(1) Key application

 We will crack down on medical crimes and ensure the safety of medical treatment.

A targeted distribution and control database for medical offenders has been established, and real-time distribution and control has been carried out in cooperation with local public security departments.

Control duty crimes and control improper competition.

To control the medical representatives entering the diagnosis and treatment area of the hospital, and to help solve the problems of irregular operation and out-of-order competition in the field of drug circulation.

Put an end to professional medical trouble, protect personal safety.

Combat frequent occupational medical trouble, improve the response speed of events, from passive response to active prevention.

Standardize the treatment process and harmonize the doctor-patient relationship.

Focus on preventing scalpers, medical care and other special groups that interfere with the normal order of medical treatment.

We will strengthen supervision and safeguard medical insurance funds.

To realize the comparison between the medical patients and the medical insurance information database, and eliminate the phenomenon of fraudulent use of medical insurance cards.

Accident prone serious mental disorder control.

In combination with the "Xueliang Project", we will ensure that the entry and exit places of patients with severe mental disorders are properly detected and controlled.

(2) The application breakthrough of face recognition in the medical industry is based on three points

 Obtaining the information of the target object: Because of the different administrative systems, it is difficult for the medical industry to obtain the information of the target object, which requires the relevant administrative units to carry out key coordination work.

The target information includes but is not limited to: face photos, portrait photos, basic information of personnel, personnel dynamics, etc.

Face recognition algorithm is further improved: the accuracy of the current face recognition algorithm has reached a quite high level, false positives, missing positives have been controlled in the acceptable range;

More advanced algorithms can obtain more value information from unstructured videos/pictures and realize different applications from more dimensions.

The improvement of managers' thinking and level: artificial intelligence and face recognition are revolutionary and subversive technologies, which can bring huge improvement to the medical industry.

How to apply face recognition to all aspects of the medical industry requires managers and technology providers to expand their thinking and work together.

(3) The prospect of face recognition in the medical industry

 Docking with the public security video surveillance and medical police linkage platforms: the system meets the existing standards and requirements of the public security, and can be seamlessly connected with the video surveillance and medical police linkage platforms of the public security organs in the future, and push the alarm information and related videos and pictures to the police stations in the jurisdiction to realize linkage.

Face identity verification: enter the photo of the target person, you can know the identity of the person and whether it belongs to the key control personnel, whether it has been to the hospital, and its appearance time, frequency.

Can be used to screen suspicious personnel, find their activity pattern.

Personnel track playback: Input the photo of the target person, you can inquire whether the person has been to the hospital, where he has been.

This function can restore the action track of specific personnel, for the suspect behavior research and forensics.

Docking access control system: Docking with access control system, reserving advanced functions such as face-brushing, face-checking, etc., to facilitate the access management of office area, operating room, pharmaceutical warehouse, inpatient department and other areas.

Connect with the card swiping system: Connect with the second-generation certificate, medical insurance card and other card swiping systems, compare the collected facial photos with the photos stored on the certificate, and verify the real identity of the card swiper.

 From: www.zyjjhome.com

Gouku Technology, focus on human biometric identification products research and development and manufacturing, committed to the development of the best fingerprint module and face recognition module.

e-mail: mark@zyjjhome.com

[fingerprintmodule]

The application of face recognition module in Finance

(1) Real-name authentication Financial institutions traditionally use artificial visual judgment, text message verification, binding bank cards and other means for real-name authentication. These traditional methods have problems such as low accuracy, poor customer experience and high cost, which have caused great troubles to the business development of financial enterprises. Real-name…

[fingerprintmodule]

The application of face recognition module in Finance

(1) Real-name authentication

 Financial institutions traditionally use artificial visual judgment, text message verification, binding bank cards and other means for real-name authentication.

These traditional methods have problems such as low accuracy, poor customer experience and high cost, which have caused great troubles to the business development of financial enterprises.

Real-name authentication based on face recognition has the advantages of high accuracy (100 million people only two people look the same), good customer experience (authentication speed, less customer operation), low cost (compared with traditional authentication), has been adopted by many leading financial enterprises.

 (2) face recognition in the bank remote account application

 During remote account opening, financial institutions can conduct online identity authentication through intelligent terminals. Using face recognition technology to open account can greatly improve the security and timeliness of business handling, and save a lot of manpower.

 (3) Brush face to withdraw money

 In this respect, the face has replaced the bank card, only need face + password to complete the withdrawal.

In the first two aspects, facial recognition technology has been widely adopted by major banks in China. In terms of facial withdrawal, Agricultural Bank of China and China Merchants Bank took the lead in launching facial withdrawal function in ATM.

 From: www.zyjjhome.com

Gouku Technology, focus on human biometric identification products research and development and manufacturing, committed to the development of the best fingerprint module and face recognition module.

e-mail: mark@zyjjhome.com

[fingerprintmodule]

The difference and advantages and disadvantages of semiconductor fingerprint module and optical fingerprint module in intelligent lock

The difference and advantages and disadvantages of semiconductor fingerprint module and optical fingerprint module in intelligent lock The Internet of Things industry of the Internet of everything is in full swing, smart home has also become the current hot point, from the “first year” of smart home in 2014, the landing of smart home is not satisfactory, but smart door lock is certainly one of…

[fingerprintmodule]

The difference and advantages and disadvantages of semiconductor fingerprint module and optical fingerprint module in intelligent lock

The difference and advantages and disadvantages of semiconductor fingerprint module and optical fingerprint module in intelligent lock

The Internet of Things industry of the Internet of everything is in full swing, smart home has also become the current hot point, from the "first year" of smart home in 2014, the landing of smart home is not satisfactory, but smart door lock is certainly one of the few in the whole industry trend is really promoted to succeed in a large range of market landing one of the smart subclasses. The current popularity of intelligent door locks can be detected from exhibitions, market feedback, real estate and e-commerce platforms. Almost half of each exhibition are large and small brands of smart door lock manufacturers, Tmall, Jingdong e-commerce home intelligent door lock list is the normal. With the maturation of the fingerprint lock industry and the improvement of people's living standards, more and more consumers have chosen fingerprint lock when decorating or changing the lock. At present, the market is mainly divided into two categories, one is semiconductor fingerprint recognition, the other is optical fingerprint recognition two categories. But many consumers and even some fingerprint lock practitioners are difficult to distinguish between the two differences and advantages and disadvantages.

Optical fingerprint module

Advantages: Optical fingerprint sensors are reliable, inexpensive, and wear resistant.

Disadvantages: 1. The fingerprint image recognition rate of dirty fingers and dry fingers with covering is very low; 2. Poor adaptability to temperature and other environmental factors. However, due to the limitation of optical path, the size of non-distortion collector is larger. It usually has severe optical distortion. 3. There are often traces left on the surface of the acquisition window. 4. CCD devices may reduce image quality and false fingerprints due to aging life

Semiconductor fingerprint module (you should be familiar with this, mobile phone fingerprint recognition is semiconductor) Advantages: 1. High recognition rate. 2, can automatically end the image acquisition, and the image quality is getting better and better. 3, anti-counterfeiting fingerprint ability is strong. 4, strong anti-static ability. 5, ultra-thin volume: can be embedded in a variety of terminal products.

Disadvantages: 1, easy to be affected by static electricity, so that the sensor sometimes can not read the image, or even destroy the image. 2. Insufficient wear resistance. Which affects its lifespan. 3, the price is more expensive than the optical fingerprint sensor.

From: www.zyjjhome.com Gouku Technology, focus on human biometric identification products research and development and manufacturing, committed to the development of the best fingerprint module and face recognition module. e-mail: mark@zyjjhome.com

[fingerprintmodule]

History of facial recognition technology development

History of facial recognition technology development

1. Understanding of facial recognition

Face recognition is a biometric technology that automatically identifies individuals based on their facial features (such as statistical or geometric features). It is also known as face recognition, portrait recognition, facial recognition, facial recognition, etc. The commonly referred to face recognition is the abbreviation for identity recognition and verification based on optical facial images.

Face recognition utilizes cameras or cameras to capture images or video streams containing faces, and automatically detects and tracks faces in the images, thereby performing a series of related application operations on the detected face images. Technically, it includes image acquisition, feature localization, identity confirmation and search, and so on. Simply put, it means extracting facial features from photos, such as eyebrow height, mouth angle, etc., and then outputting the results through feature comparison.

2. A brief history of the development of facial recognition

First stage (1950-1980s) Primary stage

Face recognition is regarded as a general pattern recognition problem, and mainstream technologies are based on the geometric structural features of the face. This is mainly reflected in people's research on silhouettes, where a large amount of research has been conducted on the structural feature extraction and analysis of facial silhouette curves. Artificial neural networks were once used by researchers in face recognition problems. In addition to Bledsoe, Goldstein, Harmon, Kanade Takeo and other researchers engaged in AFR research earlier. Overall, this stage is the initial stage of facial recognition research, and there are not many significant achievements, and there have been few practical applications.

The second stage (1990s) climax stage

Although this stage is relatively short in time, face recognition has developed rapidly, and many classic methods have emerged, such as Eigen Face, Fisher Face, and elastic graph matching; Several commercially available facial recognition systems have emerged, such as the most famous Visionics (now Identix) FaceIt system. In terms of technical scheme, linear subspace discriminant analysis, statistical appearance model and statistical pattern recognition method of 2D face image are the mainstream technologies in this stage.

Phase 3 (late 1990s to present)

The research on face recognition continues to deepen, and researchers have begun to pay attention to face recognition problems facing real conditions, mainly including the following four aspects of research:

1) Propose different facial space models, including linear modeling methods represented by linear discriminant analysis, nonlinear modeling methods represented by Kernel method, and 3D face recognition methods based on 3D information.

2) Thoroughly analyze and study the factors that affect facial recognition, including illumination invariant facial recognition, pose invariant facial recognition, and expression invariant facial recognition.

3) Utilize new feature representations, including local descriptors (Gabor Face, LBP Face, etc.) and deep learning methods.

4) Utilize new data sources, such as video based facial recognition and facial recognition based on sketching and near-infrared images.

 From: www.zyjjhome.com

Gouku Technology, focus on human biometric identification products research and development and manufacturing, committed to the development of the best fingerprint module and face recognition module.

e-mail: mark@zyjjhome.com

[fingerprintmodule]

History of facial recognition technology development

History of facial recognition technology development 1. Understanding of facial recognition Face recognition is a biometric technology that automatically identifies individuals based on their facial features (such as statistical or geometric features). It is also known as face recognition, portrait recognition, facial recognition, facial recognition, etc. The commonly referred to face…

[fingerprintmodule]

The main indicators of facial recognition technology

The main indicators of facial recognition technology

Key indicators in facial detection

Example: In a captured image of a camera, there are a total of 100 faces. The algorithm detects 80 faces, of which 75 are real faces and 5 are mistakenly recognizing road signs as faces.

Detection rate: Identify correct faces/all faces in the image. The higher the detection rate, the better the performance of the detection model.

Misdetection rate: recognition of incorrect faces/recognized faces. The lower the false detection rate, the better the detection model performance.

Misdetection rate: Unrecognized faces/all faces in the image. The lower the missed detection rate, the better the detection model performance.

Speed: The time from the completion of image acquisition to the completion of facial detection. The shorter the time, the better the detection model effect.

In this practical case: detection rate=75/100, false detection rate=5/80, missed detection rate=(100-75)/100

Key indicators in facial recognition

Out of 1000 sample images, there are a total of 600 positive samples. There are a total of 100 images with a similarity of 0.9, of which 99 are positive samples. Although the accuracy of the 0.9 threshold is very high, at 99/100; However, the number of correct outputs with a 0.9 threshold is very small, only 99/600. This can easily lead to missed identification.

Detection rate: Identify correct faces/all faces in the image. The higher the detection rate, the better the performance of the detection model.

Misdetection rate: recognition of incorrect faces/recognized faces. The lower the false detection rate, the better the detection model performance.

Misdetection rate: Unrecognized faces/all faces in the image. The lower the missed detection rate, the better the detection model performance.

Speed: The time from the completion of image acquisition to the completion of facial detection. The shorter the time, the better the detection model effect.

In this practical case: detection rate=75/100, false detection rate=5/80, missed detection rate=(100-75)/100

Key indicators in facial matching

Out of 1000 sample images, there are a total of 600 positive samples. There are a total of 100 images with a similarity of 0.9, of which 99 are positive samples. Although the accuracy of the 0.9 threshold is very high, at 99/100; However, the number of correct outputs with a 0.9 threshold is very small, only 99/600. This can easily lead to missed identification.

Precision: Number of samples identified as correct/Number of samples identified=99/100

Recall rate: Number of samples identified as correct/number of correct samples in all samples=99/600

False Acceptance Rate:

i. Definition: Refers to distinguishing two photos with different identities as the same identity, the lower the better

ii.      FAR = NFA / NIRA

In the formula, NIRA represents the number of inter class tests, which is different from the number of tests between similar classes. For example, if there are 1000 recognition models and 1000 people need to be identified, and each person only provides one material to be identified, then NIRA=1000 * (1000-1). NFA is the number of error acceptances.

Iv. FAR determines the security of the system, while FRR determines the ease of use of the system. In practice, the risk associated with FAR is much higher than that of FRR. Therefore, in biometric systems, FAR is set to a very low range, such as 1/10000 or even 1/million. Under fixed FAR conditions, FRR is below 5%, which is the only practical value of such systems.

FRR False Reject Rate:

i. Definition: Refers to distinguishing two photos with the same identity into different identities, the lower the better

ii.      FRR = NFR / NGRA

Iii. In the above equation, NFR is the number of intra class tests, which refers to the number of tests within the same category. For example, if there are 1000 recognition models and 1000 people need to identify, and each person only provides one material to be recognized, then NIRA=1000. If each person provides N images, then NIRA=N * 1000. NFR is the number of incorrect rejections.

 From: www.zyjjhome.com

Gouku Technology, focus on human biometric identification products research and development and manufacturing, committed to the development of the best fingerprint module and face recognition module.

e-mail: mark@zyjjhome.com

[fingerprintmodule]

The main indicators of facial recognition technology

The main indicators of facial recognition technology Key indicators in facial detection Example: In a captured image of a camera, there are a total of 100 faces. The algorithm detects 80 faces, of which 75 are real faces and 5 are mistakenly recognizing road signs as faces. Detection rate: Identify correct faces/all faces in the image. The higher the detection rate, the better the performance…

[fingerprintmodule]

The process of facial recognition

The process of facial recognition 1. Face collection (1) Introduction Different facial images are captured through camera lenses, such as static images, dynamic images, different positions, and different expressions. When the collected object is within the shooting range of the device, the collection device will automatically search for and capture facial images. (2) The main influencing factors of face collection Image size: If the face image is too small, it will affect the recognition effect, while if the face image is too small, it will affect the recognition speed. The minimum recognized face pixels for non professional facial recognition cameras are generally 60 * 60 or 100 * 100 or above. Within the specified image size, the algorithm is more likely to improve accuracy and recall. The image size is reflected in the actual application scenario as the distance between the face and the camera. Image resolution: The lower the image resolution, the more difficult it is to recognize. The size of the image, combined with the resolution of the image, directly affects the recognition distance of the camera. The maximum distance for a 4K camera to see a face is 10 meters, while a 7K camera is 20 meters. Lighting environment: Overexposed or dark lighting environment can affect facial recognition performance. You can use the built-in functions of the camera to supplement or filter light to balance lighting effects, and you can also use algorithm models to optimize image lighting. Blur level: The actual scene mainly focuses on solving motion blur, and the movement of the face relative to the camera often produces motion blur. Some cameras have anti blurring capabilities, but in the case of limited cost, consider optimizing this problem through algorithm models. Occlusion level: The best image is one with unobstructed facial features and clear facial edges. In practical scenarios, many faces are obstructed by obstacles such as hats, glasses, masks, etc. This data needs to be determined based on algorithm requirements whether to retain training. Collection angle: The best angle for the face relative to the camera is the front face. But in actual scenes, it is often difficult to capture the front face. Therefore, the algorithm model needs to train data that includes left and right faces, upper and lower faces. The angle of camera placement in industrial construction must meet the requirements of the angle between the face and the camera within the algorithm recognition range. 2. Face detection (1) Introduction Accurately demarcate the position and size of the face in the image, and extract useful information from it (such as histogram features, color features, template features, structural features, and Haar features), and then use the information to achieve the purpose of face detection. (2) Face Key Point Detection (Face Alignment) Automatically estimate the coordinates of facial feature points on facial images. (3) Mainstream methods Based on the detected features, the Adaboost learning algorithm (a classification method that combines some weaker classification methods to create a new strong classification method) is used to select some rectangular features (weak classifiers) that best represent the face. The weak classifier is constructed into a strong classifier using a weighted voting method, Then, concatenate the trained strong classifiers to form a cascaded hierarchical classifier, effectively improving the detection speed of the classifier. The recent genres of facial detection algorithm models include three types and their combinations: the Viola-Jones framework (with average performance and decent speed, suitable for use on mobile and embedded devices), DPM (with slower speed), and CNN (with good performance). 3 Facial Image Preprocessing (1) Introduction Based on the results of facial detection, the image is processed and ultimately serves the process of feature extraction. (2) Reason The original images obtained by the system are often unable to be directly used due to various conditions and random interference. It is necessary to perform image preprocessing such as grayscale correction and noise filtering in the early stages of image processing. (3) Main pre-treatment process Face alignment (get the image with correct face position), light compensation of face image, gray scale transformation, histogram equalization, normalization (get standardized face images with consistent size and same gray scale range), geometric correction, median filtering (smooth the image to eliminate noise), sharpening, etc. 4 facial feature extraction (1) Introduction The features that facial recognition systems can use are usually divided into visual features, pixel statistical features, facial image transformation coefficient features, facial image algebraic features, etc. Facial feature extraction is the process of modeling facial features, also known as facial representation, based on certain features of the face (2) Method of facial feature extraction Knowledge based representation methods (mainly including geometric feature method and template matching method): Based on the shape description of facial organs and their distance characteristics, feature data that is helpful for facial classification is obtained. The feature components usually include Euclidean distance, curvature, and angle between feature points. The face is composed of parts such as the eyes, nose, mouth, chin, etc. The geometric description of these parts and their structural relationships can be used as important features for facial recognition, and these features are called geometric features. Algebraic feature based or statistical learning based representation methods: The basic idea of algebraic feature based methods is to transform high-dimensional descriptions of faces in the spatial domain into low-dimensional descriptions in the frequency domain or other spaces. The representation methods are linear projection representation methods and nonlinear projection representation methods. The main methods based on linear projection include principal component analysis, also known as K-L variation, independent component analysis, and Fisher linear discriminant analysis. There are two important branches of nonlinear feature extraction methods: kernel based feature extraction technology and manifold learning dominated feature extraction technology. 5 Matching and Recognition The extracted facial feature value data is searched and matched with the feature templates stored in the database. By setting a threshold and comparing the similarity with this threshold, the identity information of the face is judged.

From: www.zyjjhome.com Gouku Technology, focus on human biometric identification products research and development and manufacturing, committed to the development of the best fingerprint module and face recognition module. e-mail: mark@zyjjhome.com

[fingerprintmodule]

The process of facial recognition

The process of facial recognition Face collection(1) IntroductionDifferent facial images are captured through camera lenses, such as static images, dynamic images, different positions, and different expressions. When the collected object is within the shooting range of the device, the collection device will automatically search for and capture facial images.(2) The main influencing factors of…

[fingerprintmodule]

Three common techniques of face recognition

Three common techniques of face recognition With the rapid development of communication equipment and camera technology, face recognition technology has become the main application support or important configuration of many products. The following content on the current face recognition of the three common technology for a systematic introduction, hope to provide you with a reference. As a hot…

[fingerprintmodule]

Three common techniques of face recognition

Three common techniques of face recognition

With the rapid development of communication equipment and camera technology, face recognition technology has become the main application support or important configuration of many products.

The following content on the current face recognition of the three common technology for a systematic introduction, hope to provide you with a reference.

As a hot technology in Internet field, face recognition technology has been widely used in Internet products and many fields in People's Daily life.

Face recognition has evolved into a broad concept, offering different products and services through several different technologies.

Different people in different working environments speak face recognition, their expectations of products and behind the technology will also reach a great degree of difference.

For example, the Face recognition technology needs of Internet finance risk control personnel is based on face recognition technology of real-name authentication or Face ID technology-based login authentication.

Most of the relevant staff in the operation and management of new media enterprises want face recognition technology, which requires the development and application of image face-changing marketing based on face detection information technology.

Understanding and analyzing role requirements and providing implementation solutions are essential skills for product managers.

If you are not familiar with the various product technologies and applications of face recognition, you may not be able to accurately understand what the requirements of the role of the first consideration, thus delaying the project.

The following content will be for face recognition of three kinds of information technology analysis.

One, face detection

Face detection is a technology to detect the location of the face in the image, which is also the initial meaning of face recognition.

Face detection is used to determine the size and position of the face in the image, that is, to solve the problem of "where the face is", the real face area is cut out from the image, for the subsequent face feature analysis and recognition.

After many years of technological improvement and research and development, our country has been developed by the realization of multiple face recognition detection in a picture, as well as human face detection in network video stream.

The first application of face detection was in the field of photography. We all know that previous cameras had a portrait mode. The basic principle of portrait mode is to process the image in the viewfinder frame.

Recognize the area of the image that belongs to a natural person and set the camera to focus on that area so that the face in the image is clear.

Face detection technology enterprises currently have the following two development research directions:

1. Face attribute recognition

Face attribute recognition is to analyze the region of the face image to get a series of attributes such as gender, age, expression and race.

The main reason for product application research is that we can establish customer portrait database and carry out big data portrait through rapid development of images, so as to realize precision marketing.

2. Face feature localization

Face feature localization refers to the accurate positioning of face features in the recognized face image region, in order to obtain the coordinates and feature values of face features in the image.

The most mainstream product application is social technology products commonly used for picture use beauty and entertainment face change.

Two, face comparison

Face comparison is to measure the similarity of two face pictures, and determine whether the natural person in the two face pictures serves the same person's product.

The technology needs to use the services provided by face detection and similarity algorithm to judge similarity.

It has also progressed to the point where we can compare instructional video streams and images, as well as develop multiple sensors to match facial signals.

It should be noted that face matching information technology enterprises can not in order to ensure 100% accuracy, product design must set such a threshold, similarity is higher than the threshold as we judge face matching (both judge for the same person).

The first application of face matching is supposed to be the notices of wanted criminals in ancient China, which usually attach face images and features of criminals together so that people can analyze face matching together.

In criminal investigation technology, the research and comparison between the photos of cyber criminal suspects and some basic feature image data is conducive to the rapid development of enterprises to determine the identity of criminal suspects and promote case investigation.

Three, in vivo detection

Vivisection detection is the pointer to face recognition in the process of face to do further detection, determine whether the object to be recognized is a real person.

At present, the methods of attacking face recognition system can be summarized into three categories: photo, video and mask. Among these three categories, mask is the most difficult to solve, mainly because its authenticity is close to real people.

At present, there are three main methods for in vivo detection. One is based on planar two-dimensional RGB cameras, the other is based on infrared cameras, and the third is a in vivo detection scheme based on three-dimensional depth cameras.

Our faces appear to be static, but in fact they are not.

From a micro point of view, our faces are also in constant motion, always showing some micro expressions that we are not aware of.

It may be difficult to see with great precision, but the high-resolution cameras are able to accurately capture micro facial expressions, such as the slight movements of the eyelids and eyeballs, and the contraction of the muscles of the lips and surrounding cheek skin.

Using specific physical characteristics, and a combination of physical characteristics, the vivisection system can be very good at distinguishing between live and fake objects.

Infrared face detection is mainly based on optical flow method.

Optical flow method uses the time-domain variation and correlation of pixel intensity data in image sequence to determine the "motion" of each pixel position, that is, the operation information of each pixel point is obtained from the image sequence, and the data is statistically analyzed by using Gaussian difference filter, LBP feature and support vector machine.

At the same time, the optical flow field is sensitive to the movement of objects, and the eye movement and blinking can be detected uniformly by the optical flow field.

This method of in vivo detection can be blind test without the user feeling.

The in vivo detection scheme of 3D depth camera is mainly based on extracting the 3D information of N (256 recommended) feature points of living and non-living face regions, and conducting preliminary analysis and processing of the geometric structure relationship between these points.

The raised areas were extracted from the depth image according to the curvature of the surface, and the EGI features were extracted for each area, and then the spherical correlation was used for reclassification and recognition.

Verify identity with a human face

In vivo detection technology combined with face detection and face comparison technology can achieve a more reliable Internet authentication solution.

Verify identity with a human face refers to the user through a selfie video or a selfie, and citizenship ID photo 1:1 face verification to confirm the user's identity.

And a vivisection test to confirm whether the current user is a person and a real person.

Facial recognition products are widely used, such as annual social security personnel identity verification, online lending system real-name authentication, online insurance sales real-name authentication, remote bank/securities account opening.

From: www.zyjjhome.com

Gouku Technology, focus on human biometric identification products research and development and manufacturing, committed to the development of the best fingerprint module and face recognition module.

e-mail: mark@zyjjhome.com