Research on facial recognition systems began in the 1960s. After the 1980s, it improved with the development of computer technology and optical imaging technology. However, it was not until the late 1990s that it truly entered the initial application stage, mainly implemented by the technologies of the United States, Germany and Japan.

Face recognition specifically refers to computer technology that uses the analysis and comparison of facial visual feature information for identity verification. It belongs to biometric recognition technology and distinguishes individual organisms (generally referring to humans) based on their own biological characteristics. First, determine whether there is a human face. If there is a human face, further provide the position, size of each face and the position information of each major facial organ. Based on this information, the identity features contained in each face are further extracted and compared with the known faces to identify the identity of each face.

1. Analysis of the Principle of Face Recognition Technology

face recognition is mainly divided into three processes: face detection, feature extraction and face recognition.

Face detection: Face detection refers to the detection and extraction of face images from input images. Usually, haar features and the Adaboost algorithm are used to train a cascade classifier to classify each piece in the image. If a certain rectangular area passes through the cascade classifier, it is discriminated as a face image.

Feature extraction: Feature extraction refers to representing facial information through certain numbers, and these numbers are the features we need to extract. Common facial features are divided into two categories: one is geometric features, and the other is representational features. Geometric features refer to the geometric relationships among facial features such as eyes, nose and mouth, such as distance, area and Angle, etc. Because the algorithm utilizes some intuitive features, the amount of calculation is small. However, due to the inability to precisely select the required feature points, its application scope is limited. In addition, when the lighting changes, the face is blocked by external objects, or the facial expression changes, the features change significantly. So, this type of algorithm is only suitable for rough recognition of face images and cannot be applied in practice.

2. Application prospects of facial recognition technology

1. Enterprise and residential safety and management. Such as facial recognition access control and attendance systems, facial recognition anti-theft doors, etc.
2. Electronic passport and ID card. This might be the largest application in the future. The International Civil Aviation Organization (ICAO) has determined that from 2010, all 118 member countries and regions must use machine-read passports, with facial recognition technology being the first recommended recognition mode. This regulation has become an international standard.
3. Public security, justice and criminal investigation. Such as using facial recognition systems and networks to hunt down fugitives across the country.
4. Self-service. For instance, in an ATM of a bank, if a user's card and password are stolen, the cash can be fraudulently withdrawn by others. If face recognition is applied simultaneously, this situation can be avoided.
5. Information Security. Such as computer login, e-government and e-commerce.

3. Smart Bank Facial Recognition Solution

The facial recognition solution for smart banks centers on user experience and relies on technical support such as intelligent dynamic facial recognition ID verification management systems, facial recognition attendance record systems, exclusive identity management systems, access control management systems, and video early warning systems to create a safer and more convenient smart financial service system. Realize a series of exclusive services such as counter real-name account opening, remote real-name account opening, real-name payment, etc., as well as intelligent access control, intelligent attendance, visitor record, VIP identification, and precise push of financial information advertisements, to ensure the security of fund transactions, optimize management, facilitate business, and provide considerate services. This is for banks to expand business areas, reshape service processes, and improve service quality. An important means to enhance market competitiveness.

I. Overview of the Plan

VIP customers are an important customer resource for each bank. Every bank is constantly innovating services to attract VIP customers to open deposit, loan, investment and wealth management services, etc. New services are easy to launch, but how to better serve these VIP customers and improve the retention rate is a difficult problem that banks are constantly thinking about at present. During the service process, when it comes to the identification and welcoming of VIP customers, it is still basically at the stage of passive waiting. How to identify those VIP customers who have made significant contributions to the bank at the first time and provide them with more considerate services is the key to doing a good job in VIP customer marketing and improving business quality.

Ii. Scheme Design

The facial recognition system for VIP customers in banks is an intelligent facial recognition system that utilizes facial biometric recognition technology. It sets up collection points in key areas of the bank, compares the collected facial information features of customers with the existing facial information features of VIP customers, obtains successful results and notifies the corresponding personnel.

The VIP bank customer facial recognition system is divided into three parts: the collection client end, the branch service end, and the facial recognition system server. Architecture deployment can be divided into two types: centralized and distributed.

Centralized deployment involves storing the facial feature data of VIP customers in the authentication platform server of the bank's data center. The facial feature data of customers captured by the branch end is sent to the authentication platform server, where it is uniformly compared. This method features unified data management and low hardware investment costs, making it suitable for customers with relatively few service outlets and small amounts of VIP customer data.

Distributed deployment: The facial feature data of VIP customers is stored in the authentication platform server of the bank's data center. It supports automatic synchronization to the servers of each branch or the authentication servers at the municipal level at the same time every day. The facial feature data of customers captured by the branch terminals is directly compared on the branch servers or the authentication servers at the municipal level. Send the comparison results to the authentication platform server for record and preservation, and send the bank's message to notify the platform. This method features decentralized data storage and high hardware investment costs, making it suitable for customers with numerous service outlets and large amounts of VIP customer data.

Iii. Characteristics of the Plan

High accuracy, offline operation: World-leading algorithms completely solve cross-age issues and small image recognition problems. No need to connect to the police to access large images of certificates, it can still 100% identify the authenticity of certificates and whether they belong to the person.

Blacklist warning, access control: Pioneering blacklist warning and automatic recognition of the white list for door opening, effectively ensuring the personal and property safety of customers.

Hierarchical management, face query: Due to the small amount of collected data, there is no pressure on storage. It adopts a three-level storage system of front end, terminal and platform, which is convenient for quick query and data backup after the fact. When handling secondary business, it can be quickly identified.

Fast recognition, voice prompt: The original face recognition algorithm can determine whether it is the real person in as fast as 0.2 seconds. Combining multiple recognition modes to meet the usage requirements of different scenarios. It can be combined with additional voice announcements to make it easy for the identifiers to operate and pass through quickly.

Visible light, multi-person recognition: Visible light face recognition technology based on deep learning has low environmental requirements, can be used under various lighting conditions, conforms to human eye habits, and can recognize more than 10 people at the same time.

System networking and data analysis: Mature product system-level application solutions make every identification device a data collection terminal, providing effective data for big data analysis, event early warning, and accident prevention.

Iv. System Functions

1. The system is equipped with a complete set of sub-management modules, including organization management, personnel management, parameter management, menu management, and customer management.

2. The system supports 1:1 comparison service and 1-n comparison service.

3. Supports multiple interface methods such as: Webservice interface method or Socket interface method;

4. The software is easy to operate, has a friendly display interface, and supports continuous and stable application for 7*24 hours.