Application of facial recognition in elevator registration and access control
ApplicationofFacialRecognitioninElevatorRegisterAccessManagement
He Xiaohu and Sun Entao
Shanghai STEP Electric Co., Ltd.
【summary】
Facial recognition has been used in the security industry for many years. With the development of facial recognition technology, its application scope has expanded further, especially with deep learning technology driving a leap forward in both accuracy and precision. Elevators, as essential vertical transportation tools in buildings, maintain a growth rate of 5% to 10% annually. Different applications have different requirements for elevator access control. Important buildings, such as financial institutions, government buildings, and high-end private residences, have stricter requirements for elevator access control. Traditional elevator registration and access control methods, such as IC cards, have many limitations due to inherent technological drawbacks. Applying facial recognition technology to elevators can greatly improve recognition accuracy, reduce the risk of hacking, and ensure genuine elevator registration and access control.
Keywords:
Facial recognition elevator registration access management
1. Project Background
With economic development, many high-end residential communities and intelligent property management systems exist in China. However, elevator access control in these communities still relies on traditional registration and access management methods. Currently, mainstream IC card-based intelligent elevator management solutions in China have significant security vulnerabilities and drawbacks, such as the ease with which IC cards can be lost, counterfeited, or stolen, thus limiting their widespread adoption. To address this, a more scientific and secure elevator registration and access control solution needs to be developed. This solution should integrate automatic control technology, network communication, and sensor technology to comprehensively manage elevator usage and security, ensuring excellent confidentiality and safety, and providing uninterrupted usability, thus becoming an important component of smart communities.
Facial recognition technology is a biometric identification technology based on facial features. It involves using cameras or webcams to capture images or video streams containing faces, automatically detecting and tracking faces within the images, and then performing facial recognition. This process is also commonly referred to as image recognition or facial identification. Deep learning-based methods have become a significant development trend and direction in the field of facial recognition technology, leading to its widespread application in elevator access control. Furthermore, with technological advancements and the increasing prevalence of applications, the cost of building large-scale, distributed facial databases and recognition systems is continuously decreasing, while the accuracy of recognition is constantly improving.
During the market promotion and application process, the following advantages of facial recognition in elevator registration and access management were found:
(1) Contactless, the user does not need to have direct contact with the device;
(2) Non-mandatory; the facial image information of the person being identified can be actively obtained;
(3) Concurrency, that is, in actual application scenarios, multiple faces can be sorted, judged and recognized;
(4) Not easy to counterfeit, no risk of theft; can be used and swiped at will;
However, there are also many challenges:
(1) It is sensitive to the surrounding light environment, which affects the accuracy of recognition;
(2) Factors such as hair, ornaments, etc. that obscure the human face, and the aging of the face, need to be compensated by artificial intelligence (e.g., by recognizing some key features of the face).
(3) When used in elevator control systems, the acquisition of facial images is easily affected by factors such as ambient light, angle, and distance, which can cause instability in facial recognition elevator control.
However, with the development of facial feature recognition technology, local key part recognition technology, machine learning technology, etc., the impact of the above drawbacks is becoming smaller and smaller.
The application of facial recognition technology in the elevator registration and access management system (hereinafter referred to as the "system") can turn public elevators into private ones, greatly enhancing the security of property management, improving management level and efficiency, changing the image of property management, and making it more humane; reducing elevator operating costs, reducing elevator energy consumption, reducing the number of elevator maintenance and repairs and costs, and extending the service life of elevators; controlling personnel entry and exit is uncopyable and highly secure.
2. System Design
The facial recognition elevator registration and access control system consists of a facial recognition all-in-one machine and a facial data collector installed in the elevator car. Suitable for high-security environments, it requires users on each floor to verify their identity before use, effectively preventing unauthorized entry and improving community security.
Elevator registration access management includes in-car registration access management and out-of-hall registration access management. This article only discusses elevator registration access management using facial recognition control inside the elevator car.
2.1 System Components
The elevator facial recognition registration and access control system mainly consists of a facial recognition terminal, an infrared sensing unit, a protocol conversion board, and an elevator control system. The facial recognition terminal comprises a camera, an image processing unit, a logic processing unit, and a protocol processing unit. The infrared sensor determines whether there is someone in the elevator car, and its signal directly enters the facial recognition terminal. The protocol conversion board converts the RS232/RS485 protocol of the facial recognition terminal into the CANBUS protocol and converts elevator data sent by the elevator control system into RS232 protocol packets for transmission to the facial recognition terminal, ultimately enabling data interaction between the facial recognition terminal and the elevator control system.
The protocol conversion board receives CANBUS data from the elevator control system in real time, including the elevator's basic status, such as the direction of travel, the physical floor the elevator is on, and the elevator's operating mode (automatic, driver-operated, locked, independent, fire-fighting, etc.). Simultaneously, the protocol conversion board receives the registration status of in-car commands to confirm whether the commands registered by the facial recognition terminal have been successfully received and executed. In the communication link with the facial recognition terminal, the protocol conversion board periodically transmits the elevator's basic status and command registration information obtained from the elevator control system to the facial recognition terminal in real time.
The facial recognition terminal dynamically scans facial images inside the elevator car. When a registration command or permission granting command is triggered, it notifies the protocol conversion board via the RS232 interface. After the protocol conversion board confirms receipt, it returns response data. Otherwise, the facial recognition terminal will continue to request the registration command or permission granting command at regular intervals until the response is successful or the request times out.
2.2 Working Principle
A facial recognition terminal is installed inside the elevator car, using a dynamic acquisition method to scan faces. Installed on the upper inner side of the car, the terminal scans a passenger's face as they enter. If the detection stage identifies the scanned object as a face, the recognition stage begins. In this stage, the terminal compares the facial information locally or via the network, temporarily storing the results in a comparison list. Simultaneously, using infrared sensing, if a person is detected in the car and no valid face is found in the comparison list, the facial recognition terminal sends a door-hold command to the elevator control system. The elevator main controller then controls the door operator to keep the elevator door open and issues a buzzer signal to remind the passenger to exit the car, as shown in Figure 2. When a valid face is detected, two methods are used for access registration: the first is to grant access to the corresponding floor. Since facial information is bound to the passenger's floor information during system entry, the system automatically grants access to the floor button corresponding to that face upon recognition, allowing the passenger to automatically register the in-car command button. The second method is automatic registration of the floor corresponding to the facial information, eliminating the need for manual registration by the passenger. The two methods are implemented using different protocols.
It is worth noting that when the elevator is in certain special states, such as fire return, firefighter, elevator lock, maintenance, or malfunction, the facial recognition terminal's registration authority function will automatically become invalid, and the elevator's registration authority will be completely revoked by the elevator control system in order to complete higher priority operations.
This project employs a feature-based approach for face recognition. This detection method can detect faces not only from existing facial features but also from their geometric relationships. In contrast to knowledge-based methods, it seeks invariant facial features for face detection. Currently, many experts have proposed methods that first detect facial features and then infer the presence of a face. Based on the extracted features, a statistical model is established to describe the relationships between features and determine the presence of faces. This project utilizes a professional face recognition API (Application Programming Interface) to extract various feature values of faces from images using relevant parameters (as shown in Table 1). The API returns results in the format shown in Table 2. The specific implementation algorithm is only applied and not designed in this project.
The return code indicates the result of this request. If the execution is successful, the face pose angle information is valid.
2.3 Facial Information Update Process
Facial recognition data collection is a prerequisite for facial recognition elevator registration and access control. The system needs to obtain information about authorized personnel beforehand, such as facial information, floor information, and other user information. This information must be accurately entered into the management system before use. Based on the different storage locations for personnel information, this project designs two solutions from a practical application perspective: local storage and network storage. The local solution, as the name suggests, stores user information and the comparison algorithm on the facial recognition terminal. When facial information needs updating, it is manually updated to the facial recognition terminal via a mobile storage device. This solution is less convenient due to the need for human intervention, and updating the facial recognition algorithm is also inconvenient. However, the local solution eliminates the need for additional network equipment, reducing usage and operating costs. The network solution, on the other hand, is superior to the local solution in terms of facial information collection and algorithm updates. The facial recognition terminal is only responsible for facial collection and recognition; the facial comparison operation is implemented by a distributed service platform, which returns the recognition results. Therefore, the facial recognition terminal does not require additional large-capacity storage units and high-performance comparison units, making it convenient to use and install.
3. System Parameters
3.1 Electrical Parameters
Since the facial recognition terminal is installed inside the elevator car, its performance must meet the environmental requirements of the car. Additionally, it must support the input of signals such as infrared sensors to determine whether the car is occupied. Furthermore, the communication interface and power supply voltage must be user-friendly and compatible. The electrical parameters of the facial recognition terminal in this project are as follows:
The face recognition terminal is powered by DC12V, and the maximum operating current of the main control board is <500mA.
The logic board has two dry contact outputs; the output characteristics of each contact are: on-resistance less than 5Ω; off-resistance greater than 10MΩ.
Output drive capability: AC125V0.3A/DC110V0.3A/DC30V1A; supports dry contact input for use as input of signals such as infrared sensing.
Operating environment: Temperature: 0 to ±60 degrees Celsius; Relative humidity: 20%-90% non-condensing; meets the environmental conditions for use in buildings.
Storage environment: Temperature: -10 to +90 degrees Celsius; Relative humidity: 20%-90% (non-condensing);
External communication interfaces: RS232, RS485, TCP/IP, connected to the elevator control system via protocol conversion.
3.2 Face Recognition Parameters
One of the key components of a facial recognition terminal is the camera (reader), and its performance directly determines the effectiveness of facial recognition. The parameters of the facial recognition reader used in this project are shown in Table 3.
The facial recognition system in this project offers a superior user experience. From capturing faces inside the elevator car and extracting features to completing the recognition and comparison, the response time is less than 1 second. The system can also be trained based on the frequency of residents' entry and exit to improve the pass rate of user feature recognition.
The backend has high load concurrency capabilities, supports load balancing, and has strong concurrency capabilities.
4. Conclusion
The application of facial recognition technology in elevator registration and access control can significantly improve community security. By authorizing passengers to use elevators through facial recognition, it provides a better solution for buildings with high security requirements. As an important component of security technology, facial recognition technology has gradually evolved from a system component to embedded, independent operation, and then to modularization and chip-based implementation. Its accuracy and anti-spoofing performance are constantly improving, providing strong technical support for the widespread application of facial recognition in elevators and becoming an important part of building safe and smart communities.
References:
[1] Chen Yaqian, Lei Kaibin. A review of face recognition technology [J]. Journal of Southwest University for Nationalities (Natural Science Edition), 2007.
[2] Zhao Minghua. Research on face detection and recognition technology. Doctoral dissertation of Sichuan University. 2006, 10.
Mailing address: He Xiaohu, No. 599, Meiyu Road, Jiading District, Shanghai, June 13th
Contact number: 15000770442
