CN106887059A - A kind of intelligent electronic lock system based on face recognition - Google Patents

Abstract

The present invention is based on intelligent equipment, mainly uses the camera to obtain the image information of the target, extracts the facial image of the person for the dynamic target, and links it with the door lock by analyzing the facial features of the target and the database. Image comparison to determine whether the intruder is an unknown person from outside, so as to realize the intelligent switch of the door lock. The present invention improves the problems of high cost and poor environmental adaptability in the current face recognition access control system, adopts ARM Cortex‑M3 embedded chip design, reduces hardware cost and operating power consumption, and improves hardware performance. At the same time, based on the MATLAB image processing platform, an improved 2DPCA face recognition algorithm is proposed. At the same time, the algorithm based on the skin color feature is used to locate the face information, reduce the influence of the environmental background, and improve the face recognition rate. Compared with the traditional 2DPCA algorithm, the improved algorithm proposed in this paper takes less time. At the same time, the success rate of face matching reaches 87%, the misjudgment rate is lower than 3%, and the system security reaches more than 97%.

Description

An intelligent electronic door lock system based on facial recognition

technical field

The invention relates to an intelligent identification device, in particular to an electronic door lock system based on facial recognition.

Background technique

With the development of social economy and technological innovation, people have gradually reached a high living standard in a well-off society, and their awareness of security is constantly improving. Therefore, how to use new technical means to design a safer, more convenient and reliable security system and improve the quality of life of residents has become the focus of attention in the field of intelligent security. Among them, the door lock, as the first checkpoint in the intelligent security system, is the top priority of the entire security system. The identity verification methods of the traditional door lock system include: key, password, magnetic card, etc. However, these verification methods are separable from the user and are easy to be copied, deciphered and embezzled, which cannot fully meet the modern security concept. Nowadays, human biometrics such as fingerprints, faces, irises, etc. have been widely used in the field of intelligent security because of their uniqueness, invariance and inseparability. Therefore, designing an intelligent door lock identification system for security and defense has important economic value and practical significance for the development of society.

Contents of the invention

In today's society, intelligent security equipment based on biometric technology is becoming more and more popular. Due to the characteristics of uniqueness, invariance and inseparability of biometric technology, it is an important feature that other technologies lack. Facial recognition is also one of the widely researched topics. In recent years, the smart door lock system based on facial recognition has also begun to heat up. However, because its market price is very expensive, and the recognition success rate is greatly affected by factors such as the environment and light, it is still very difficult to promote. Compared with other security systems, its market Occupancy rate is slightly low, but the development space is huge.

The present invention is realized through the following technical solutions: adopting ARM Cortex-M3 embedded chip design, reducing hardware cost and operating power consumption, and improving hardware performance. At the same time, based on the MATLAB image processing platform, an improved 2DPCA face recognition algorithm is proposed. In the sample training stage, different feature subspaces are constructed for each type of person, and in the face image recognition stage, the image to be recognized is projected to each Matching is carried out in the feature subspace of a class of people, and the maximum matching degree is found to determine the belonging of the person to be recognized. At the same time, the algorithm based on skin color features is used to locate facial information, reduce the influence of environmental background, and improve the facial recognition rate.

On the basis of facial recognition, the equipment is used to identify and analyze the people outside the door, and then compare it with the photos of people who can pass through, and then judge whether they are consistent, and then determine whether to unlock and allow entry.

An electronic door lock system based on facial recognition, including stm32 main control module, camera, serial port communication module, human body infrared sensor module, storage module, electric lock drive module, display module, sound and light prompt module, power conversion module, power supply switch, remote control receiving module, sound and light alarm module, camera and motor drive module, it is characterized in that when the door lock system is working, other modules transmit information to the main control module, which is fed back to each module after being processed by the main control module, and executes corresponding Work. This hardware system has strong adaptability and transplantation characteristics. It can be transplanted in different environments without changing the hardware system. It can be expanded into a door lock system in different environments only by modifying the software parameters.

Preferably, in the hardware system, the camera module, serial port module, human body infrared sensor module, TFT-LCD module, etc. transmit information to the main control module, and after being processed by the main control module, the information is fed back to each module to perform corresponding work.

Preferably, the hardware system includes nine components including an electronically controlled lock drive module, a TFT-LCD display module, an audible and visual prompt module, a DC power conversion module, a camera module, a CH340G serial communication module, a human body infrared sensor module, and an SD card storage module. Composed of hardware modules.

Preferably, the software system program part is mainly divided into two parts, MATLAB program design and STM32 master control program design.

Preferably, the MATLAB program design can be divided into four major blocks such as serial port communication, face detection, face matching, and user database management, so that the upper computer can design the monitoring software and monitoring program of the facial recognition intelligent door lock system based on the MATLAB software The interface is simple and clear. Users only need to click a few buttons in the program interface to realize real-time monitoring of the system, and form a complete facial recognition intelligent door lock system with the lower computer STM32.

Preferably, the STM32 main control program design corresponds to each hardware module of the hardware system, including a face positioning camera program, a motor lock driver module program, a TFTLCD display module driver program, an SD card storage module program, a human body infrared detection program, CH340 communication module program and low power mode program, etc.

The working principle of the invention is: the work of each sub-module is regulated by the main control module, and the modules work in coordination with each other. Firstly, through the human body infrared sensing module, it is detected in real time whether someone is approaching the door lock system. If a human body signal is detected, the main control chip in the standby state is activated, and then the OV7670 positioning camera module completes the initialization and prepares to acquire facial images. Using TFT-LCD Assisted by the LCD screen, it can display the position of the facial image in real time, which is convenient for the user to locate and obtain the avatar. At the same time, it is accompanied by Chinese prompts on the TFT-LCD display screen to remind the user of the current task execution status. Each time the face recognition door lock system is activated, there are three recognition opportunities. If the three recognitions fail, the system will brake and enter the standby state, waiting for the next Once the human body infrared sensor module reacquires the trigger signal, the motor lock drive module mainly receives instructions from the STM32 main control, and executes the door opening and closing procedures. The USB high-definition camera is the main camera for facial image acquisition. After the facial headshot is taken, it is sent to the MATLAB program on the PC for facial recognition. The SD storage module will store the pictures obtained for each recognition in real time, so as to preserve evidence and view records at any time.

Compared with the prior art, the present invention has the following beneficial effects:

1. According to the system analysis and design goals, complete the hardware system design based on the ARM Cortex-M3 main control, implement facial positioning, TFT-LCD intelligent prompts, and switch lock functions.

2. Based on the intelligent monitoring software designed on the MATLAB image processing platform, the facial detection based on skin color features and the improved 2DPCA facial recognition algorithm are completed. At the same time, it can manage the user database in real time and view the monitoring records.

3. Good human-computer interaction interface design. By learning the GUI design of MATLAB, complete the design of the monitoring system software interface, including optimizing buttons, simplifying the operation sequence, and optimizing the display frame.

4. The low power consumption mode of the hardware system. The system will be activated only when the infrared sensor of the human body is triggered. After three identification failures, the system will automatically enter the low power consumption standby mode. consumption settings.

Description of drawings

Figure 1 shows the overall composition framework of the hardware system.

Figure 2 is the system hardware connection diagram.

Fig. 3 is the overall program flow chart of the monitoring software of the intelligent door lock system based on facial recognition.

Figure 4 is the interface diagram of the monitoring software of the smart door lock system based on facial recognition.

Figure 5 is the operation diagram of the door lock system.

detailed description

The present invention will be further described in detail below in conjunction with the embodiments and the accompanying drawings, but the embodiments of the present invention are not limited thereto.

Example

See Figures 1 and 2, the overall composition framework of the hardware system and the system hardware connection diagram, the STM32F103ZET6 in the STM32 MCU enhanced series products, it is a 32-bit micro CPU based on the ARM Cortex-M3 core, with a processing frequency of 72MHz and up to 64KB SRAM And 512KB FLASH, rich timer resources, including 2 advanced timers, 4 general-purpose timers, 2 basic timers, communication interface includes 5 serial ports, 1 USB interface, 3 SPI interfaces, 2 IIC In addition, there are 1 12-bit DAC, 3 12-bit ADCs, 1 FSMC interface, etc. The chip has a total of 144 pins, of which 112 are general-purpose IO ports, and adopt the LQFP-144 package structure. The electric control lock drive module adopts TB6612FNG DC motor drive chip, its MOSFET-H bridge structure has high current drive capability, and has dual-channel output, which can drive two DC motors simultaneously through PWM. Each channel can output up to 1.2A continuous drive current, which can control DC motor forward/reverse/stop/brake, supports up to 100kHZ PWM frequency, has on-chip low-voltage detection and thermal shutdown protection circuit, operating temperature: -20～ 85°C, using SSOP24 chip package. The LCD screen adopts ALIENTEK's 2.8-inch TFT-LCD. The LCD screen comes with an LCD driver chip IL9320, powered by a 5V power supply, with a working current of 130mA~350mA, a resolution of 320*240 and 16-bit true color display. Therefore, the power consumption of this display screen is low, which meets the design requirements. In this design, dual cameras are used to collect data. The face positioning camera uses the ALIENTEK OV7670 camera module, and the face acquisition camera uses a USB high-definition camera. The face positioning camera OV7670 is directly controlled by the STM32 microprocessor, and the images it acquires are displayed on the TFT-LCD in real time, which is convenient for users to locate their own face positions. The face acquisition camera is controlled by the Matlab software of the PC, and its main function is to acquire facial images for recognition processing. The requirements for the installation of the two cameras are that the obtained external images should be the same, and the installation height, inclination angle, and environment should be consistent. The sound and light prompt module includes a buzzer circuit and a common anode red and green dual-color LED light circuit. When the hardware system is activated, the buzzer will give a short beep to remind the user. When the face recognition is successful, the buzzer will also beep once. At the same time, the green light is on, indicating that the verification has passed, and the system executes the unlock command. When the facial recognition fails, the buzzer will beep twice, and the red light will be on at the same time, indicating that the verification failed, and the user needs to re-authenticate. The RCW-0506 human body infrared sensor module used in this design is an automatic control module based on infrared technology. It adopts the probe LHI778 imported from Germany. It works with ultra-low power consumption, small size and high sensitivity. It is widely used in security products, human body sensor lighting, industrial Automatic control, etc. This design also adopts battery power supply mode. The standby current of RCW-0506 human body infrared sensor module is less than 50uA, which is very suitable for battery power supply circuit. The hardware system is in standby state during non-working hours. At this time, the infrared module will be the only active monitor. , Its low power consumption performance makes the standby low power consumption of the whole system remarkable, making it possible to use dry batteries for power supply in this design.

Refer to Figure 3 and 4, the overall program flow chart of the monitoring software of the intelligent door lock system based on facial recognition and the interface diagram of the monitoring software of the intelligent door lock system based on facial recognition. The port is connected, then start the monitoring program on the MATLAB side, click the program initialization button, and wait for the initialization to complete. At this time, the real-time camera screen will appear in the facial image collection frame, and then click Run Monitoring, the program will enter real-time monitoring. Start the hardware part of the door lock system, and the STM32 will enter the standby state. Only the human body infrared sensor module maintains normal detection work. When someone enters the infrared sensor module detection range and is continuously detected for more than 2 seconds, the STM32 main control module will be woken up. , it takes about 2 seconds to initialize the hardware system. After the initialization is completed, the LCD screen will prompt the user to locate the face head and prepare to obtain the face picture for recognition. When the user's face is located, the OV7670 camera and USB HD camera will capture the user Locate the avatar, after the photo is captured, the buzzer will give a short beep, and the LCD will refresh and display the user avatar just captured. At this time, MATLAB on the PC will execute the face detection and facial recognition program, and the LCD will display "Recognizing, please wait... " and other statements, wait for the command after the PC end has completed the recognition. If the recognition result is a legitimate user, the STM32 will control the electric lock drive module to perform the door opening action after receiving the door opening command. Otherwise, the STM32 will receive the door lock command and the LCD will display " Recognition failed, please try again!" At this time, the user needs to reposition the face and take a photo again for detection. If the three recognitions fail, the STM32 system will enter the standby mode, and the human body infrared sensor module will also enter a 10-second lockout period. During this period of time, the infrared sensing module of the human body will no longer detect external signals, and the user can only wait for the next trigger time to take pictures and identify them again.

In addition, users can perform various database management operations on the face recognition intelligent door lock system monitoring software on the PC side: for example, to view the status of current legal users, click "View Legal Users", and a pop-up window will automatically display the avatar pictures of all current legal users. When you need to register a new user, click the "Register New User" button, and a pop-up window will prompt "Continuously register 50 face pictures, please click 'Confirm' to continue." At this time, the program will call the USB HD camera to take pictures of the new user and build a picture library, about 30 seconds after the image is acquired, a pop-up window will prompt "The new user information has been entered, please click 'OK' to continue". At this time, the new user is already in the facial recognition system. Click the program initialization again to update the legal user Information, when you need to delete a legitimate user, click the "Delete User" button, the system will pop up the root directory of the user's picture library, just "delete" the character you want to delete. The above is the function of the user database management module. In addition to the addition, deletion, and modification of user data, the system also sets up a "view record" button, which is convenient for checking the user's entry record, and can also be used as a home monitoring record. Door records, which can be used as one of the important materials to solve the case.

PCA-based facial recognition algorithm: This paper mainly uses PCA (Principal Component Analysis) for facial recognition, which is a method based on multivariate statistics. PCA mainly uses the idea of dimensionality reduction to study how to use a few principal components to explain the variance of multiple variables, where each principal component can reflect most of the information in the original variable, and the information contained is not related to each other. PCA is also commonly used in pattern recognition technology, data compression technology and data feature extraction technology because principal component analysis converts data into low-dimensional and makes people more intuitive to see the data structure.

The basic principle of the traditional PCA method is: use K-L transformation to extract the main feature components in the face, construct the eigenface space matrix, and project the image to be recognized onto the feature space during face recognition to obtain a set of projection coefficients, and then pass each Individual facial features are compared and identified. Using this de-correlation coordinate transformation method, the mean square error error before and after image compression is minimized, and it still has a good recognition ability after being transformed into a low-dimensional space.

The facial recognition method in this paper is improved based on the traditional 2DPCA algorithm. The traditional 2DPCA constructs a feature subspace with different facial images in the training phase. During facial recognition, the facial image to be recognized is projected 23 onto the feature subspace. Compare. When there are many different types of persons in the sample base, if different types of persons are jointly constructed into a feature subspace, the intra-class Euclidean distance is relatively small, the Euclidean distance between different classes is relatively large, and the recognition threshold T is determined according to the maximum Euclidean distance between samples, so the recognition threshold T is often too large, resulting in misjudgment. Therefore, this paper proposes a 2DPCA face recognition algorithm based on multiple feature subspaces. In the sample training stage, different feature subspaces are constructed for each type of person, while in the face image recognition stage, the to-be-recognized The image is projected into the feature subspace of each type of person for matching. If the matching is successful, the image to be recognized belongs to this type of person, otherwise the image to be recognized is considered to be an illegal person.

Figure 5 is the operation diagram of the door lock system.

The above is a preferred embodiment of the present invention, but the embodiment of the present invention is not limited by the above content, and any other changes, modifications, substitutions, combinations, and simplifications that do not deviate from the spirit and principles of the present invention are all Replacement methods that should be equivalent are all included within the protection scope of the present invention.

Claims (7)

1. An electronic door lock system based on facial recognition, including stm32 main control module, camera, serial port communication module, human body infrared sensor module, storage module, electric lock drive module, display module, sound and light prompt module, power conversion module , power switch, remote control receiving module, sound and light alarm module, camera and motor drive module, characterized in that, when the door lock system is working, the camera module, serial port module, human body infrared sensing module, TFT-LCD module, etc. transmit information to the main control The module, after being processed by the main control module, feeds back to each module and performs corresponding work. This system has strong adaptability and transplantation characteristics. It can be transplanted in different environments without changing the hardware system, and only needs to modify the software parameters. It can be expanded into a door lock system in different environments. 2. The electronic door lock system based on facial recognition according to claim 1, characterized in that, using the STM32F103ZET6 chip with the core ARM Cortex-M3 as the framework, compared with the embedded chip, its execution code efficiency is higher, and it supports external Memory expansion, support for low power consumption modes, etc. 3. The electronic door lock system based on facial recognition according to claim 2, characterized in that the system uses an electronically controlled male lock, because the door lock system involves a large number of hardware modules and peripheral devices, which need to be integrated into the door The inside of the lock is designed with an electronically controlled male lock solution, which avoids cumbersome wiring and installation procedures, and is more portable. The embedded system is fully integrated into the door lock, and the user only needs to replace the original lock body when installing. 4. the electronic door lock system based on facial recognition according to claim 1, is characterized in that, this system has adopted double camera to gather data, and wherein facial positioning camera adopts ALIENTEK OV7670 camera module, and facial acquisition camera adopts USB high-definition camera. 5. The electronic door lock system based on facial recognition according to claim 1, characterized in that the RCW-0506 human body infrared sensor module used is an automatic control module based on infrared technology, and the German imported probe LHI778 is used, which is ultra-low power Power consumption, small size, high sensitivity. The sound and light prompt module includes a buzzer circuit and a common anode red and green dual-color LED light circuit. When the hardware system is activated, the buzzer will give a short beep to remind the user. When the face recognition is successful, the buzzer will also beep once. , at the same time the green light is on, indicating that the verification has passed, and the system executes the unlock command. When the face recognition fails, the buzzer will beep twice, and the red light will be on at the same time, indicating that the verification has failed and the user needs to re-authenticate. 6. The electronic door lock system based on facial recognition according to claim 1, characterized in that, based on the facial detection method based on skin color features, by analyzing the color features, it is found that under the constraints of YCgCr and YCgCb color spaces, it can be well Satisfy the skin color detection of Asians, and under the YCgCr and YCgCb color spaces, the influence of brightness is relatively small, and the face position can be quickly detected even when the face is in a complex background. . 7. The electronic door lock system based on facial recognition according to claim 4, characterized in that, an improved 2DPCA facial recognition algorithm is proposed, and in the sample training stage, different feature subspaces are constructed for each class of characters respectively, and In the facial image recognition stage, the image to be recognized is projected into the feature subspace of each type of person for matching, and the maximum matching degree is found to determine the identity of the person to be recognized.