Design of fingerprint recognition access control system based on ARM

Abstract: Traditional identity recognition technology can no longer meet the needs of modern society, so fingerprint recognition technology has been widely used. This article introduces the principles and processing methods of fingerprint recognition, and points out a software and hardware design method for a fingerprint recognition access control system based on the embedded ARM9 architecture.
Keywords: ARM; fingerprint recognition; S3C2440AL; TFS-D0303

0 Introduction With
the rapid development of modern society, identity confirmation is required in many occasions, and traditional identity recognition technology can no longer meet social requirements. Human physical characteristics are non-replicable, so people began to study biometrics, and fingerprints have the characteristics of uniqueness, lifelong invariance, and difficulty in forging, and are highly secure, so they have been widely used. In some confidential departments, such as banks, hotels, computer rooms, etc., access control systems are generally installed. Access control systems are systems that manage and control the entrances and exits of important passages to ensure the safety of people's lives, work and property. Access control systems based on fingerprint recognition technology are high-tech security facilities that improve the security of the system. As an embedded system processor, ARM has the characteristics of high performance, low power consumption, and low cost, so it has been widely used in the field of industrial control, imaging and security products. This paper introduces the fingerprint recognition principle and processing method based on embedded ARM9 architecture, as well as the software and hardware design method of fingerprint recognition access control system.

1 Fingerprint recognition principle and processing method
Fingerprint recognition technology analyzes the local features of fingerprints and extracts detailed feature points from them, so as to reliably confirm personal identity. Fingerprint recognition technology mainly involves four functions: reading fingerprint images, extracting features, saving data and comparing. First, find the "feature points" from the acquired fingerprint image, and then establish the fingerprint feature data of the user's live fingerprint based on the characteristics of the feature points (a one-way conversion that can be converted from fingerprint images to feature data but not from feature data to fingerprint images). Since two different fingerprints will not produce the same feature data, the feature data of the collected fingerprint image and the fingerprint feature data stored in the database are pattern matched to calculate their similarity, and finally the matching results of the two fingerprints are obtained, and the user's identity is identified based on the matching results.

1.1 Fingerprint image acquisition
Optical imaging has a long history, which can be traced back to the 1970s and is based on the principle of total reflection of light. Optical fingerprint sensors are generally used to collect fingerprints. The fingerprints obtained have the characteristics of clear images, low power consumption, and high stability. This system uses the TFS-D0303 optical fingerprint sensor.
1.2 Implementation of fingerprint recognition algorithm
The effectiveness of the fingerprint recognition algorithm directly affects the accuracy of fingerprint recognition and the security and stability of the fingerprint recognition access control system, and plays a decisive role in the security and reliability of the entire access control. The processing process of the fingerprint recognition algorithm mainly includes: image processing, texture refinement, feature extraction and feature matching (see Figure 1).

Preprocessing is an indispensable step in fingerprint recognition. Its purpose is to remove the noise introduced when the image is acquired and eliminate the influence of low-quality images so that the subsequent links can correctly perform feature extraction and classification recognition. The preprocessing process mainly includes steps such as image segmentation, smoothing, image enhancement, binarization and thinning. Each step can improve the quality of the image and facilitate the subsequent work.
The influence of ambient temperature or dry fingers may cause discontinuity of the lines in the fingerprint image. For fingerprint images with discontinuous lines, smoothing filtering is generally required, which is also to blur the image so that the broken lines are connected after blurring. In this algorithm, a low-pass filter is used to smooth the direction information of each block of fingerprint images at the block level to correct inaccurate calculation results.
The specific method is to first calculate the projection components of 0(x, y) on the X-axis and Y-axis:

where H(u, v) is a two-dimensional low-pass filter; w1×w1 is a 5×5 filter size; w is the image sub-block size (this algorithm takes w=10 pixels).

After the grain thinning, the grain lines of the binary image are thinned into connected line segments with a width of only one pixel. The original fingerprint image, the enhanced binary image, and the thinned binary image are shown. The feature point information of the fingerprint is extracted from the thinned binary image. The bifurcation points or endpoints are found from the thinned binary image, and then the grain trajectory is found from these points. Through these points, the shape of the grain is calculated. These shape data, point types, and point positions are recorded as feature points of the fingerprint image.
Feature matching is determined based on the maximum matching point support number of the feature points of the two images. Assuming there are two images A and B, first, take each feature point of image A as the reference, and calculate the distance and direction of each feature point in image B that is the same as the reference feature point in image A to each feature point in image B. Based on this distance and direction, take the reference point in image A as the origin, calculate a point in image A, and then determine whether it is also a feature point of image A. If so, the matching support number is increased by 1. Calculate in sequence according to the feature points in Figure A, and the maximum value is the maximum support matching points of the two images. If the maximum matching support points are greater than the specified value, the two images are considered to match. Otherwise, they do not match.

2 System Hardware Design
ARM microprocessors based on RISC architecture are widely used in many fields due to their small size, low power consumption, and fast execution speed. This system uses Samsung's embedded microprocessor S3C2440AL, which is a high-performance, low-power, and powerful embedded application processor product. It uses the Intel X-Scale microarchitecture system framework and integrates many commonly used peripheral interfaces. It is powerful. The main frequency of S3C2440AL is 400MHz, and the highest is 533MHz; TFS-D0303 fingerprint sensor consists of 256×300 capacitive sensor arrays, with a resolution of up to 500dpi, an operating voltage range of 3.3~5V, an 8-bit ADC inside the sensor, and 2 sets of sampling and holding circuits. The block diagram of the entire hardware system is shown in Figure 2.

The system memory interface uses 128MB SDRAM memory, supports 16, 64, 128, 256MB DRAM technology, 4 SDRAM areas, each area supports 64M memory. The clock is enabled (a CKE pin is used to set the entire SDRAM interface to self-refresh), supports up to 6 static memory devices (SRAM, Flash, ROM), and supports 2 PCMCI/CF slots. The clock uses a 3.6864M-Hz oscillator with a core PLL and a peripheral PLL to generate various operating frequencies. The 32.768kHz oscillator can drive the real-time clock, power manager and interrupt controller. The power controller can control fast/run, idle and sleep working modes. The LCD controller supports passive (DSTN) and active (TFT) LCD displays, with a maximum resolution of 800×600×16, and 2 dedicated DMA channels, allowing the LCD controller to support single-layer or dual-layer displays. The real-time clock (RTC) can generate periodic interrupts to wake up the application processor from sleep. The serial port communication USB slave module complies with the USB specification and supports version v1.1. It supports up to 16 endpoints and provides a 48MHz internal clock.

3 System software design
The software design of this system mainly includes the following processes: power-on system initialization, fingerprint recognition, and control of electric lock opening. The accuracy of the system depends largely on the fingerprint recognition algorithm. After the program is initialized, it enters the low-power mode and waits for various specific operations. When fingerprints are collected, it enters the fingerprint data collection and processing module. After the processing is completed, it returns to the low-power waiting mode and waits for other operations. Similarly, when there is card reading data or clock setting response, it enters the corresponding operation module for processing.
3.1 Initialization of S3C2440AL
System initialization For embedded systems, there must be system initialization code in its program memory (as shown in Figure 3). After the system executes the initialization code, other applications can be executed correctly. The initialization is automatically completed after the system is reset. The initialization of the system must include the following initialization codes, including setting the interrupt vector table, initializing registers, initializing stack pointer registers, initializing ports, and changing the working mode of the processor. The system can only open interrupts in user mode. In the initialization of the fingerprint sensor, the control registers should be initialized step by step as required.

3.2 Fingerprint collection
Fingerprint collection can work in interrupt mode or query mode. This system adopts the query mode. The program flow is as follows: first initialize each register, write the control word to the corresponding register, set the parameters for collecting fingerprints, and when the fingerprint is automatically collected and enters the data register, the fingerprint data is stored in the specified storage space.
3.3 Communication with the computer
RS232 serial port is used for communication between S3C2440AL and the host computer. When there are too many images to store, the images can be stored on the host computer. The UART of S3C2440AL can work in interrupt mode or DMA mode. This system uses interrupt mode, and can also use query mode. The program can control the port baud rate, data width (5, 6, 7, 8 bits), stop bit (1 to 2 bits), polarity control and other communication protocol settings.

4 Conclusion
This article introduces the software and hardware design of the fingerprint recognition access control system based on ARM. The ARM architecture has a wealth of interfaces and is playing an increasingly important role in the fields of access control and monitoring. Specific embedded systems that are integrated and portable are increasingly appearing in the field of engineering applications. The fingerprint recognition access control system implemented in this paper has the advantages of being lightweight, easy to install, and low-cost, and has good development prospects.