Design and implementation of fingerprint recognition system based on ARM9

Biometric identification technology is the use of human body inherent physiological characteristics (such as fingerprints, face, red film, etc.) and behavioral characteristics (such as handwriting, sound, gait, etc.) to conduct personal identification.

Biometrics is more secure, more confidential and more convenient than traditional identification methods. Biometric identification technology has the advantages of easy forgetting, good anti-counterfeiting performance, easy forgery or theft, carrying "carry" and anytime anywhere.

The working principle of biometrics is to use biometric devices to sample biometric features, extract their unique characteristics and transform them into digital codes, and further make the code a feature template, and when people authenticate with the identity device, the recognition device acquires its characteristics and is compared with the feature template in the database. To determine whether a match is made, thereby deciding to accept or reject the person. In many biometric identification techniques used for authentication, fingerprint identification technology is the most convenient, reliable, non-invasive and inexpensive solution at present.

Fingerprint, as the most obvious appearance feature in human body, has the advantages of uniqueness, universality, uniqueness and easy collection. Fingerprint identification technology using human fingerprint stability and uniqueness of the physiological characteristics of people as a "live identity card", and fingerprint is irreplaceable, so that the security of identification by fingerprint is greatly improved, and with the development of image processing pattern recognition method and fingerprint sensor technology is maturing, Fingerprint identification method in finance, public security, access control, household registration management and other fields have a good application prospects. Fingerprint acquisition is relatively easy, and the fingerprint recognition algorithm is more mature. Because fingerprint recognition has the advantages of fast scanning fingerprint, convenient and small size, fingerprint identification technology has gradually entered the civil market and applied to many embedded devices, but how to improve the recognition rate and stability of fingerprint recognition system, There are a number of technical challenges in reducing costs and extending stability and node distribution.

Therefore, this paper studies the core of microprocessor at91sam7x256 with ARM core, external extended fingerprint sensor MBF200 to form fingerprint identification server hardware, system software porting real-time multitasking operating system Μc/os-ⅱ, file system, LwIP, application software to realize fingerprint recognition. The method has the advantages of low cost, less resource and strong expansibility.

1 principle and hardware design of distributed fingerprint recognition system

Fingerprint identification technology mainly involves 4 functional modules: Reading fingerprint images, extracting features, preserving data and matching. Through the fingerprint reading device to read the human fingerprint image, and then the original image of the initial processing to make it clearer, and then through the fingerprint identification software to establish fingerprint characteristics data. The software finds data points called "nodes" (minutiae) from fingerprints, which are the bifurcation, termination, or coordinate positions of the fingerprint lines, which have more than 7 unique characteristics. Typically there are 70 nodes on the finger, so this method produces approximately 500 data. This data, often called a template. The method of computer fuzzy comparison. Compare the two fingerprints of the template, calculate their similarity, and finally get two fingerprint matching results.

The hardware circuit is implemented with microprocessor at91sam7x256 as the core, the peripheral circuit mainly includes fingerprint identification Module MBF200, Ethernet physical Layer (PHY) transceiver RTL8201BL, large-capacity data flashat45dbl61d, hardware calendar clock device DSl302, Power supply circuit, reset and clock circuit, 1 shown.

　　

1. 1 at91sam7x256 device and MBF200 module application

AT91SAM7X256 is a 32-bit ARM7TDMI-based microprocessor introduced by Atmel Corporation. It also integrates the on-chip flash and the KB SRAM on a piece of chips without the need for external expansion memory. The interior is also integrated with USB2. 0 device ports, as well as rich on-chip peripheral resources, powerful. The at9lsam7x256 reset controller can manage the power-up sequence of the chip and the entire system. The microcontroller has an embedded 10/100 MB/s Ethernet (Ethernet) MAC, CAN, full speed (MB/s) USB2. 0, designed for a wide range of networked real-time embedded systems, its performance is stable, powerful, can be widely used in the field of protocol conversion, communication, industrial control. Using at91sam7x256 to develop fingerprint identification system can effectively control the cost. Industrial networks require extreme stability, but experiments have shown that over 60% of bus bandwidth usage can cause conflicts.

MBF200 is an advanced solid-state fingerprint sensor developed by Fujitsu, which, in addition to the automatic detection of fingerprints, also comes with a variety of interface modes for capacitive sensors, the capacitive sensor array consists of two-dimensional metal electrodes, all metal electrodes act as a capacitive plate, the contact finger acts as a 2nd capacitive plate, The passivation layer on the device surface acts as the insulating layer of the two plates. When the finger touches the surface of the sensor, the roughness of the fingerprint creates a change of capacitance on the sensor array, causing the voltage to change on the two-dimensional array and to form a fingerprint-sensing image. Capacitive solid-state devices with standard C13MS technology with a resolution of up to 1 dpi, and a sensor area of. Cmxl. . With automatic fingerprint detection capability, including 8-bit analog-to-digital converter, can provide 3 kinds of bus interface form. The power consumption under 5 V operating voltage is less than MW.

1. 2 Ethernet Interface Circuit design

The at91sam7x256 is internally integrated with a MAC controller to support the Mii interface and the Rmii interface. The RTL820LBL is an industrial-grade 10/100 MB/s Low Power Ethernet transceiver with a MII interface, a MHz clock output, an intelligent power-down mode that provides a reliable, high-quality network solution for the system, and an Ethernet that supports real-time transmission for factory enterprises and other harsh operating environments. IEEE compliant

802. Technical standards for 3u. The Ethernet interface circuit is shown in schematic diagram 2.

　　

2 software design of distributed fingerprint identification system

2. 1μc/os-ⅱ System porting

Because the system hardware platform chooses the embedded microprocessor at91sam7x256 the RAM, the flash and so on the resources are very limited, considers various factors, chooses the Μc/os-ⅱ as the embedded operating system, the TCP/IP protocol expands realizes. The Μc/0s-ⅱ operating system is a real-time multi-tasking operating system that is open source, portable, curable, cropped and head-up. Its most important feature is the source code is open, most of the source code is written in ANSI C. Although Μc/0s-ⅱ provides the main services such as time management, task synchronization, task management and memory management, it is highly scalable. The scalable top-level services are: device drivers, file systems, graphics systems and TCP/IP protocol systems, and because of their performance can be comparable to many high-end commercial software products, and even some performance than they are better, so it is a huge advantage to attract many developers. Μc/0s-ⅱ is a preemptive real-time multitasking operating system kernel designed for microcontroller systems and software development, which is the background program that is executed first after the start of the microcontroller, as the framework of the whole system runs through the system, for the real-time and stability of high-demand data acquisition system, The introduction of Μc/os-ⅱ will undoubtedly greatly improve its performance.

2. 2 Can bus interface communication module

The can protocol is based on an open system interconnection model of ISO, taking 3 layers: Physical layer, Data link layer and application layer. The functions of the physical layer and the data link layer can be implemented by the can interface device, while the application layer's function is done by the application. The function of the CAN bus interface communication module is to receive and transmit can bus data. The main operations include the initialization of can controllers and the operation of receiving and transmitting data on the CAN bus. The information can be received by the can controller, and is automatically completed by the can controller.

2. 3 Ethernet Communication Module implementation

Ethernet is the most common communication protocol standard used in communication network, which defines the type of cable and signal processing method used in communication network. Ethernet uses the carrier frame listening multi-access (CSMA/CD) mechanism with collision detection, which is a broadcast network. Data must be sent and received on an Ethernet IEEE802. 3 protocol to proceed. The implementation of the software is mainly the initialization of the module, data transmission and reception of 3 parts. The on-chip ported Clinux system contains the TCP-IP protocol stack, and the Ethernet controller is integrated within the at91sam7x256. Therefore, as long as the gateway system is turned on, the initialization of the module is completed.

2. 4 Fingerprint identification module software implementation

　　fingerprint recognition algorithm is the core of fingerprint recognition system. The fingerprint recognition algorithm adopted in this system is: extracting ridge line direction, ridge Line frequency, Gabor filtering, feature extraction, and fingerprint database for feature matching. The system has two functions: fingerprint identification and fingerprint template storage. The fingerprint Identification software module is shown in Workflow 3.

　　

After acquiring the fingerprint information, image enhancement is needed, in which the core problem is the preprocessing of fingerprint image, which is aimed at reducing the noise and improving the image quality, so that the feature can be extracted. The fingerprint texture consists of an alternating ridge line and a valley line, which contain a lot of information such as texture direction, texture density and so on. This information displays different features in different regions. The fingerprint image enhancement algorithm is realized by using the regional difference of image information.

This system reference the fingerprint image texture frequency information, with Gabor transform this can simultaneously to the direction of the local structure of the image and spatial frequency analysis of the optimal filter as a template, thus greatly improving the effect of the enhancement algorithm. The method for extracting ridge lines is:

1) The fingerprint image is segmented into a small enough sub-block to satisfy the condition of the texture approximation parallel in the block;

2) for each sub-block of each point P (s,t) (s,t=o,i...w-1) using Sobel operator to calculate its X-direction gradient gx and y-direction gradient gy respectively;

3) Calculation formula for each sub-block direction θ (m,n):

In the formula,..

Gabor transform has the characteristics of optimal time-domain and frequency-domain connection resolution, and can analyze the orientation and spatial frequency of the local structure of the image at the same time, so the ridge direction of the fingerprint image and the ridge line frequency information are well taken into account. As the direction of the filter in the direction perpendicular to the direction of the sub-block lines, the ridge frequency is used as the filter frequency to construct the filter. In this system, the fingerprint matching is based on the feature point set matching calibration algorithm, the algorithm is simple comparison logic and subtraction operation, no need to use DSP processing unit. The use of the ARM7 device in this design can work better. As the fingerprint recognition has a certain degree of rejection, so if the identification results to reject this person, the 3 consecutive times are refused to be established.

3 concluding remarks

This distributed fingerprint recognition system can be applied to the monitoring and access management of large enterprises based on ARM7 core multi-node. The system can realize the field bus and Ethernet interconnection communication problem of fingerprint data acquisition, so that each node can connect to the backbone Ethernet network, facilitate the management and update of fingerprint database, and can carry out the feature matching query remotely. The innovation of this design is: 1) Adopt high integration high-performance ARM7 processor at91sam7x256 solution, take full advantage of at91sam7x256 provides full-duplex Ethemet controller, can controller, simple structure, low cost, with commercial value. ; 2) Design the specific system Ethernet, can bus and power supply hardware circuit diagram, 3) transplant Μc/0s-ⅱ, give the specific fingerprint identification algorithm and processing scheme, achieve stability and low-cost combination.

Design and implementation of fingerprint recognition system based on ARM9