CN102799869A - Embedded fingerprint identification system based on FPGA - Google Patents

Abstract

The invention discloses an FPGA-based embedded fingerprint identification system, which belongs to the technical field of computer programs, and in particular relates to an FPGA-based embedded fingerprint identification technology combining software and hardware. It includes a fingerprint identification algorithm IP core and a fingerprint acquisition instrument controller, and is characterized in that: the fingerprint identification algorithm IP core has been optimized in algorithm, and VerilogHDL is used for design optimization to form a fingerprint identification IP core with independent intellectual property rights. At the same time, the collector controller has the function of fingerprint quality evaluation, and the system introduces a fingerprint classification method to build a fingerprint recognition system framework that is easy to cut and combines high-efficiency software and hardware. Provides a high-performance FPGA-based embedded fingerprint recognition technology, which has high-quality fingerprint image collection, fast processing speed, low power consumption, strong scalability, and is convenient for secondary development. It can be widely used in fingerprint locks for mobile devices.

Description

An Embedded Fingerprint Recognition System Based on FPGA

technical field

The invention belongs to the technical field of computer programs, and in particular relates to an FPGA-based embedded fingerprint recognition technology combining software and hardware.

Background technique

At present, companies and research institutions engaged in the research of fingerprint identification algorithms are developing rapidly. Among them, companies such as Identicator and Secugen in the United States, and Segam in France, their strong economic strength and long-term technology accumulation make them in the world's leading position in this field; The fingerprint security system adopts the open source IP processor core Leon2; in addition, for example, Taiwan's Startek and South Korea's Pefis are also doing well. The current academic authority in the field of fingerprint recognition is A. K． Professor Jain is from the Pattern Recognition and Image Processing Laboratory of the Department of Computer Science, Michigan State University. His team has published dozens of papers on IEEE and made outstanding contributions to the development of fingerprint recognition. Now, the fingerprint identification technology has matured day by day. Although there are some shortcomings that need to be improved, it is an indisputable fact that it has been accepted by the market.

At present, fingerprint identification technology is widely used in products such as safes, mobile phones and laptops that require passwords to start. Philips and Motorola have launched business mobile phones such as W727 and ME860 respectively, Lenovo has launched a fingerprint recognition keyboard, and Yizhichan has launched a U disk computer fingerprint encryption lock; and Apple will also support fingerprint recognition on IOS5.1; The application of identification technology has been unknowingly integrated into people's life. Some supermarkets at home and abroad have begun to use fingerprint recognition systems to speed up billing operations and use them to open and close temporary lockers. Fingerprint recognition systems will be the primary identity verification tool.

In our country, the vast majority of companies that launch fingerprint identification products have introduced foreign technologies, and only companies that have truly mastered core technologies and owned independent intellectual property rights are Peking University High-Tech, Zhongzi Hanwang, Yuean Group, and Hangzhou Shengyuan. In terms of academics, the Institute of Automation of the Chinese Academy of Sciences represented by Tian Jie, the Department of Computer Science of Peking University, the Department of Automation of Tsinghua University, and Beijing University of Posts and Telecommunications are in the leading positions in China.

At present, the hardware platform of the embedded fingerprint recognition system is mainly based on the high-speed DSP processor (mostly the TMS320 series DSP application of TI Company), with the DSP chip as the core, it controls the fingerprint collector to collect fingerprints, and the RAM stores the code and data at runtime. Both operation and control are completed by DSP, mainly considering algorithm optimization and memory optimization. In recent years, ARM has been widely used in embedded control, consumer electronics, network communication, mobile applications and other fields. In China, ARM processors have been used as hardware platforms to develop fingerprint recognition systems. Some universities have done some research. The operation of the above two technical algorithms can only be executed sequentially, and the speed of the identification system is limited. However, FPGA (Field Programmable Gate Array) has the characteristics of parallel processing and fast operation speed. It can perform fingerprint identification module for different environments and different user groups. Crop, expand, update online. These features make it easy for developers to design, modify, and upgrade. This is also one of the purposes of this study, to further study the ASIC integrated circuit technology to make the IP core into a very large scale integrated circuit---fingerprint identification chip. In this way, only one chip can complete the work of the fingerprint identification system, which greatly shortens the development cycle, and at the same time expands the applicable scope and market application field of fingerprint identification.

Contents of the invention

The purpose of the development of the present invention is to overcome the above-mentioned defects in the prior art, and to provide a mobile device that can be used for fingerprint identification, and also as a prototype of application-specific integrated circuit development, and has low power consumption, stable performance, and fast identification speed. Effectively compare and record fingerprints.

A kind of embedded fingerprint identification system based on FPGA of the present invention comprises fingerprint identification algorithm IP core and fingerprint acquisition instrument controller, described fingerprint identification algorithm IP core has carried out algorithm optimization, adopts Verilog HDL to carry out design optimization, forms fingerprint identification IP core. At the same time, the collector controller has the function of fingerprint quality evaluation, and the system introduces a fingerprint classification method to build a fingerprint recognition system framework that is easy to cut and combines high-efficiency software and hardware.

The fingerprint identification algorithm IP core optimizes the fingerprint enhancement algorithm, including using cumulative gradients when calculating the direction field, optimizing the operator according to the symmetry of the Gabor function, and correlating the size of the filtering area with the frequency of the ridge. Its function is to simplify the calculation of the direction field and reduce the calculation of the Gabor coefficient, thereby reducing the calculation amount of the filtering algorithm as a whole, and the filtering area is more reasonable. The efficiency of the enhanced algorithm is improved as a whole.

The design optimization of the fingerprint identification algorithm IP core includes adopting pipeline design and serial-to-parallel conversion module to form the fingerprint identification IP core. Its role is to increase the speed of algorithm processing data by means of resource exchange rate, thereby increasing the speed of fingerprint recognition algorithm.

The fingerprint collector controller has the function of fingerprint quality evaluation, evaluates parameters such as over-humidity and over-dryness, fingerprint effective surface, fingerprint alignment, etc., and collects high-quality fingerprint images. Its role is to provide high-quality fingerprint input images for the fingerprint identification system and improve the performance of system identification.

In the fingerprint identification system architecture combining software and hardware, fingerprint classification is implemented by software, and fingerprint identification is implemented by hardware circuit IP core, which runs in parallel. Its role is to parallelize the architecture and improve the performance of the fingerprint identification system.

The beneficial effects of the present invention are:

1. The fingerprint identification algorithm of the present invention uses Verilog HDL to realize the IP core, which provides a model basis for the design of professional chip for fingerprint identification.

2. The optimization work of fingerprint identification IP core in the present invention also has the effect of reference in other IP core designs.

3. The optimization strategy of the fingerprint enhancement algorithm designed by the present invention has versatility in the algorithm optimization process, and the complex algorithm is optimized into an algorithm suitable for hardware circuit implementation, and the effect is obvious.

4. The software and hardware architecture constructed by the present invention, the software part and the hardware part are all implemented on the FPGA board. This architecture is advanced, easy to cut, and can be upgraded online, which fully utilizes the advantages of FPGA technology.

The technology we provide can be applied to fingerprint identification modules with poor battery life such as mobile devices. It can also quickly carry out the secondary development of the fingerprint recognition system, or carry out the design of a special fingerprint recognition chip.

Description of drawings

Fig. 1 is the system hardware frame of the present invention;

Fig. 2 is the flowchart of fingerprint pretreatment algorithm;

Fig. 3 is a system block diagram of the present invention;

Fig. 4 is a system work flowchart of the present invention;

Fig. 5 is a serial-to-parallel conversion module;

Figure 6 Fingerprint identification algorithm IP core top-level module;

Figure 7 is a comprehensive diagram of the normalization module;

Figure 8 is a comprehensive diagram of the enhancement module;

Figure 9 is a comprehensive diagram of the binarization module;

Figure 10 is a comprehensive diagram of the refinement module;

Fig. 11 Comprehensive diagram of feature extraction module;

Figure 12. Synthesis diagram of feature matching module.

In the figure: A is the collector controller; B is the fingerprint identification algorithm IP core.

Detailed ways

Example:

As shown in Fig. 1, 2, 6: a kind of software and hardware of the present invention combines the embedded fingerprint identification technology based on FPGA, comprise fingerprint identification algorithm IP core and fingerprint acquisition instrument controller, it is characterized in that: described fingerprint The identification algorithm IP core has been optimized by algorithm, and Verilog HDL is used for design optimization to form a fingerprint identification IP core with independent intellectual property rights, as shown in Figure 6-12. At the same time, the collector controller has the function of fingerprint quality evaluation, and the system introduces a fingerprint classification method to build a fingerprint recognition system framework that is easy to cut and combines high-efficiency software and hardware.

The fingerprint identification algorithm IP core optimizes the fingerprint enhancement algorithm, including using cumulative gradients when calculating the direction field, optimizing the operator according to the symmetry of the Gabor function, and the size of the filtering area is related to the frequency of the ridge line. When calculating the direction field, the gradient accumulation method is used instead of the traditional double angle and gradient vector square method, which greatly reduces the amount of calculation; the mathematical analysis of the Gabor function fully excavates the symmetry of the Gabor function, and it can be known that Gabor is Even symmetry, and the origin is symmetric in the xy coordinate system, so that only the pixel value of one quadrant can be calculated to obtain other pixel values, thereby reducing the amount of repeated calculations and shortening the filtering time. The filtering areas τ ₁ and τ ₂ represent the range of the rectangle, which is related to the ridge frequency f (8 is taken in this paper). τ ₁ =0.8f(i,j), τ ₂ =1.4f(i,j).

The design optimization of the fingerprint identification algorithm IP core includes the use of pipeline design and serial-to-parallel conversion modules, as shown in Figure 5, to form a fingerprint identification IP core with independent intellectual property rights, as shown in Figures 6-12. The function of the serial-to-parallel conversion module is to reduce the delay caused by reading data and improve the speed of algorithm processing data by means of resource exchange rate, thereby increasing the speed of fingerprint recognition algorithm. The IP core can be tailored according to the user's needs, leaving only the part that the user needs.

The fingerprint collector controller has the function of fingerprint quality evaluation, evaluates parameters such as over-humidity and over-dryness, fingerprint effective surface, fingerprint alignment, etc., and collects high-quality fingerprint images. Its role is to provide high-quality fingerprint input images for the fingerprint identification system and improve the performance of system identification.

则反应图像前景和背景的对比度，

值越大，对比度越明显。而对于湿度较大的指纹，采集图像灰度均值较小，方差也较小；对于太干的指纹，采集的图像灰度均值较大，方差较大。根据以上特性，设置灰度均值的阈值t1(下界)，t2(上界)，以及方差阈值v，进行联合决策。将指纹图像做三类情况处理：The mean M ₀ and variance of the image can be obtained. The average gray value M ₀ reflects the lightness and darkness of the image, the larger the M ₀ , the brighter the image; otherwise, the darker the image. variance

It reflects the contrast between the foreground and background of the image,

The larger the value, the more obvious the contrast. For fingerprints with high humidity, the average value of the gray scale of the collected image is small, and the variance is also small; for fingerprints that are too dry, the average value of the gray scale of the collected image is large, and the variance is large. According to the above characteristics, the thresholds t1 (lower bound), t2 (upper bound) and variance threshold v of the gray mean value are set for joint decision-making. The fingerprint image is processed in three types of situations:

指纹图像太亮，且对比度不高，即指纹太干，需湿润；1.

The fingerprint image is too bright and the contrast is not high, that is, the fingerprint is too dry and needs to be moistened;

指纹图像太暗，且对比度不高，即指纹太湿，需擦干；2.

The fingerprint image is too dark and the contrast is not high, that is, the fingerprint is too wet and needs to be wiped dry;

指纹明暗合适，且对比度高，即指纹质量好。3.

If the fingerprint is light and dark, and the contrast is high, the quality of the fingerprint is good.

The gray variance can be used to interpret the background and foreground of the fingerprint, so as to count the effective area of the fingerprint. By calculating the offset between the centroid of the effective area and the centroid of the original fingerprint image (width/2, length/2), the degree of offset of the user's finger is judged, so as to achieve the purpose of guiding the user to correct.

$\left\{\begin{array}{c}{\mathrm{<span\; class="notranslate">\; x</span>}}_{\mathrm{<span\; class="notranslate">\; bias</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; Fgr</span>}}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; /</span>{\mathrm{<span\; class="notranslate">\; N</span>}}_{\mathrm{<span\; class="notranslate">\; Fgr</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; width</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 2</span>}\\ {\mathrm{<span\; class="notranslate">\; Y</span>}}_{\mathrm{<span\; class="notranslate">\; bias</span>}}<span\; class="notranslate">\; =</span>\underset{\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; \&Element;</span>{\mathrm{<span\; class="notranslate">\; R</span>}}_{\mathrm{<span\; class="notranslate">\; Fgr</span>}}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; p</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; /</span>{\mathrm{<span\; class="notranslate">\; N</span>}}_{\mathrm{<span\; class="notranslate">\; Fgr</span>}}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; length</span>}<span\; class="notranslate">\; /</span>\mathrm{<span\; class="notranslate">\; 2</span>}\end{array}\right.$

Among them, x(p) and y(p) are the coordinates of point p; N _Fgr is the number of pixels in the foreground area. (width, length) are the width and height of the image respectively. Prompts are given to the user according to the relationship between the horizontal offset X _bias , the vertical offset Y _bias and the thresholds T _x and T _y respectively.

1. If |X _bias |>T _x and X _bis <0, it means that the fingerprint is too far to the left;

2. If |X _bias |＞T _x and X _bis ＞0, it means that the fingerprint is too much to the right;

3. If |Y _bias |>T _y and Y _bias <0, it means that the fingerprint is too high;

4. If |Y _bias |>T _y and Y _bias >0, it means that the fingerprint is too low.

In the above four cases, the user needs to adjust the position of the finger to collect the fingerprint again.

The architecture of the fingerprint recognition system combining software and hardware is shown in Figure 3. Fingerprint classification is realized by C software, which is transplanted into an embedded development environment. Fingerprint recognition is realized by hardware circuit IP core and runs in parallel. When the IP core preprocesses the fingerprint data, it starts fingerprint classification, and the classification result is stored in the designated location of Flash. The parallel architecture improves the performance of the fingerprint identification system. The engineering process of the system operation is shown in Figure 4.

Claims (5)

1. an embedded fingerprint identification system based on FPGA, comprising fingerprint identification algorithm IP core and fingerprint acquisition instrument controller, it is characterized in that: described fingerprint identification algorithm IP core has carried out algorithm optimization, adopts Verilog HDL to carry out design optimization, A fingerprint identification IP core is formed, and the collector controller has the function of fingerprint quality evaluation, and the system introduces a fingerprint classification method to build an easy-to-cut, efficient software and hardware combined fingerprint identification system framework. 2. the embedded fingerprint identification system based on FPGA according to claim 1, is characterized in that: described fingerprint identification algorithm IP core, has optimized fingerprint enhancement algorithm, comprises when calculating direction field and adopts accumulative gradient, according to The Gabor function symmetry optimizes the operator, and the size of the filtering area is related to the ridge frequency. 3. the embedded fingerprint identification system based on FPGA according to claim 1, is characterized in that: described fingerprint identification algorithm IP core, design optimization comprises adopting assembly line design, serial-to-parallel conversion module, and forms the fingerprint of independent intellectual property right Identify the IP core. 4. the embedded fingerprint recognition system based on FPGA according to claim 1, is characterized in that: described fingerprint acquisition instrument controller has the function of fingerprint quality evaluation, is too wet and dry, fingerprint effective surface, fingerprint alignment And other parameters are evaluated, and high-quality fingerprint images are collected. 5. The embedded fingerprint identification system based on FPGA according to claim 1, characterized in that: the fingerprint identification system framework of the combination of described software and hardware, fingerprint classification adopts software to realize, and fingerprint identification adopts hardware circuit IP core to realize, parallel run.

Images