CN101226588A - A fingerprint recognition method and device based on a field programmable gate array chip - Google Patents

Abstract

The invention discloses a fingerprint identification method and device based on a field programmable gate array chip, which is realized by a hardware description language, and sequentially performs image distribution, fingerprint image enhancement, and minutiae extraction operations on fingerprint image data read by a fingerprint sensor interface. Store the fingerprint template containing fingerprint minutiae information in the external memory chip through the external memory interface; in minutiae matching, perform minutiae matching on the minutiae template generated by the input fingerprint and the stored minutiae template to confirm whether they belong to the same finger. The field programmable gate array chip is respectively connected with the fingerprint sensor, the fingerprint template storage circuit, the serial communication circuit and the power management circuit. The invention improves the fingerprint identification speed in the embedded system environment, reduces the temporary data storage space required in the fingerprint identification process, and reduces the hardware cost of the fingerprint identification system. Higher fingerprint recognition performance can be achieved on lower cost embedded systems.

Description

A fingerprint recognition method and device based on a field programmable gate array chip

technical field

The invention relates to a fingerprint identification method, in particular to an FPGA-based fingerprint identification method and device.

Background technique

Biometric identification technology is an emerging technology gradually formed with the continuous development of computer science and technology, especially the development of computer image processing and pattern recognition. In recent years, biometric identification technology, especially fingerprint identification technology has become more and more mature, and the core technology has fully reached the practical level. Compared with other biometric identification methods, such as palmprint recognition, iris feature recognition and other biometric identification methods, fingerprint recognition is more important in terms of uniqueness of features, persistence of features, reliability of feature recognition, and user acceptability. has considerable advantages. As countries around the world pay more and more attention to various security issues, more and more fingerprint identification products have been put on the market.

At present, the existing fingerprint identification products are basically divided into two categories from the fingerprint identification processing platform: one is based on the computer as the computing processing platform; the other is based on the embedded system as the processing platform. Both products have advantages and disadvantages. Because the first type of product uses a computer as a processing platform, it is stronger than the second type of product in terms of data processing speed and fingerprint recognition performance. The second type of products is superior to the first type of products in terms of product cost performance and product applicability because the embedded system is used as the processing platform. Moreover, the number of products in the second category is far more than that in the first category.

Because regardless of high-performance computer processors or low-performance embedded processors, they all belong to general-purpose processors. Therefore, the above-mentioned two types of products all have a common feature in the realization of the fingerprint identification method, that is, all products use a general-purpose processor as a computing platform for the fingerprint identification method. Its characteristic is that it must rely on pre-compiled programs to complete predetermined operations. In different fingerprint identification products, fingerprint identification methods are implemented using various programming languages, such as C language and various assembly languages. It is compiled into different machine codes through a dedicated program compilation environment and stored in the program storage chip of the product. The corresponding processing program is read from the program storage chip through the control of the general processor to complete the operation of fingerprint identification. The implementation of this fingerprint identification method that relies on software programming completely depends on the performance of the general-purpose processor, and a general-purpose processor with lower performance will affect the performance of fingerprint identification products. The implementation of this fingerprint identification method greatly limits the performance of fingerprint identification products using low-cost embedded processors, thereby affecting the application range of fingerprint identification products.

Contents of the invention

The implementation of the existing fingerprint identification method completely depends on the performance of the general-purpose processor, which limits the performance of the fingerprint identification product and limits the application of the fingerprint identification product. The purpose of the present invention is to adopt a technology that does not rely on programming languages and general-purpose processors. The scheme provides a method and device for realizing fingerprint recognition on a field programmable gate array chip based on a hardware description language.

In order to achieve the stated purpose, a first aspect of the present invention provides a fingerprint identification method based on a field programmable gate array chip, the steps comprising:

Fingerprint registration steps: implemented by the hardware description language, image distribution, fingerprint image enhancement, and minutiae extraction operations are performed sequentially on the fingerprint image data read by the fingerprint sensor interface functional unit, and finally the fingerprint minutiae information is included in the external memory interface functional unit The fingerprint template is stored in the external memory chip;

Fingerprint matching step: implemented by the hardware description language, it is to perform minutiae matching on the minutiae template generated by the input fingerprint and the stored minutiae template in the minutiae matching function unit, and identify whether the fingerprint belongs to the same finger.

Preferably, the fingerprint registration step further includes: using image overlapping block technology, extracting image blocks from the input fingerprint image in units of windows in the image distribution function unit, and extracting image blocks in the fingerprint enhancement function unit and minutiae point extraction function unit The image block is used as the object to perform fingerprint image enhancement and local minutiae point extraction operations. In the minutiae point extraction functional unit, the local minutiae point information extracted from each image block is reconstructed to complete the input fingerprint minutiae point template extraction operation.

Preferably, said field programmable gate array chip-based fingerprint image enhancement also includes the following processing steps:

Step 1: Using multi-directional gradient estimation, calculate the direction of the image pixel in the input image block;

Step 2: Using image filtering based on geometric characteristics, the input image block is filtered using the set filtering templates in different directions;

Step 3: Thinning the filtered input image block by using the morphological template.

Preferably, the minutiae extraction adopts a minutiae reconstruction method, and the local minutiae information extracted from each image block is reconstructed into a complete fingerprint template according to the image block division method in the image allocation operation.

Preferably, the field-programmable gate array chip-based identification fingerprint operation has the following characteristics:

S1: Build a temporary data storage area inside the field programmable gate array chip, and store image blocks to be processed, local detail point information, and temporary data of fingerprint templates to be matched during the fingerprint recognition operation process;

S2: The pipeline operation mode is adopted in the process of fingerprint recognition operation, and the same fingerprint image block is sequentially processed in the fingerprint enhancement function unit and the minutiae point extraction function unit to complete a local minutiae point information extraction operation.

Preferably, the minutiae point matching adopts a point matching method, and the similarity between two minutiae point templates is calculated as:

In order to achieve the stated purpose, the second aspect of the present invention provides a fingerprint identification device based on a field programmable gate array chip, including:

The fingerprint sensor is connected with the field programmable gate array chip, and is used to convert the fingerprint information into an image data stream and transmit it to the field programmable gate array chip;

The power management circuit is connected with the field programmable gate array chip to provide the power required by the fingerprint identification device;

The fingerprint template storage circuit is connected with the field programmable gate array chip, and is used for storing the fingerprint template data generated by the field programmable gate array chip;

The field programmable gate array chip is respectively connected with the fingerprint sensor, the power management circuit, the fingerprint template storage circuit, and the serial communication circuit, and is used to receive the information of the above-mentioned components to complete the fingerprint recognition operation;

The serial communication circuit is connected with the field programmable gate array chip, and is used for communicating with the computer, receiving control commands from the computer and sending fingerprint recognition results to the computer.

Preferably, the field programmable gate array chip includes:

The fingerprint sensor interface functional unit is used to read fingerprint image data;

The image distribution functional unit is respectively connected with the fingerprint sensor interface functional unit and the temporary data buffer functional unit, and generates fingerprint image blocks by reading the fingerprint image data;

The fingerprint image enhancement functional unit is connected with the temporary data buffer functional unit, and performs fingerprint refinement operation on the fingerprint image block data in the temporary data buffer to generate fingerprint refinement information;

The external memory interface functional unit is connected with the temporary data buffer functional unit, and is used to read the stored minutiae template into the temporary data buffer, or write the user fingerprint template generated by fingerprint registration into the external fingerprint storage chip;

The temporary data buffer functional unit is used to store the temporary data generated by each functional unit of the image distribution functional unit, the fingerprint image enhancement functional unit, the minutiae extraction functional unit, and the minutiae matching functional unit;

The minutiae point extraction functional unit is connected with the temporary data buffer function unit, processes the fingerprint refinement information input by the temporary data buffer area, and obtains the fingerprint minutiae information;

The minutiae point matching functional unit is connected with the temporary data buffer function unit, and is used for performing minutiae point template generated by the fingerprint image read in by the image distribution function unit and the stored minutiae point template read in by the external memory interface function unit point match.

Preferably, the fingerprint image enhancement functional unit includes:

The fingerprint image direction field module is connected with the temporary data buffer, and is used to extract the direction of the specified pixel in the fingerprint image block;

The fingerprint image filtering module is connected with the fingerprint image direction field module, and uses the filtering template to perform filtering operation on the fingerprint image block;

The fingerprint image thinning module is connected with the fingerprint image filtering module, performs thinning operation on the fingerprint image block, obtains the fingerprint thinning image and inputs it to the temporary data buffer.

Beneficial effects of the present invention: the fingerprint identification method of the present invention is based on the characteristics of the Field Programmable Gate Array chip (FPGA), and uses a hardware description language to divide the fingerprint sensor interface, image distribution, image enhancement, and detail points in the FPGA chip Functional units such as extraction, temporary data buffer, minutiae point matching, and external memory interface realize the functions of fingerprint registration and fingerprint matching. At the same time, in the process of fingerprint recognition operation, pipeline operation mode is adopted to sequentially perform fingerprint enhancement and fingerprint recognition on the same fingerprint image block. Minutia point extraction and other operations, and gradually complete the reconstruction of fingerprint minutiae points, greatly reducing the temporary data storage space required to process a complete fingerprint image.

The technical scheme of the invention solves the problem that the implementation of the existing fingerprint identification method completely depends on the performance of the general-purpose processor, which limits the performance of fingerprint identification products and limits the application of fingerprint identification products. The fingerprint identification method of the invention improves the fingerprint identification performance in the embedded system environment, reduces the temporary data storage space required in the fingerprint identification process, and reduces the hardware cost of the fingerprint identification system.

The technical scheme of filtering the input image block by adopting the filtering template of the present invention reduces the time used for filtering processing and improves the fingerprint recognition speed.

Description of drawings

Fig. 1 is a block diagram of a fingerprint identification device realized using the method of the present invention;

Fig. 2 is a functional block diagram of fingerprint identification according to the present invention;

Fig. 3 is a schematic diagram of image overlapping block technology;

Fig. 4 is the image block output by the fingerprint image distribution functional unit;

Fig. 5 is the processing procedure of fingerprint image enhancement functional unit;

Fig. 6 is the fingerprint image enhancement processing result;

Fig. 7 is the fingerprint minutiae point information structure;

Fig. 8 is a schematic diagram of fingerprint minutiae point reconstruction;

Fig. 9 is a test performance curve diagram of the fingerprint recognition method of the present invention;

Detailed ways

The present invention will be described in detail below in conjunction with the accompanying drawings. It should be noted that the described embodiments are only intended to facilitate the understanding of the present invention, rather than limiting it in any way.

The fingerprint identification method workflow of the present invention is divided into following two stages:

1. Fingerprint registration stage:

The fingerprint identification method realized by the hardware description language performs operations such as image distribution, image enhancement, and minutiae point extraction on the fingerprint image input by the fingerprint sensor interface function unit 51 in the fingerprint registration stage, and finally through the external memory interface function unit 54. The fingerprint template of the minutiae information is stored in the fingerprint template storage circuit 3 .

In the fingerprint registration stage, overlapping image block technology and pipeline operation mode are adopted. In the image distribution function unit 52, the input fingerprint image is extracted in units of windows, and in the fingerprint enhancement and minutiae point extraction function unit 56, image The operation is performed in units of blocks, and the data of the same image block is processed sequentially to complete a detail point extraction operation. In the minutiae extraction function unit, the local minutiae information extracted from each image block is reconstructed to complete the minutiae template extraction operation of the input fingerprint. This processing method greatly reduces the temporary data storage space required to process the complete fingerprint image.

It contains the following steps in order:

(1) Perform block processing on the fingerprint image

The fingerprint identification method of the present invention is aimed at the characteristics of the FPGA computing platform and considering the cost factor, and does not use an additional system memory to store temporary data. A 20K byte internal temporary data storage area is built by the logic unit inside the FPGA chip 4 . In the process of fingerprint recognition operation, image blocks to be processed, local detail point information, and temporary data of fingerprint templates to be matched are stored; FPGA as a processing chip does not exchange data with external data memory. For the fingerprint image data input by the sensor interface function unit 51, the block operation of the fingerprint image is completed in the image distribution function unit. The image distribution function unit 52 divides the fingerprint image into image data blocks of size w×w, where w=64 pixels. As shown in FIG. 3 , the sequence of image block data acquisition during fingerprint image processing is from left to right and from top to bottom.

In order to avoid block effects in the subsequent calculation process, the image allocation function unit 52 divides the fingerprint image block area by overlapping windows. The image block window moves w/2 pixel distance each time to ensure that the data of two adjacent fingerprint image blocks overlap with w/2 pixel width data. Figure 4 shows fingerprint image patches acquired sequentially using overlapping windows.

(2) Enhance the fingerprint image

Limited by the cost of the FPGA chip 4, it is impossible to save the processing results of all fingerprint image blocks in the temporary data storage area combined and generated inside the chip. Utilize the method described in the present invention, the processing object of fingerprint image enhancement function unit 53 is fingerprint image block, after finishing fingerprint thinning operation, output the result of fingerprint image block thinning to minutiae extraction function unit 56, delete all simultaneously Temporary data in preparation for processing the next fingerprint image block. The processing procedure of fingerprint image enhancement function unit 53 is as shown in Figure 5, comprises the following steps:

(2.1) Find the direction field of the fingerprint image

The direction of the pixel point in the image block is obtained by using the method of multi-directional gradient estimation. By calculating the 4 basic directions within the neighborhood of 8 pixels:

$\left\{\begin{array}{c}\mathrm{<span\; class="notranslate">\; <figure-callout\; id="1"\; label="o"\; filenames="A20071006282700201.png"\; state="\{\{state\}\}">o</figure-callout></span>}\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; I</span>}\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; +</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; I</span>}\mathrm{<span\; class="notranslate">\; 8</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; |</span>\\ \mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; I</span>}\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; +</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; I</span>}\mathrm{<span\; class="notranslate">\; 6</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; |</span>\\ \mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; 3</span>}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; I</span>}\mathrm{<span\; class="notranslate">\; 4</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; +</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; I</span>}\mathrm{<span\; class="notranslate">\; 5</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; |</span>\\ \mathrm{<span\; class="notranslate">\; o</span>}\mathrm{<span\; class="notranslate">\; 4</span>}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; I</span>}\mathrm{<span\; class="notranslate">\; 2</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; |</span><span\; class="notranslate">\; +</span><span\; class="notranslate">\; |</span>\mathrm{<span\; class="notranslate">\; I</span>}\mathrm{<span\; class="notranslate">\; 7</span>}<span\; class="notranslate">\; -</span>\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; |</span>\end{array}\right.$

Obtain the basic direction factors of the pixel points, o1, o2, o3, o4 correspond to the direction factors of the four basic directions of 0°, 45°, 90°, and 135° respectively. Calculate the main direction of the pixel point by the following formula:

o(x,y)=α1(x,y)×o1+α2(x,y)×o2+α3(x,y)×o3+α4(x,y)×o4

o(x, y) represents the main direction of the pixel, and [α1, α2, α3, α4] is the weight parameter of the basic direction factor [o1, o2, o3, o4], which are equal to [1, 32, 96 respectively , 64].

(2.2) Direction filtering

The directional filtering method based on geometric characteristics is adopted, and the filter equation of the following form is used:

分别表示局部指纹脊线频率和方向， $f=\frac{1}{6};$ δ_x、δ_y分别表示高斯包标准偏差，都等于4。根据图像块的范围，求出针对不同方向的滤波模板，选择滤波模板的大小为k×k，其中k＝16。滤波卷积方程如下：Among them, x, y represent the pixel position; f,

represent the local fingerprint ridge frequency and direction, respectively, $f=\frac{1}{6};$ δ _x , δ _y represent the standard deviation of the Gaussian package, both equal to 4. According to the range of the image block, filter templates for different directions are obtained, and the size of the filter template is selected as k×k, where k=16. The filter convolution equation is as follows:

$\mathrm{<span\; class="notranslate">\; f</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\underset{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&Subset;</span>\mathrm{<span\; class="notranslate">\; D.</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; \&CenterDot;</span>\mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span>$

where D represents the image block area. Using the template to perform the filtering operation of the fingerprint image block can speed up the execution speed of the fingerprint filtering process.

(2.3) Fingerprint image refinement

The enhanced fingerprint image block is thinned by morphological method, and the fingerprint image block is corroded by four morphological templates. The four templates represent four different processing directions, namely 0°, 90°, 180°, and 270°. The processing template T1 in the 0° direction is as follows, and the processing templates T2, T3, and T4 in other directions can be obtained by rotating the T1 template:

I1==0 I2==0 I3==0 I4==0 NA I5==0 16＞K 17＞K 18＞K

Among them, NA represents the central pixel to be processed, and k represents the operating threshold. When the 8 neighboring pixels of the central pixel meet the template condition, set it to 0, indicating that it is not a valid ridge. Use the template to perform repeated point-by-point operations in the image block area until the image block thinning operation is completed.

The final result of fingerprint image enhancement processing is shown in Figure 6.

(3) Extract detail points

Fingerprint minutiae information includes minutiae coordinate information, local ridge direction, and minutiae type. Due to the use of overlapping window technology to process fingerprint images, the minutiae information can only be accurately extracted within the image block. At the same time, it is limited by the temporary data storage area, so the present invention does not use global minutiae features, such as the number of crossed ridges and other information. The complete minutiae feature vector is {x, y, θ, t}, as shown in Figure 7, where x, y represent the position of the minutiae, θ represents the direction of the minutiae, and t represents the type of the minutiae.

After the extraction of local fingerprint minutiae in the image block is completed, since the processing of this fingerprint image block is completed, the temporary storage area occupied by this operation needs to be prepared for the processing of the next image block. Therefore, the local fingerprint minutiae points extracted by each image block are saved and reconstructed to ensure the integrity of the minutiae set.

See Figure 8 for a schematic diagram of the minutiae point reconstruction process. Increasing the number of minutiae points extracted can improve the fingerprint recognition performance of the method of the present invention, but processing too many minutiae points during the reconstruction of minutiae points will occupy too much temporary data storage area. Therefore, under the premise of ensuring the performance of fingerprint recognition, the number of minutiae points contained in the complete minutiae point set is limited within 100.

2. Fingerprint matching stage:

In the fingerprint matching stage, the fingerprint identification method implemented by the hardware description language first obtains the minutiae template from the fingerprint image read by the fingerprint sensor interface function unit 51 through the above steps (1) to (3). In the minutiae matching function unit 57, the minutiae template generated by the input fingerprint and the stored minutiae template read in by the external memory interface function unit 54 are subjected to minutiae matching to confirm whether they belong to the same finger.

The matching of fingerprint minutiae set mainly utilizes the characteristics of minutiae structure, which is point matching. This structure is based on the local structure of minutiae points, which is composed of the minutiae feature vectors obtained in the above step (3). The matching similarity between the minutiae template generated by the input fingerprint and the stored minutiae template is calculated by the following formula:

Among them, P _im ^and P _j ^l are the i-th point in the input minutiae template and the j-th point in the stored minutiae template; PWR is the coefficient of the normalized weight value, and the value is 100; THR is the experience threshold, which is taken as The value is 50; k represents the number of matching minutiae points between two fingerprint templates.

The above minutiae structure is matched one by one on the input minutiae template and the stored minutiae template, and at this time, the accumulated matching points N and the matching similarity sum PS of a record are obtained. The final matching score is determined by the following formula:

Score=PS/N

By comparing the Score with the preset fingerprint matching threshold, it is finally determined whether the input fingerprint is consistent with the stored fingerprint template. According to experience, the fingerprint matching threshold is set to 0.4. When the Score is greater than this threshold, it is considered to be the fingerprint of the same finger, otherwise it is not considered to be the fingerprint of the same finger. By adjusting the fingerprint matching threshold, the recognition performance of the system using the above-mentioned fingerprint recognition method, such as rejection rate and false recognition rate, can be adjusted, so that the fingerprint recognition system can adapt to different application fields.

The method of the present invention uses c language to carry out the fingerprint identification performance test on the computer, using the fingerprint image collected by the capacitive fingerprint sensor, the image size is 256 * 256 pixels, and the fingerprint database size is 100 people * 8 fingers. After testing, the fingerprint identification performance EER=2.55%, and the test performance curve is shown in FIG. 9 .

As shown in FIG. 1 , the device of the present invention includes: a fingerprint sensor 1 , a power management circuit 2 , a fingerprint template storage circuit 3 , an FPGA chip 4 , and a serial communication circuit 5 . The device receives commands from the PC through the serial port communication circuit 5, and performs fingerprint registration and fingerprint matching operations.

The fingerprint sensor 1 described in this device is a fingerprint collection chip based on CMOS technology, and the type of the chip can be planar or strip scanning. These sensors are characterized by their small size and are suitable for use in various embedded systems. The fingerprint sensor 1 enters the collected image data into the fingerprint sensor interface functional unit 51 of the FPGA chip 4 through an 8-bit parallel data line.

The power management circuit 2 described in this device is composed of a DC voltage conversion chip. This circuit converts the externally provided DC power into the DC voltage required by other circuits in the device to supply power to the entire device.

The fingerprint template storage circuit 3 in the device uses an electrically erasable flash memory (FLASH) to store registered user fingerprint templates.

The FPGA chip 4 reads and writes the FLASH chip through the external storage area interface function unit 54 . In the FLASH chip, the fingerprint template is stored in a data table structure. Each record unit in the table contains user ID number information and user fingerprint template information. Each record unit in the data table is stored in the order in which it is stored and written. The size of each record unit in the table is fixed. During the fingerprint registration operation, the external storage area interface function unit 54 writes the generated fingerprint template and the corresponding user ID number into the FLASH chip. During the fingerprint matching operation, according to the command of the FPGA chip 4, the external storage area interface function unit 54 sequentially reads each recording unit according to the storage order of the records, and performs the template matching operation.

The serial communication circuit 5 described in this device is composed of a serial communication interface chip. The circuit can transmit data between the computer and the FPGA chip 4 . The commands sent by the computer are transmitted to the FPGA chip 4 through the serial communication circuit 5, so that the fingerprint identification device performs corresponding fingerprint registration or fingerprint matching operations. The FPGA chip 4 sends the result of whether the fingerprint registration is successful or whether the fingerprint matching is successful to the computer through the circuit.

This device divides the fingerprint sensor interface functional unit 51, the image distribution functional unit 52, the fingerprint image enhancement functional unit 53, the external memory interface functional unit 54, the temporary data buffer 55, and the minutiae extraction function in the FPGA chip using a hardware description language. The unit 56, the minutiae point matching function unit 57, etc. realize the functions of fingerprint registration and fingerprint matching. The connection mode of each functional unit is shown in Figure 2:

Fingerprint sensor interface function unit 51, for reading in fingerprint image data;

The image distribution function unit 52 is connected with the fingerprint sensor interface function unit 51 and the temporary data buffer function unit 55 respectively, and will read in the fingerprint image data to generate the fingerprint image block;

The fingerprint image enhancement function unit 53 is connected with the temporary data buffer function unit 55, and the fingerprint image block data in the temporary data buffer is carried out the fingerprint thinning operation, and generates the fingerprint thinning information;

The external memory interface functional unit 54 is connected with the temporary data buffer functional unit 55, and is used for reading the stored minutiae point template into the temporary data buffer, or writing the user fingerprint template generated by fingerprint registration into the external fingerprint storage chip;

The temporary data buffer function unit 55 is used to store the temporary data generated by each functional unit of the image distribution function unit 52, the fingerprint image enhancement function unit 53, the minutiae extraction function unit 56, and the minutiae matching function unit 57;

The minutiae point extraction functional unit 56 is connected with the temporary data buffer function unit 55, processes the fingerprint refinement information input by the temporary data buffer area, and obtains the fingerprint minutiae point information;

The minutiae point matching function unit 57 is connected with the temporary data buffer function unit 55, and is used for the minutiae point template generated from the fingerprint image read in by the image distribution function unit and the stored minutiae point template read in by the external memory interface function unit Perform minutiae matching.

In the fingerprint image enhancement function unit 53, the direction field estimation module 531, the image filtering module 532, and the image thinning module 533 are divided to complete the image enhancement operation of the fingerprint image block, as shown in Figure 5, including:

The fingerprint image direction field module 531 is connected with the temporary data buffer 55, and is used to extract the direction of the specified pixel in the fingerprint image block;

The fingerprint image filtering module 532 is connected with the fingerprint image direction field module 531, and utilizes the filtering template to perform filtering operation on the fingerprint image block;

The fingerprint image thinning module 533 is connected with the fingerprint image filtering module 532 , performs thinning operation on the fingerprint image block, acquires the fingerprint thinning image and inputs it to the temporary data buffer 55 .

The operation mode of this device is divided into fingerprint registration and fingerprint matching. The external computer sends the operation command to the FPGA chip 4 through the serial communication circuit 5, and the FPGA chip 4 controls the device to perform corresponding operations.

During the fingerprint registration operation, the user's fingerprint information is collected by the fingerprint sensor 1 , and the fingerprint image data is sent to the FPGA chip 4 . Inside the FPGA chip 4, the fingerprint sensor interface functional unit 51 inputs the complete fingerprint image to the image distribution functional unit 52, and the image distribution functional unit 52 divides the complete fingerprint image into image blocks of a fixed size by using image overlapping block technology, As the operation object of subsequent processing such as image enhancement and minutiae extraction. In the subsequent processing process, the pipeline operation mode is adopted, and the fingerprint image enhancement function unit 53 and the minutiae point extraction function unit 56 sequentially process the same fingerprint image block to complete a minutiae point extraction operation.

In the process of fingerprint image enhancement, the fingerprint image block sequentially undergoes direction field extraction, image filtering, and image thinning operations to generate a temporary fingerprint thinning image. Then the local fingerprint minutiae information is generated in the minutiae extraction function unit 56 and stored in the temporary data buffer. After the minutiae extraction function unit 56 completes the processing of all image blocks of a fingerprint image, it reconstructs the local fingerprint minutiae information according to the order in which the image is divided by the image distribution unit 52 to obtain a fingerprint minutiae template. It is stored in the external fingerprint storage chip by the external memory interface function unit 54. So far, a fingerprint registration operation is completed, and the FPGA chip 4 sends registration success information to the computer through the serial communication circuit 5 at the same time.

During the fingerprint matching operation, the device first extracts the fingerprint template from the fingerprint image input by the user. Then by controlling the external memory interface function unit 54 to read a fingerprint template data from the external fingerprint storage chip according to the order of saving. The fingerprint matching is completed in the minutiae point matching function unit 57, and the similarity value of two fingerprint templates is obtained, and compared with the preset threshold. If the matching value is greater than the threshold, it indicates that the input user is a registered user, and the FPGA chip 4 sends the matching success information to the computer through the serial communication circuit 5 . If the matching value is less than the threshold, it indicates that the input user fingerprint template does not match the fingerprint template read this time. At this time, the external memory interface functional unit 54 will sequentially read the next fingerprint template data from the external fingerprint storage chip, and transmit it to the minutiae matching functional unit 57 for another fingerprint matching. If all the fingerprint templates in the external storage chip do not match the fingerprint templates of the input user, it indicates that the current user is not a registered user, and the FPGA chip 4 sends the matching failure information to the computer through the serial communication circuit 5.

This device uses Verilog hardware description language to test on the xc2v8000 chip of Xilinx Company, the FPGA external clock is 70MHz during the test, and the resource of the used FPGA chip 4 is about 34%. After testing, the average time required to register a fingerprint is 1 second, and the average time required to match two fingerprint templates is 1 millisecond.

The above is only a specific implementation mode in the present invention, but the scope of protection of the present invention is not limited thereto. Anyone familiar with the technology can understand the conceivable transformation or replacement within the technical scope disclosed in the present invention. All should be covered within the scope of the present invention, therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims (9)

1. fingerprint identification method based on field programmable gate array chip is characterized in that step comprises:

Fingerprint register step: realize by hardware description language, the fingerprint image data that is read in by fingerprint sensor interface function unit carried out successively image distributes, fingerprint image strengthens, minutiae point is extracted operation, finally the fingerprint template that will comprise the details in fingerprint dot information by the external memory interface functional unit is stored in the external memory chip;

The fingerprint matching step: being realized by hardware description language, is that minutiae point template that in minutiae point matching feature unit the input fingerprint is generated and the minutiae point template of having stored are carried out the minutiae point coupling, and whether the identification fingerprint belongs to same finger.

2. fingerprint identification method according to claim 1, it is characterized in that, described fingerprint register step also comprises: adopt image overlapping piecemeal technology, fingerprint image to input in image distribution function unit is that unit extracts image block with the window, in fingerprint enhancement function unit, minutiae point abstraction function unit be that object carries out that fingerprint image strengthens, local detail point extracts and operates with the image block, in minutiae point abstraction function unit, will be reconstructed, finish the operation of input fingerprint minutiae template extraction by the local detail dot information that each image block extracts.

3. fingerprint identification method according to claim 1 is characterized in that, described fingerprint image enhancing based on field programmable gate array chip also comprises following treatment step:

Step 1: adopt multi-direction gradient to estimate the direction of image slices vegetarian refreshments in the calculating input image piece;

Step 2: adopt image filtering, utilize the filtering template of the different directions that has been provided with that the input picture piece is carried out filtering based on geometrical property;

Step 3: adopt the morphology template, to carrying out Refinement operation through the input picture piece after the Filtering Processing.

4. fingerprint identification method according to claim 1, it is characterized in that, described minutiae point is extracted the minutiae point reconstruct mode that adopts, and will be reconstructed into complete fingerprint template according to image block dividing mode in the image batch operation by the local detail dot information that each image block extracts.

5. fingerprint identification method according to claim 1 is characterized in that, described identification fingerprint computing based on field programmable gate array chip has following feature:

S1: programmable gate array chip internal build ephemeral data memory block at the scene, in the fingerprint recognition calculating process, deposit pending image block, local detail dot information, fingerprint template ephemeral data to be matched;

S2: in the fingerprint recognition calculating process, adopt the pipeline operation pattern, in fingerprint enhancement function unit, minutiae point abstraction function unit, handle same fingerprint image piece successively, finish a local detail dot information and extract operation.

6. fingerprint identification method according to claim 1 is characterized in that, described minutiae point coupling adopts a some matching way, and the similarity of two minutiae point templates is calculated as:

7. the fingerprint identification device based on field programmable gate array chip is characterized in that, comprising:

Fingerprint sensor (1) is connected with field programmable gate array chip (4), is used for converting finger print information to image data stream and transfers to field programmable gate array chip (4);

Electric power management circuit (2) is connected with field programmable gate array chip (4), and the power supply that provides fingerprint identification device required is provided;

Fingerprint template memory circuit (3) is connected with field programmable gate array chip (4), is used for the fingerprint template data that storage is produced by field programmable gate array chip (4);

Field programmable gate array chip (4) is connected with fingerprint sensor (1), electric power management circuit (2), fingerprint template memory circuit (3), serial communication circuit (5) respectively, and the information that is used to receive above-mentioned these parts is finished the fingerprint recognition operation;

Serial communication circuit (5) is connected with field programmable gate array chip (4), is used for and the computing machine communication control command of receiving computer and send the fingerprint recognition result to computing machine.

8. fingerprint identification device according to claim 7 is characterized in that, comprising at described field programmable gate array chip (4):

Fingerprint sensor interface function unit (51) is used to read in fingerprint image data;

Image distribution function unit (52) is connected with ephemeral data buffer zone functional unit (55) with fingerprint sensor interface function unit (51) respectively, will read in fingerprint image data and generate the fingerprint image piece;

Fingerprint image enhancement function unit (53) is connected with ephemeral data buffer zone functional unit (55), and the fingerprint image blocks of data in the ephemeral data buffer zone is carried out the fingerprint thinning operation, generates fingerprint thinning information;

External memory interface functional unit (54) is connected with ephemeral data buffer zone functional unit (55), the minutiae point template that is used for having stored is read in the ephemeral data buffer zone, and perhaps the user fingerprints template that fingerprint register is generated writes in the outside fingerprint storage chip;

Ephemeral data buffer zone functional unit (55) is used for the ephemeral data that each functional unit of memory image distribution function unit (52), fingerprint image enhancement function unit (53), minutiae point abstraction function unit (56), minutiae point matching feature unit (57) produces;

Minutiae point abstraction function unit (56) is connected with ephemeral data buffer zone functional unit (55), and the fingerprint thinning information of ephemeral data buffer zone input is handled, and obtains the details in fingerprint dot information;

Minutiae point matching feature unit (57) is connected with ephemeral data buffer zone functional unit (55), is used for the minutiae point template that the fingerprint image that is read in by image distribution function unit is generated and carries out minutiae point by the minutiae point template of having stored that the external memory interface functional unit reads in mating.

9. described according to Claim 8 fingerprint identification device is characterized in that, fingerprint image enhancement function unit (53) comprising:

Fingerprint image orientation field module (531) is connected with ephemeral data buffer zone (55), the direction of the image block middle finger fixation vegetarian refreshments that is used to take the fingerprint;

Fingerprint image filtration module (532) is connected with fingerprint image orientation field module (531), utilizes the filtering template that the fingerprint image piece is carried out filtering operation;

Fingerprint image refinement module (533) is connected with fingerprint image filtration module (532), and the fingerprint image piece is carried out Refinement operation, obtains the fingerprint thinning image and inputs to ephemeral data buffer zone (55).

Images