A design scheme of optical fingerprint identification system

With the development of the application of electronic information technology, many occasions need to identify the specific user groups or identity records, such as access control system, attendance system, security certification system, in a variety of systems used in various forms of technology, such as retinal recognition, face recognition, fingerprint identification, RFID applications such as radio frequency identification. Among them, the biometric recognition method has been recognized and accepted by more and more people because of its convenience and high security, especially the fingerprint identification technology, which has been developed into one of the most widely used biometric recognition techniques. Therefore, the research of fingerprint recognition system based on embedded architecture has practical significance and broad application prospect.
1 system Overall structure
The system adopts optical fingerprint sensor (optical GC0307 CMOS Image Acquisition chip of Nechangko Microelectronics Co., Ltd.) and the 32-bit high performance MCU stm32f205re of arm Cortex M3 core STMicroelectronics, which uses Sobel edge detection operator, Gabor filter, image binary image acquisition and processing algorithm to identify the fingerprint image, built a small volume of embedded fingerprint identification module, with building block embedded, micro-power, program interface simple and easy to use, easy to two times development, high accuracy, high cost-effective and so on.
2 system hardware circuit design
The whole system is designed to form an integrated optical fingerprint identification module. The module design adopts the principle of optical dark background imaging, and adds a special living detection chip to solve the problem of residual fingerprint error identification and rubber false fingerprint while solving the dry finger effect.
Figure 1 shows the schematic diagram of the application circuit for the optical GC0307 CMOS Image Acquisition chip of Galaxycore Microelectronics Co., Ltd. The CMOS image acquisition chip is a high-precision, low-power, micro-volume high-performance camera built-in components, it enables the realization of high-quality VGA image CMOS image sensor with a highly integrated image processor, embedded power supply and high-quality lens group, output JPEG image or image video stream, support 8/10 Digital transmission of JPEG images and YCBCR interfaces, providing a complete imaging solution.



CMOS image capture core function output serial data PIN, clock signal pin, reset PIN, serial bus pin, etc. are connected to the Gpio port of the stm32f205re, through the GPIO port analog timing read the CMOS chip captured image information. Since the Stm32f205re Gpio Port works at a frequency of up to 10 MHz, the timing can be simulated very accurately and efficiently, and the original image of 640x480 can be captured in the main processor stm32f205re for image processing at a speed of up to S/sec at a rate of up to/from.
3 System software function design
The fingerprint image acquisition process of this system is shown in 2. In the part of system software design, four-point regularization algorithm is adopted for distortion correction.


The transformation from (x, y) to (u,v) can be obtained by the formula (1) and the formula (2), where A ~ H is determined by the optical path, which can be determined by the specific measurement data, and the original data can be obtained by actual measurement. Figure 3 shows the difference in effect between the original image and the distortion before and after the correction. Through the transformation, the distorted image can reach the resolution of up to a few dpi, which lays the basic conditions for the subsequent acquisition of high-quality image processing data.

It is then fed into the algorithm processing. Because the image processing algorithm of the embedded system must be small in computation and occupy a small amount of RAM memory, the system can run in the single-chip microcomputer system with limited computing performance, and therefore, it replaces the point direction by the small block direction and reduces the ram occupancy.
In the aspect of image enhancement, the image can be divided into small pieces with L as long, and then the mean variance of each block is obtained by the following formula:

According to the experimental data measurement and analysis, when the aver>36, can be considered as an image in the region, otherwise considered to be the background. The mean variance is used to distinguish the front and rear view, and the contrast of the image can be judged accordingly. According to the contrast difference to enhance the image, can make different exposure brightness of the image has been uniformly enhanced. The original image is processed by the algorithm, and the effect is compared before and after the extraction process, and the effect is shown in 4.

The original Sobel operator is as follows:

The improved Sobel operator is:

The improved Sobel operator can increase the accuracy of the direction field, and the measured pass rate is increased from 93.3% using the standard Sobel operator to 95.8%. Figure 5 shows the change.

5 It is shown that the improved Sobel operator, on the basis of the original Sobel operator, can significantly divide the area of the correct image, and can extract the correct direction almost in the whole picture area. The image is both Gabor filtered and image data binary. The fingerprint image belongs to the texture image, the texture image adopts Gabor filter, the point direction of each point is increased along the direction, and the direction of normal direction is weakened. The Gabor filter is a good way to stitch the broken lines, filter out the ambient noise, and finally make the two-window mean threshold two value for the Gabor filtered image:

Threshold 1: The mean-value operator matrix: the unit matrix of the 7x7.

Threshold 2: mean-value operator matrix: 3x3 Unit matrix.

The specific operation expression is as follows:

When the value of each point g (x, Y) >p (x, y) is assigned a value of g (x, y) = 1, otherwise the assignment is 0, in order to obtain a two value of the final result, the extraction of the image for the measured effect of the comparison of 6 is shown.

Figure 7 is finally based on the image texture of the thickness of the binary image, and based on the endpoint and intersection point extraction feature points.

Through the above steps, you can extract valid feature information from the original image. The characteristic information describes the location and direction of the feature points, and finally forms a feature template with a size of not more than 512 bytes. Fingerprint comparison is based on the characteristics of the template, the construction of two points formed by the pole pair set, and the rod length of the bar, the end of the direction of the rod and the angle of the information is already relative amount, independent of the position. Ideally, the same fingerprint, each of the two images that can be found, is mathematically equal to every amount (length, angle) of the rod pair. With this as the basic mathematical model, the entire alignment algorithm is constructed.

4 Conclusion

In this paper, the design of optical fingerprint identification system based on ARM , after physical testing, module input user fingerprint image time is ~ 1% MS, the average 4.2 ms can be compared to a fingerprint, support 1:1 fingerprint verification and 1:n fingerprint search. In the hardware design leads to the communication terminal, the system supports 3.3V TTL serial communication, the module can be registered through the serial port, delete the specific user, delete all users, reset module, get the total number of users, access to user rights, 1:1 ratio, 1:N, set the serial baud rate, read the image and extract the eigenvalues, Obtaining 30 General or extended function commands such as images, which can satisfy most fingerprint applications, can be well applied to the embedded field, thus confirming the feasibility of this scheme.

A design scheme of optical fingerprint identification system