OMAP-based embedded fingerprint recognition system design and application

Introduction

Biometric authentication technology is an important product in the information age.

Unique human biological characteristics and stability, therefore, the people developed a fingerprint, face, voice, Iris, genetic (DNA) architecture, and other biometric technology. In various biological characteristics, excellent fingerprint features, fingerprint identification technology is maturing. Therefore, for car security field of market needs, this article has designed a new OMAP embedded fingerprint recognition system. The system of recognition and high processing speed, scalability, and certainly has broad application prospects and research value.

OMAP open multimedia applications platform

TI's OMAP (Open Multimedia Applications Platform) platform is based on the DSP application open multimedia platform.

It uses dual-core structure, high performance and low power DSP core and control performance strong ARM microprocessor, with a high integration, hardware reliability and stability, speed, data processing capability, low power consumption, openness, good, etc. OMAP platform through advanced operating system platforms has not only open, and open the ARM DSP. Through DSP/BIOS bridge, DSP resources like ARM peripherals through the operating system API is invoked. DSP/BIOS bridge in the OMAP platform implements a duo of seamless connectivity. OMAP platform support WinCE, EPOC, Nucleus, VxWorks and Linux and other operating systems, provides easy-to-use open programming environment.

OMAP platform with dual core technology makes the operating system's efficiency and more multimedia execution of code optimization.

Real-time tasks completed by DSP, non-real-time tasks and system control work completed by the ARM, so that the system power consumption to a minimum, successfully resolved the performance and the best combination of power consumption.

System hardware design

OMAP-based embedded fingerprint identification system (for example in car application) mainly by OMAP5912 embedded microprocessor, FPS200 fingerprint sensor, man-machine interface, CAN bus interface, USB communication interface, power management, and other modules.

The system architecture as shown in Figure 1.

OMAP5912 is the core of the entire system.

It has to control fingerprint sensor FPS200 fingerprint sample acquisition, establishment of fingerprint sample libraries, and fingerprint library management and maintenance; on the other hand it also controls the DSP core processing complex of fingerprint identification algorithm for fingerprint image processing, fingerprint feature extraction and matching. In the system's operation, in particular the establishment of friendly man-machine interface. ADS7846 touchscreen controller and OMAP5912 under the control of, the user can touch-screen and LCD for man-machine conversation quickly and easily, such as user personal identification code (PIN) certification, fingerprint library management and maintenance. Open systems for information resource sharing, CAN bus interface makes the system and car multiple controller (i.e. electronic control unit ECU) for communication. USB communication interface design makes use of embedded system more flexible, convenient, and a variety of types of peripherals communicate possible. Powerful power management chip core for system TPS65010 provides the required 1.6V, 1.8V, 3V, 3.3V etc working voltage, and reasonably for power distribution. Because the fingerprint image data volume, fingerprint algorithm complexity, design requires a large storage space, so the system extends 1 NOR Flash and 1 DDR SDRAM. System accessible via JTAG interface for online debugging.

OMAP5912 embedded microprocessor

OMAP5912 microprocessor is 192MHz of DSP cores and TMS320C55x 192MHz low power consumption and enhanced ARM926 CPU dual core application processor consisting of.

Flow controller (Traffic Controller, or TC) is used to control access to external memory, the maximum operating frequency is 75MHz.

TC provides fast external memory interface (EMIFF), external memory slow interface (EMIFS) and internal memory interface (IMIF). Of these, you can connect with SDRAM EMIFF, and can only be used with Flash and EMIFS speed slower ROM connection. OMAP5912 there is internal memory, 192K by ARM and DSP sharing. TC can be thought of as storage resource guarding, by its internal arbitration to determine which kernel has the right to access these resources, and whether you can have two parallel access to these resources.

OMAP5912 Central has a high-performance 9 channel system direct memory access (DMA).

The 9 channel DMA allows different port transfers data between without ARM interference. DMA can jobs port including EMIFF, EMIFS, IMIF and peripheral components.

FPS200 fingerprint sensor

Veridicom Inc. FPS200 is based on standard CMOS technology fingerprint sensor.

Use semiconductor Silicon capacitive effect, Silicon sensors become a plate capacitor, the finger is another plate, the fingerprint ridge and Valley line relative to the smoothing of Silicon capacitive sensor, you can create an 8-bit grayscale image of the fingerprint. FPS200 fingerprint sensor operating voltage 5V at 3.3V-, low-power, high efficiency, smaller (1.28cm1.50cm) surface access and optical technology equally good or even better (500DPI of higher resolution) image quality.

In the hardware design, notably in OMA

P5912 and increase between 1 FPS200 74LV245 transceiver to resolve timing conflicts, will not repeat them here.

Man-machine interface design

Because of the touch screen convenience lightweight, input, this design uses four-wire resistive touch screen input device as man-machine interface.

HMI system by OMAP5912 microprocessor, touchscreen controller, 4-wire resistive touch-screen and LCD 4 parts. Touch screen controller adopts Burn-Brown ADS7846 chip company, it is a typical with continuous approximation of registers of 12-bit ADC, internal comes with 2.5V reference voltage, there is a standard and microprocessor connected SPI according to the interface. ADS7846 under serial data input DIN received different command to open the corresponding switch channel and receive the returned analog voltage, through A/d conversion came to the corresponding digital content, and then through the serial data output DOUT back to the microprocessor. Touch screen controller ADS7846 and microprocessor OMAP5912 connection circuit shown in Figure 2.

　　

CAN bus interface design

Embedded fingerprint recognition system complete owner identity authentication, you need to auto engine electronic control system sending (EEC), thereby igniting control signal to start the engine.

Therefore, this system CAN-bus interface was designed to enable communication with the EEC. CAN bus interface with microprocessor circuit usually include CAN controller and CAN transceiver, this design uses a Microchip's MCP2510 CAN Protocol controller and Philips PCA82C250 CAN transceiver. Connection relationship as shown in Figure 3.

USB communication interface design

In order for the embedded system with a variety of types of peripherals to communicate design extends USB communication interface.

OMAP5912 microprocessor built-in corresponds to a USB1.1 USB host controller, and the design support master-slave dual USB communication mode, only need to use a USB interface adapter can easily convert master-slave mode. Note that when you select a Client mode, you need to set FUNC_MUX_CTRL_D [5: 3] = 000.

Software design process

OMAP5912 software architecture

OMAP5912 software structure based on two operating system: Linux operating system based on arm and DSP-based DSP/BIOS.

Connect two operating system core technology is a DSP/BIOS bridge. DSP/BIOS bridge provides a way to use DSP seamless interface that allows developers to GPP (generic processor) on the use of standard application programming interface to access and control DSP runtime environment. Use TI CCS Company (Code Composer Studio) integrated development environment, from a developer's perspective, it seems only GPP OMAP processor completes all processing capabilities. In this way, developers do not need to separate the two processors, allowing programming programming work greatly simplified. In OMAP architecture, developers can treat as a single GPP on dual-processor OMAP platform for programming.

System software flowchart

In order to prevent the theft of fingerprints and use fingerprint film on fingerprint recognition system for deceptive attacks, take the "personal identification code (PIN) + fingerprint," two-factor authentication measures is very necessary.

User input PIN information to boot the system find the fingerprint library corresponding to a template, and the collected user's fingerprint 1 on 1 in the "best" match, so that both can prevent deceptive attacks, and do not need each to 1 to n successive match, you can efficiently and quickly complete the fingerprint identification process. System software main program flow shown in Figure 4.

Fingerprint identification process

1) image preprocessing.

Image preprocessing consists of three steps, namely, image segmentation and image enhancement and binary. Segmentation is read and cut input of fingerprinting, the basic no loss of useful fingerprint information reduce subsequent steps in the amount of data to be processed; image enhancement to the segmentation of the fingerprint image to smooth, sharpen, filtering, and other processing to improve picture quality; binary that is 8-bit greyscale fingerprint image converted to 0 and 255 binary image, use a threshold selection method of local threshold.

2) detailing.

In order to further compress the data, you also need the binarization image refinement process. Refine keep ridge of connectivity, the direction does not change, you should keep the center of the Ridge will remain basically unchanged. Refinement of Ridge will bypass and huokou, therefore must be refined after the denoising.

3) feature extraction.

Fingerprint feature information from the refinement of binarization image extraction. From thinning of binarization image find fork point or endpoint, and then start looking for from these patterns. Through these points, lines, shapes are calculated. These shape data, point of type and location of the point as the fingerprint image feature points is stored.

4) feature match.

In the extract fingerprint minutiae, you will get a set of vector points. Likewise, the fingerprint database to store the fingerprint template is a set of vector points, to determine whether two fingerprint match is transformed into judgement vector points two group matches. Fingerprint matching is based on two images feature points of maximum number of matches to determine the supported, if the maximum number of matches is greater than the required support, consider two fingerprint match.

FPS200 register settings

During initialization, to fingerprint sensor FPS200 corresponding register write control word,

Set the parameters of the fingerprint. One of the most critical is the DCR, DTR, PGC these three registers of parameter settings. This article has been through the trial registers of parametric optimum setting, Figure 5 shows the three registers respectively adjusted parameter value on image quality.

Conclusion

This article is based on embedded technology OMAP, design a complete, stand-alone embedded fingerprint recognition system, the system has been successfully tested in the lab, and is applying for approval.

As biometric authentication technology for the rapid development of the system expansion space will become more and more broad, system design and perfect. In-vehicle multimedia network development has become a trend, the system can be further developed GPS vehicle navigation system, OMAP platform open multimedia applications for future system expansion reserve enough space. Believe that the system will have a huge market potential.