CN105631388A - Linux-based embedded fingerprint identification system - Google Patents

Abstract

The invention provides a linux-based embedded fingerprint identification system. The system is realized based on the important biological identification technology as the most mature branch of the current biological identification technology development. In the linux environment, an embedded fingerprint identification system is configured. The system mainly has the functions of acquiring fingerprint images, pre-processing the images, extracting feature points and the like.

Description

Fingerprint-based User Authentication for Embedded Systems based on LINUX

Technical field

Fingerprint-based User Authentication for Embedded Systems based on LINUX relates generally to computer application system aspect field.

Background technology

Fingerprint identification technology (FingerprintVerificationTechnology) be utilize somatic fingerprint feature carry out authentication ' kind technology is current biological detection }, study the most deep, most widely used general, develop the most ripe technology. In recent years, along with ecommerce, ridicule upper trade, ridicule ��: banks etc. develop rapidly and people's improving constantly the requirement of personal information, and Fingerprint-based User Authentication for Embedded Systems will have ample scope for one's abilities. Research fingerprint recognition system, holds society's demand to fingerprint recognition, makes the exploitation application of embedded system can tightly grasp the developing direction of current finger print identification. and foot has giant's using value.

Summary of the invention

Do not have to find the application materials about this system aspects by national patent retrieval.

Purpose foot one minutiae point automatically extracted out of judgement of the algorithm for recognizing fingerprint of native system in given position and, J upwards actually foot streakline end points, streakline divide justice point or false detail point, provide more reliable foundation for follow-up details in fingerprint Point matching link. The main contents of algorithm include choosing and top process of minutiae point neighborhood. Neighborhood characteristics analysis etc.

One. the acquisition of fingerprint image:

The collection of fingerprint image is the important component part of Automated Fingerprint Identification System (AFIS). Fingerprint sensor can be divided into optical pickocff, heat sensitive sensor, voltage sensitive sensor and sonac etc. according to the little same of detected object. Difference according to device, raised path between farm fields is divided into cmos device sensor and CCD device sensor again. Their operation principle is all that after inspection, biological characteristic is converted into the image information that system may identify which. In fingerprint recognition system, reliable and cheap image capture device is system normal operation, reliably key.

Adopt capacitive fingerprint sensing device technology herein. Capacitance sensor is to be integrated with more than 10 ten thousand capacitance sensors at single wafer L, it it is the surface of insulation outside it, when the finger of user is placed on above, finger skin constitutes another pole of capacitor array, and the capacitance of capacitor is different due to the distance (ridge of fingerprint and paddy are relative to the distance of another pole) between conductor. By measuring, interrogate medium and small same capacitance and obtain complete fingerprint image. The more typical product of the type is the FPS200 fingerprint sensor of Veridicom company, in fingerprint collecting process, regulates the parameters such as capacitor discharge time according to feedback information to strengthen its sensitivity. Its area only has 1.5cm*1.5cm, integrated 90000 electric capacity, and with high-speed a/d converter part. The resolution of this product 6 & is 500dpi. System provides USB interface, SPI to meet r] and 8 BITBUS network interfaces. Capacitive fingerprint sensing device has difficult ten deceptions, eliminate daily fingerprint change, volume is little, high-resolution, the impact of light light around elimination, fingerprint image picture element is the highest, induction zone is not easy to wear, the raw l of the l of I:l is as features such as sizes. Advantage and moderate price thereof due to capacitive fingerprint sensing device. become the sensor of main flow in fingerprint recognition system.

Two. the pretreatment of image:

The image obtained has a lot of noise, and this causes mainly due to working environment at ordinary times, and such as, finger is dirty, and finger has knife injury, scar, trace, dries, moistens or tear. Expecting more visible clean being not easy to of ratio, this needs fingerprint image is carried out pretreatment. First have to be normalized, the segmentation of image enhaucament, image, binaryzation, refinement and refinement post processing. After refinement, generally also the image processed is carried out again refinement post processing. Because, four noise likes such as refined image also jagged, short-term, breakpoint, aperture they can form the characteristic point of vacation, it is therefore necessary to remove.

Three. the extraction of characteristic point:

The streakline of fingerprint image is moved towards in feature extraction, and streakline breakpoint, cross point etc. can fully represent that the form of this fingerprint only feature numerical value of 4 property is expressed. Accuracy for comparison, it is desirable to feature extraction algorithm extracts validity feature as much as possible. The false feature caused by a variety of causes is filtered during I department. To the fingerprint image after refinement, generally conventional characteristic point can be gone out simply by the template detection of 3x3. What have then also needs to searching general characteristic point such as core point and triangulation point etc.

Main Basis following information during coupling: the ridge count etc. between the position of minutiae point, direction, the stricture of vagina type of fingerprint, the trend of crestal line, details. Fingerprint was made up of crestal line and valley line alternate interlocking, after Fingerprint Image Binarization, we can be clear that this structure, find the end points same characteristic point corresponding to the bifurcation on fingerprint valley of fingerprint ridge line E, the bifurcation same characteristic point corresponding to the end points on valley line that crestal line is foretold by analysis. Therefore all end points of fingerprint and bifurcation only need to can all be extracted by we by two width refined image extraction end points.

Four. characteristic matching:

Characteristic matching is to be compared with the eigenvalue of deposited fingerprint in fingerprint base by the eigenvalue of newly inputted fingerprint. Find out most like fingerprint as the output result identified. Namely described fingerprint authentication/identification process. Impact due to various water hole ropes. Same fingerprint inputs the feature templates very raised path between farm fields of gained for twice can be different. Therefore, as long as there being the character modules of input fingerprint to pull when pulling similar to stored mould, just say that the two fingerprint is coupling. Then the problem producing relevant weighing apparatus scape standard.

Generally, matching result represents with " matching degree ". When matching degree more than a certain explain value time, it is believed that two fingerprint matchings: contrary, when less than this value of explaining, it is believed that do not mate. Close value size to be manually set generally according to factors such as experience, security of system ranks. Explain value bigger time, security of system increases, but FRR will raise; Otherwise, system ease for use is good, but FAR to raise.

Finally, the Linux process in embedded system has been studied by we, to can automatically run, it would be desirable to rewrite its script file, in etc/profile script, does following interpolation:

PATH=/usr/qpe/bin:$PATH

LD_LIBRARY_PATH=/usr/qpe/1ib:$LD_LIBRARY_PATH

OTDIR=/usr/qpe

QPEDIR=/usr/qpe

cd/usr/qpe/bin

ExportPATHLD_LIBRARY PATHQTD{RQPEDIR file destination performs after starting automatically, comprises fingerprint recognition program in interface, and clickable icon both can make program perform.

Claims (5)

1. it is characterized in that the acquisition of fingerprint image, the pretreatment of image, the special extraction of sheet point, characteristic matching based on the Fingerprint-based User Authentication for Embedded Systems of LINUX.

2. the important component part of Automated Fingerprint Identification System (AFIS) it is mainly according to the acquisition of the fingerprint image in claim 1.

3. first have to be normalized according to the pretreatment of the image in claim 1, the segmentation of image enhaucament, image, binaryzation, refinement and refinement post processing; After refinement, generally also the image processed is carried out again refinement post processing.

4. the streakline of fingerprint image is moved towards in the feature extraction of extracting according to the special sheet point in claim 1, and streakline breakpoint, cross point etc. can fully represent that the form of this fingerprint only feature numerical value of 4 property is expressed.

5. it is that the eigenvalue of deposited fingerprint in the eigenvalue of newly inputted fingerprint and fingerprint base is compared according to the characteristic matching in claim 1.