Login| Sign Up| Help| Contact|

Patent Searching and Data

Document Type and Number:
WIPO Patent Application WO/1993/015473
Kind Code:
Fingerprint (FP) data are collected to collector (93) with best identification parameters (t, n, a, l, d, r, e) through FP data input device (52-61) in close cooperation with expert. FP data are checked, arranged in good order in a collector and stored in memory device (94, 95). Search FP is entered by FP data input device (50, 51) and matched with file FP (94, 95) at the matching device (91, 92).

Application Number:
Publication Date:
August 05, 1993
Filing Date:
December 24, 1992
Export Citation:
Click for automatic bibliography generation   Help
International Classes:
G06K9/00; G07C9/00; (IPC1-7): G06K
Foreign References:
Attorney, Agent or Firm:
Download PDF:
controller is provided to ensure the automatic feed of FP card.7. A system as claim
1. ed in Claim 2, characterized in that a control keyboard is directly installed in a FP data input device.
2. 8 A system as claimed in Claim 2, characterized in that FP minutiae positions are detected hy means of pulse received from encoder and FP minutiae positions counter circuit is introduced to prevent the mi soperat i on .
3. 9 A system as claimed in Claim 2. charaterized in that two direction linking dioά circuit is introduced in link¬ ing line between FP data input device and controller to overcome current leakage.
4. A method of defining afresh the FP identification parameters in the best mode.
5. A method as claimed in Claim 10, characterized in that the type of FP is classified into 27 classes basi¬ cally.
6. A method as claimed in Claim 10, characterized in that the minutiae positions are defined by starting point and end position.
7. A method as claimed in Claim 10, characterized in that specified regional data of FP are used as the FP identification parameters.
8. A method as claimed in Claim 10, characterized in that a polar coordinate system is introduced for the en¬ coding of FP .
9. A method for the realiaation of clustering in the 10 dimensions space by applying weight according to the type of FP in the 10 finger FP identification system.

This invention relates to a method and device for identifying f ingerpr ints .

In conventional system, the minutiae of fingerprints l ike a branch, end and so on are extracted through steps of inputting the fingerprint image by means of a ingle purpose processor, filtrating.binarying, thinning. retrieval .correction by expert and the other. and search f i gerpr i nt ( FP ) and fi le fingerprint are matched by using identification parameters like X-Y coordinates of a inul iae. i ts d i rect i on . concent rat i on , relation and so on.

As the conventional system requires very high quality of fingerprint entered, nd many identi fic t ion parameters are employed for matching, so confi urat ion of device becomes complicated and expensive.

The prior system has a low accuracy f ident i fication for fingerprint of l w quality or special pat tern.

In this invention an opt imum i ent i ficat ion parameters are. defined and the system was so arranged that FP features identification abil ity of special ists can be displayed to the maximum. thus raising the accuracy and speed.

In the FP data input devioe(52-Rl , in close cont ct with experts in paral lel , the i entificati n parameters of FP is entered direct ly with high accuracy and speed by means of key control board.

In this cheap FP data input device the FP pattern classifi¬ cation, finger No., core, direction, minutiae position etc. are so arranged that the knowledge of expert may be reflected to the maximum.

The FP data entered by FP data input device is directly encoded without special pretreating process and not passing the correction stage, and stored in file FP memory and match ¬ ed with search FP .

By using the FP identification system made up according to this invention, 3 millions of FP of 300 thousands persons have been classified and stored in the course of 5 months and the result of searching and examininsr the uncleaned or overlapped or partial FP shows that all the FP have been correctly detected as the first candidates.

Summary of the Invention

It is an object of the present invention to provide a me¬ thod and device for creating a single purpose FP data base useful for crime detection or banking business etc. while reducing the cost of device and raising the accuracy and speed of matching the search fingerprint with file FP.

Identification parameters such as FP pattern, number of finger, specific region information, rotation angle, core point, direct on, posit n of minutiae point , attribute etc. are exactly entered by means of FP data input device(50, 51).

As is seen in figure 1, plenty of FP data can be entered in parallel by connecting FP data input device with bus bar net¬ work, and the control ler(70, 71) can col lect the primary FP data entered by FP data input device to update and supplement

the fi le FP data base at any time.

The FP matching device(91 , 92) al lows to match, in close cooperation with the experts, the identi ication parameters of search FP entered by FP data input device(50, 51 ) with those of fi le FP.

The rotation of coordination system, the pairing of corres¬ ponding minutiae and evaluation of score are carried out in order to decide the best matching.

A cl ssific tion by clustering is carried out in the 10 fin ers FP system for personal veri fication, thus r ising the search speed.

Brief Description of Drawings Figure 1 shows a block diagram of FP identification system. Figure 2 shows a block diagram of input subsystem.

Figure 3 shows a block diagram of FP data input device.

" i^urc _ • ■ :, ; s .. tic:..!- -: ? - ' • " " ■- ' ^ - 'Z' J - ' ice .

Detailed Discription of the Invention

FP identification system is composed of classific tion ter¬ minal system and identific ti n princip l system (fig. 1).

The identification host system is composed of input sub¬ system and matching subsystem.

A classification terminal system is composed of FP collector (93) and several input subsystems.

Input subsystem is composed of FP data input device, col lect' or, external memory and display device (fig. 2).

FP data input device(501) is composed of FP minutiae sensor, FP card automatic feeder, FP image fi ltration and expansion-

projection part, expert interface, FP image input device, cont ¬ roller ' Z'- _...;, y

FP image fi ltration and expansion-pro ection part process a FP image in optical way to provide a FP data very exactly.

Expan ion-projection part(G04) is composed of projection screen consisting of projection lens, prism, reflector, frosted glass, and illumination unit and ventilator.

Magnification of optical system is R.7 folds.

The reflector is made of a gl ss whose surface is coated with an aluminium vapour deposition.

Projection screen is made of a frosted class whose one side was subjected to a sand b.l as t .

100* h logen lamp is used for ill mination.

There is a ventilator to entracl the heat generating from i 11 u ina ion .

A FP minutiae sensor(601) automaticall measures the minu¬ tiae position of the FP in close connection with the expert. It consists of t o linear encoder (x, y directions). Each encoder consists of a movable verni er(glass scale), a fixed leveling rod, two infrared ray luminescent elements and a pair of sensors for converting a i lacement into pulse.

The displacement is converted into a row of pulse by micro leading scale of movable vernier and two infrared rays lumi¬ nescent e1 e en t sensors . and the moving direction is determined by the phase difference pulse emitting from two sensors.

A reversible counter performs addition and subtraction by moving directions and a row of pulse, and the output value of the reversible counter represents precisely the displacement value-

, order to eliminate the cumulat ive error the encoder is furnished wi th a pair of infrared rays luminescent diodesen- sor r s ectively in the central posi t ion of x and y so as to be l n i 1 l a 1 i zed .

The Fp image reader(603) converts a FP into a mul tilevel digitalized image data to be sent to a control ler.

The resolution of FP image data put in from above-ment i ond device is 640 x 480 and i ts gray ( coneent rat i on ) is a level of G4.

FP card automat ic feeder ensures an automat i c feed of FP card according to the operation of an expert .

FP card automatic feeder(G02 ) is composed of t o sm l l pulse motors, 'nitput control lers, two reciprocat ing carriages, FP card fixing table along the direct ions x, y.

The control pulse is generated at the control ler of input sub¬ system ami then sent in.

The expert interface(GOG) consist of H ve nier m ving device, a FP projection screen and a keyboard.

Roller bearings are fixed in the vert ical and horizontal direc t ions of the vernie mov i ng device and ha 1 anee- e i gh t in a er ical direct i n, thus el i inat i g vibrat ion and p and ensur ing a smooth an r ci e displacement .

The expert di rect lv puts in the tvpe of FP(t ) . number of a finger(n) , direction of FP(d) , posi t ion of FP miniit iae(x. y) and an attribute(a) etc.

Then the expert can put in precι el \ arid quickly the FP iden¬ ti fication parameters in an interact ive way through image data of input subsy tem.

The FP image entered FP image input device(G03) is subject

to filtration and binarying and then displayed on the screen of a contro11 er .

The FP data entered is characterized in the following identi¬ fication parameters:

F(t, n. a,l, d, r, s ) (1) t: type of FP (1-27) n: number of finger (0-9) a: attribute of minutiae

(starting, ending) 1 : polar radius (1-120) d: polar angle (1-3G0) r: number of ridge (0-23) srspecifyed region (1-45) Mi . the minutiae of fingerprint is encoded by the polar radiuce li, polar angle di , number of ridge from polar point to Mi (ri) in the polar coordinate system having its attri¬ bute ai and polar point Oc ti3 auoriginal point, i .e.

Mi(ai , li. di, ri) (2)

After all .searcVi and file FP are characterized in a minut i ae co11 ect i on

M={Mi(ai .1 i ,di ,ri)l i=l,n] (3) The controller of input subsystem (27-81 ) conducts general control over the above mentioned device. The control and processing device (72-81) collect the FP identifica ion data classified through linking unit to form a primary FP file (3) and send the latter tore 'Tθr

with FP image data entered through FP image reading device to FP col 1 ect ion .

Identification principal system arranges in good order the primary FP data base.

The input subsystem of identific tion principal system exactly puts in the FP identification parameters by expert with regard to partial FP or a FP of low quality.

The expert defines the type of FP . number of a finger, core point, position and attribute, direction of minutiae etc. t'α ou-'h F image entered by input device.

The matching device (91.92) of a identification principal system decides degree of response of pair of minutiae which gives the best score with regard to a type of FP,number of a finger ratio of expansion and reduction of i age , sett ing of core point, t ransformat i on( rotat ion) of coordinate system.

Once the type of FP, number of a f inger , expans ion and reduc¬ tion ratio of an image, core point are defined, e can set the specific region of search FP,while rotating the polar coordi¬ nate system at the step ^g , and find out all file FP satisfying the condition of formular (4) with regard to the file FP having minutia number less than in the region corresponding to file FP .

Λ 1 i j=| 1 i-1 j I Tl

Λdi j=| di- j I < Td


& ai j=| a i -a j | = 0 Λ ri j=| ri-rj I / Tr .where Tl .Td,Tr represent .resho'j.d v-.lues,

Td is a dynamic threshold value changinc: according to polar ra By going through above-mentioned process. e can reduce

considerably the number of candidates of file FP matched with search F .

We seek file candidates which is best matched with the selected finite candidates through paring of minutiae and evaluation of scores. Table 1 shows the matched search FP and file FP.

Table 1 shows file minutiae proximate to search minutiae, assuming that the number of possible pairs of search minu¬ tiae and file minutiae (Si,Fj) most proximate thereto is N (for example S1.F12) on the table 2, e can form N match lists for each pair.

Table 3 represents a initial match list for match pair (S1.F12) ,(S1,F1) .

Table 4 represents a final match list for initial match list. In this way match lists are prepared for all the possible match pairs.

Scores are evaluated for each match list. If the score for each minutiae pair is obtained the order of the corres¬ ponding files minutiae in table 2 is arranged in the order of big score for match list, but not according to the dis¬ tance.

Table 5 shows the pairs of minutiae corresponding to the scor .

The most right column represents finally matching pairs.

Tiotal match score is represented as a sum of match score of each match pair.

The best matching data is those input data corresponding to the highest score among the matching scores with regard to the type of FP.number of a finger, core point, each ro-

tation of polar coordinate system within the scope of ±12 degree A θ

The FP identification system is backed with single pur¬ pose FP identification data base which detects case FP and personal data, renders services for vi ew, stati t i cs , correct i on,de1 et ing, inspect i on , s t orajrc etc, and incorpo¬ rates FP data and personal data.

The program system of this invention is so arranged as to be used conveniently for identific tion of criminals in the Legisl- 1 1ive organs and for id ntification of per¬ sons in the banking institutions by means of FP . Cl ims

1. \ fingerprint identific tion system charact rized in that it is constituted so as to input the FP data simultaneously by means of several input subsystems.

2. A system as claimed in Claim 1, characterized in that the FP data input device is consti tuted so as to ensure FP minutiae identification ability of expert to the maximum.

3. A system as claimed in Cl im 2, characterized in that it is installed so as to more precisely and smoothly the sliding device wi th vernier attached having 2 degrees of freedo .

4. A system as claimed in Cl im 2. churn teri ed in that

FP minutiae position sensor is linked to the sliding device.

5. A system as claimed in Claim 2. ch racterized in that the keyboard for FP data input is linked with the vernier s1 iding device.

6. A system as claimed in Claim 2. characterized in that