WO2006078054A1 - パターン情報登録装置、パターン情報登録方法、パターン情報登録プログラム及びパターン照合システム - Google Patents
パターン情報登録装置、パターン情報登録方法、パターン情報登録プログラム及びパターン照合システム Download PDFInfo
- Publication number
- WO2006078054A1 WO2006078054A1 PCT/JP2006/301164 JP2006301164W WO2006078054A1 WO 2006078054 A1 WO2006078054 A1 WO 2006078054A1 JP 2006301164 W JP2006301164 W JP 2006301164W WO 2006078054 A1 WO2006078054 A1 WO 2006078054A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- pattern information
- data
- information data
- pattern
- information registration
- Prior art date
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/12—Fingerprints or palmprints
- G06V40/1365—Matching; Classification
- G06V40/1371—Matching features related to minutiae or pores
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/98—Detection or correction of errors, e.g. by rescanning the pattern or by human intervention; Evaluation of the quality of the acquired patterns
- G06V10/993—Evaluation of the quality of the acquired pattern
Definitions
- Pattern information registration device Pattern information registration method, pattern information registration program, and pattern matching system
- the present invention relates to a pattern information registration apparatus, and more particularly, to a pattern information registration apparatus, a pattern information registration method, a pattern information registration program, and a pattern matching system that determine the quality of pattern information data.
- Pattern information such as fingerprints is widely used as a means of identity verification and crime prevention.
- pattern registration is generally performed first, and then verification is performed based on this pattern registration.
- the conventional pattern information registration method it is difficult to set a standard for determining the quality of pattern information data. For this reason, pattern information data that does not satisfy the accuracy required for the verification device is registered. As a result, there is a problem that it is difficult to distinguish between data when there are a large number of data. Thus, the conventional pattern information registration method cannot determine the quality of pattern information data.
- the method disclosed in Japanese Patent Laid-Open No. 8-2 6 3 6 5 8 relates to a fingerprint registration method. Therefore, a registration window that detects normal feature points and pseudo feature points of a fingerprint and has the normal feature points as the center. The number of pseudo feature points in is calculated. Then, the total number of pseudo feature points in all registered windows is summed and divided by the number of registered windows to obtain the average number of pseudo feature points in the registered window. Since the pseudo feature point is a feature point generated by cracks, wrinkles, etc., the smaller the average number of pseudo feature points, the better the image quality of the fingerprint image. Therefore, when the average number of pseudo-feature points is less than or equal to the threshold value, it is determined that the image quality of the captured fingerprint image is good and registration processing is performed.
- the method disclosed in Japanese Patent Laid-Open No. 2000-339461 relates to a method for creating a pattern dictionary, and creates an additional dictionary to reinforce the weaknesses of the pattern dictionary.
- creating an additional dictionary the feature vector of the pattern to be learned and the recognition result are input, and the neighborhood region centered on each feature vector is found.
- the neighborhood region is a set of feature vectors that satisfy a certain inclusion condition.
- an inclusion condition for example, it can be set that all feature vectors in the neighborhood area are misread data. In this way, the maximum of the neighborhood area obtained in this way is obtained and written in the additional dictionary.
- Japanese Patent Application Laid-Open No. 2002-288667 describes an example of a pattern collation apparatus that performs collation using a probability of coincidence when compared with an arbitrary pattern.
- the inspection is performed using the probability that the feature points match more than when comparing the inspection figure and the model figure. The figure is compared with the model figure.
- the method disclosed in Japanese Patent Laid-Open No. 8-2.63658 is based on a pattern information data based on pseudo feature points. This is a method of judging whether the evening is good or bad. Pseudo feature points are different from feature points (regular feature points) that are data of pattern information itself. The determination of pass / fail of pattern information data according to the method of Japanese Patent Laid-Open No. 8-2 6 3 6 5 8 is not based on the special points of pattern information data.
- the method disclosed in Japanese Patent Laid-Open No. 2 00 0-3 3 9 4 6 1 is a method for reinforcing weaknesses of the pattern dictionary, and is intended to determine whether the pattern information is overnight. Not.
- JP-A-3-2 1 8 5 75, JP-A-6 3-4 3 81, and JP-A-1-1 3 1 9 7 8 As described below, the quality of the pattern information is not judged.
- the size of the fingerprint is taken into consideration. This is because only the number of feature points is considered. In these methods, when there are more than a certain number of feature points, it is judged as good. However, in these methods, the size of the fingerprint is not taken into consideration, and therefore the quality of the pattern information data is not judged.
- An object of the present invention is to provide a pattern information registration device, a pattern information registration method, a pattern information registration program, and a pattern verification system that solve the above-mentioned drawbacks of the prior art and determine the quality of pattern information data. is there.
- the present invention for achieving the above object is a pattern information registration apparatus for selecting and registering pattern information data to be registered for use in pattern matching, Forming arbitrary pattern information data having the same number of feature points as the determination target pattern information data, calculating an identification accuracy value representing a degree of coincidence between the determination target pattern information data and the arbitrary pattern information data; The quality of the pattern information data to be determined is determined based on the identification accuracy value.
- the arbitrary number of pattern information data is formed by randomly arranging the same number of feature points in a predetermined area of the determination target pattern information data, and the same number of feature points respectively corresponds to the pattern information of the determination target.
- the identification accuracy value is calculated on the assumption that the arbitrary pattern information data matches the pattern information data to be determined.
- pattern registration is generally performed first, and verification is generally performed thereafter. At the time of pattern registration, if the identification accuracy of the registered pattern information data is high, it is easy to distinguish from many other pattern information data. On the other hand, if the identification accuracy of the registered data is low, It becomes difficult to distinguish from many other data. Therefore, it is required to obtain the identification accuracy value of the pattern information data registered at the time of data registration by determining the quality of the pattern information data.
- the identification accuracy value is defined as a probability that the pattern information data cannot be distinguished from any pattern information data having the same number of feature points, and the calculation is performed on the pattern information data to be determined. Calculate against
- the image area and the feature point are extracted from the pattern information data to be determined.
- arbitrary data having the same number of feature points as the pattern information data to be determined is formed, and the probability that the pattern information data to be determined and the arbitrary data cannot be identified is calculated.
- the situation where both data cannot be identified occurs when the arbitrary data matches the pattern information data to be registered. Therefore, it can be seen that the probability that the judgment target pattern information dender cannot be distinguished from arbitrary data is equal to the probability that the arbitrary data matches the judgment target pattern information data. From the above, it is derived that the identification accuracy value can be obtained by calculating the matching probability.
- the calculation for obtaining the probability of matching is performed as follows.
- the pattern information data is special in the image area of the pattern information data to be judged. It is formed by randomly arranging the same number of feature points as saddle points. If all of the arranged feature points are placed in the vicinity of each feature point in the pattern information to be judged, it is considered that the arbitrary data matches the pattern information to be judged.
- the accuracy of pattern information data is obtained by calculation, and the quality of the pattern information data is determined by comparing with a predetermined threshold value.
- the reason for this is to determine whether the identification accuracy of the pattern information data is good or bad by calculating the identification accuracy value of the pattern information data.
- pattern information data suitable for pattern matching can be selected regardless of the amount of pattern information data (for example, the number of feature points).
- FIG. 1 is a block diagram showing the configuration of the pattern information registration apparatus according to the first embodiment of the present invention.
- FIG. 2 is a flowchart for explaining the operation of the pattern information registration apparatus according to the first embodiment of the present invention.
- FIG. 3 is a flowchart for explaining an example of determining whether or not the fingerprint data identification accuracy is good according to the first embodiment of the present invention.
- FIG. 4 is a diagram showing the feature points of the fingerprint data to be determined according to the first embodiment of the present invention.
- FIG. 5 shows the positions of the feature points of the fingerprint data to be determined according to the first embodiment of the present invention. It is a figure which shows the example which determines with the position of the feature point of arbitrary data being in agreement.
- FIG. 6 is a diagram showing an example in which it is determined that the position of the feature point of the arbitrary data and the position of the feature point of the fingerprint data to be determined do not match according to the first embodiment of the present invention.
- FIG. 7 is a diagram showing a state in which feature points of arbitrary data are formed according to the first embodiment of the present invention.
- FIG. 8 is a diagram showing an example in which the feature points of the arbitrary data according to the first embodiment of the present invention match the feature points of the fingerprint data to be determined.
- FIG. 9 is a diagram showing an example where the feature points of the arbitrary data and the fingerprint feature points to be judged do not match according to the first embodiment of the present invention.
- FIG. 10 is a view for explaining a specific example of pass / fail judgment of pattern information data according to the first embodiment of the present invention.
- FIG. 11 is a block diagram showing the configuration of the pattern information registration apparatus according to the second embodiment of the present invention.
- FIG. 12 is a flowchart for explaining the operation of the pattern information registration apparatus according to the second embodiment of the present invention.
- FIG. 13 is a flowchart for explaining an operation example of the pattern information registration apparatus according to the second embodiment of the present invention.
- FIG. 14 is a diagram showing the fingerprint data input for the first time according to the second embodiment of the present invention.
- FIG. 15 is a diagram showing fingerprint data inputted a second time according to the second embodiment of the present invention.
- FIG. 16 is a diagram showing an example in which common features are extracted from two fingerprint data by the common feature extraction unit of the second exemplary embodiment of the present invention.
- FIG. 17 is a diagram showing a common area and common feature points of two fingerprint data extracted by the common feature extracting means according to the second embodiment of the present invention.
- FIG. 18 is a diagram showing an example of feature points when a face is used as pattern information according to the second embodiment of the present invention.
- FIG. 19 is a diagram showing an example of special points when the position of the end of the eye is used as pattern information according to the second embodiment of the present invention.
- FIG. 20 is a block diagram showing the configuration of the pattern matching system according to the third embodiment of the present invention. BEST MODE FOR CARRYING OUT THE INVENTION
- FIG. 1 is a block diagram showing the configuration of the pattern information registration apparatus 10 according to this embodiment.
- the pattern information registration apparatus 10 is configured by an input unit 2 0, a data processing unit 30, an output unit 40, and a registration unit 50.
- the input unit 20 includes an input unit 21 and a feature amount extraction unit 22. These means generally operate as follows.
- the input means 21 is realized by a scanner or the like, and has a function of inputting, for example, a fingerprint image as pattern information data.
- the feature quantity extraction means 22 has a function of extracting feature quantities included in the pattern information data.
- the feature amount is an amount for identifying an image and includes a feature point and an image area.
- the data processing unit 30 includes an identification accuracy value calculation unit 3 1 and a pass / fail determination unit 3 2.
- the identification accuracy value calculation means 31 can calculate the identification accuracy value of the pattern information input from the input unit 20.
- the pass / fail judgment means 3.2 is a means for judging pass / fail of the pattern information data.
- the discrimination accuracy value of the pattern information data obtained by the discrimination accuracy value calculation means 31 does not exceed a predetermined threshold, the pattern information data is It can be determined that the pattern information data is not good when the threshold is exceeded.
- the data processing unit 30 is realized by a computer processing device that executes the functions of the identification accuracy value calculation means 31 and the pass / fail judgment means 32.
- the output unit 40 can output the determination result made by the pass / fail determination means 32.
- the output unit 40 is provided with a display or printer.
- the registration unit 50 performs the registration process of the pattern information data determined to be good by the quality determination means 32, and stores the registered data in the database 60. Next, the operation of the pattern information registration apparatus 10 according to this embodiment will be described.
- FIG. 2 is a flowchart for explaining the operation of the pattern information registration apparatus 10 according to the present embodiment. Refer to the main part of Figure 1 as necessary.
- pattern information data is input by the input means 21 (step 2 0 1), and the feature quantity is extracted by the feature quantity extraction means 2 2 (step 2 0 2).
- the identification accuracy value calculation means 31 calculates an identification accuracy value using the feature amount of the pattern information data provided from the input unit 20 (step 20 3).
- the pass / fail judgment means 32 determines that the pattern information data is good when the discrimination accuracy value calculated by the discrimination accuracy value calculation means 31 does not exceed the threshold value. Is not good (step 2 0 4).
- step 2 0 5 When it is determined that the pattern information data is good (step 2 0 5), a determination result indicating that the pattern information is good is output by the output means 40 (step 2 0 6).
- the registered data is stored in the database 60.
- the output unit 40 outputs a determination result indicating that the pattern information data is not good (step 2 0 8).
- determining the quality of the pattern information data, and performing registration processing such a function is provided inside the computer device.
- a program application for realizing the characteristic function of the present invention is stored in a storage medium, and the program is executed by a computer device. By executing the program, it can also function as the pattern information registration device 10.
- the input means 2 1 of the input unit 20 is provided with a fingerprint sensor, Capture the fingerprint image.
- the input unit 20 can input a fingerprint image captured by a fingerprint sensor or fingerprint data after performing feature extraction via a network. It is also possible to input a fingerprint image or fingerprint data recorded in a storage device such as a memory or a hard disk inside or outside the data processing unit.
- the output unit 40 can not only output the result to a display or a printer, but can also provide the determination result to another processing apparatus through the network.
- the results can be recorded in an internal or external memory or hard disk of the data processing unit.
- the storage medium in which the registered data recorded in the database 60 connected to the registration unit 50 is written can be used in other devices.
- FIG. 3 is a flowchart for explaining an example of determining whether or not the fingerprint data identification accuracy is good according to the present embodiment. Refer to the main part of Figure 1 as necessary.
- fingerprint data is input to the fingerprint sensor provided in the input means 21 (step 3 0 1).
- the feature amount extraction means 21 extracts the fingerprint region and the number of feature points, which are the feature amounts for calculating the identification accuracy value (step 3 0 2).
- the identification accuracy value of fingerprint data can be defined with the probability that it cannot be distinguished when compared with other fingerprint data with the same number of feature points.
- the calculation of the probability that cannot be distinguished will be explained in the calculation of the identification accuracy value described later.
- the feature quantity to be used may be the feature quantity of the fingerprint data determined by the pattern information registration apparatus of the present invention. desirable.
- the feature data match / mismatch judgment criteria also coincide with the fingerprint data judgment criteria determined by the pattern information registration apparatus of the present invention.
- Equations 1 and 2 are disclosed in Japanese Patent Publication No. 2002-288667.
- N points are the feature points of the fingerprint area.
- P3 (N) (N7tR2 / S) N (Equation 3)
- the identification accuracy value of fingerprint data is defined by the probability that the fingerprint data cannot be distinguished when compared with other fingerprint data having the same number of feature points. This probability As will be described later, when the fingerprint data is compared with arbitrary pattern information data having the same number of feature points as the fingerprint data, it can be calculated as the probability that the two match.
- arbitrary pattern information data having the same number of feature points as the fingerprint data is abbreviated as arbitrary data.
- the identification accuracy value In the calculation of the identification accuracy value, first, an image region and a feature point are extracted from the pattern information data. Next, arbitrary data is formed, and the probability that pattern information data to be judged cannot be distinguished from arbitrary data is calculated. The situation in which the two data cannot be distinguished occurs when the discretionary data matches the pattern information data to be judged. Therefore, the probability that the pattern information data to be judged cannot be distinguished from the arbitrary data is equal to the probability that the arbitrary data matches the pattern information data to be judged. From the above, it can be derived that the identification accuracy value can be obtained by calculating the probability that the arbitrary data matches the pattern information data to be determined.
- arbitrary data is formed by randomly placing the same number of points as the feature points of the pattern information data in the image area of the pattern information data to be judged.
- the probability that the same number of points are arranged in the vicinity of the feature points of the pattern information data is obtained.
- the probability is calculated assuming that the vicinity of the feature point is within the radius R around the feature point.
- FIG. 4 is a diagram showing feature points of the fingerprint to be determined according to the present embodiment.
- one end point and one branch point which are characteristic points of fingerprint data, can be seen.
- FIG. 5 is a diagram illustrating an example in which it is determined that the position of the feature point 1 1 1 of the fingerprint data to be determined according to the present embodiment is the same as the position of the feature point 1 2 1 of the arbitrary data.
- the figure is an enlarged view of a part of the image that is overlaid with the fingerprint data and comparison data.
- the feature point 1 2 1 of the arbitrary data is located in the area within the radius R around the feature point 1 1 1 of the fingerprint data.
- the feature point 1 of the arbitrary data 2 1 is determined to be the same as the feature point 1 1 1 of the fingerprint data.
- Figure 6 shows the position of the feature point 1 2 1 of the arbitrary data and the fingerprint data to be judged according to this example. It is a figure which shows the example which determines with the position of the overnight feature point 1 1 1 not matching.
- the feature point 1 2 1 of the arbitrary data is located outside the area within the radius R around the feature point 1 1 1 of the fingerprint data.
- the position of feature point 1 2 1 is determined not to match the position of feature point 1 1 1 of fingerprint data.
- the identification accuracy value of fingerprint data is defined by the probability that it cannot be distinguished when fingerprint data is compared with arbitrary data. Considering that the fact that fingerprint data cannot be distinguished from arbitrary fingerprint data means that they match, fingerprint data can be calculated by calculating the probability that both match with the feature points of the actual fingerprint data. Can be obtained.
- FIG. 7 is a diagram showing how the feature points 1 2 2 of arbitrary data are formed according to the present embodiment.
- the feature points 1 2 2 of the fingerprint data there are five fingerprint data feature points 1 2 2 in the fingerprint region 1 0 1.
- the five feature points 1 1 2 of the fingerprint data are rearranged in Rashidam.
- the feature point 1 1 2 of the fingerprint data moves to the position of the feature point 1 2 2 of the arbitrary data.
- the feature point 1 2 2 of the arbitrary data is inside the circle centered on the feature point of the fingerprint data 1 1 2, the feature point 1 2 2 of the arbitrary data and the feature point of the fingerprint data 1 1 2 match If the feature point 1 2 2 of the arbitrary data is outside the circle centered on the feature point of the fingerprint data 1 1 2, the radius R of the circle for determining that they do not match Is set by the identification accuracy value calculation means 3 1 (step 3 0 3).
- FIG. 8 is a diagram showing an example in which the feature point 1 2 2 of the arbitrary data according to the present embodiment and the feature point 1 1 2 of the fingerprint data to be judged match.
- the feature points 1 2 and 2 of the arbitrary data are both within the radius R around the feature points 1 1 and 2 of the fingerprint data. It can be seen that the positions of the feature points 1 1 and 2 are the same at all feature points.
- the probability that the state of Fig. 8 will be realized can be calculated by setting the radius R in Equation 3 above.
- FIG. 9 is a diagram showing an example in which the feature point 1 2 2 of the arbitrary data according to the present embodiment does not match the feature point 1 1 2 of the fingerprint data to be determined.
- the feature points of arbitrary data 1 2 2 one feature point is located outside the area within radius R around the feature point of fingerprint data 1 "" Even 1 1 2 ". Point 1.2 2 and fingerprint data feature 1 1 2 do not match.
- the identification accuracy value of the fingerprint data is calculated by the identification accuracy value calculation means 3 1 ( Step 3 0 4).
- the pass / fail judgment means 3 2 compares the discrimination accuracy value calculated by the discrimination accuracy value calculation means 31 with a predetermined threshold (step 3 0 5).
- the pass / fail judgment means 3 2 is determined to be good fingerprint data with high identification accuracy (step 3 0 7).
- the output unit 40 For the fingerprint data determined to have high identification accuracy, the output unit 40 outputs a determination result indicating that the identification data is good (step 3 0 8).
- the pass / fail judgment means 3 2 determines that the fingerprint data is not good and has low identification accuracy. (Step 3 1 0).
- the output unit 40 For fingerprint data determined to have low identification accuracy, the output unit 40 outputs a determination result indicating that the identification accuracy is not good (step 3 1 1).
- the number of feature points is five has been described.
- the number of feature points is one.
- the probability that arbitrary data matches fingerprint data is high. For this reason, the identification accuracy value increases and the identification accuracy decreases.
- the pattern information registration device 10 of this embodiment is a device that selects and registers pattern information data to be registered for use in pattern matching.
- the pattern information data with the same number of feature points is formed, the identification accuracy value indicating the degree of coincidence between the pattern information data to be determined and the arbitrary pattern information data is calculated, and the determination is made based on the identification accuracy value.
- the quality of the target pattern information data is judged.
- Arbitrary pattern information data is formed by randomly arranging the same number of feature points in a predetermined area of the judgment target pattern information data, and the same number of feature points is the feature point of the judgment target pattern information data. If it is placed in the vicinity of, the discriminant accuracy value is calculated assuming that any pattern information data matches the pattern information to be judged.
- a false match rate (FMR) required for an authentication device that uses fingerprint data registered by the fingerprint registration system of the present invention for authentication can be used. There is no particular problem if the identification accuracy value matches the FMR required by the authentication device.
- F M R required for the authentication device may be used as it is, or a value obtained by multiplying F M R by an appropriate safety factor (less than 1) may be used.
- the radius R of the circle is equal to each other in the circles centered on the feature points in Equation 1 and Equation 2. Therefore, the area of the fingerprint area is divided by the number of feature points, and the calculated area is set to a value smaller than the radius of a circle having an area of 1 Z 10 It is preferable.
- the pattern information registration apparatus 10 it is possible to calculate an identification accuracy value that reflects the quality of the identification accuracy using only the pattern information data to be determined.
- the identification accuracy value calculated by Equation 3 is small, so the identification accuracy is high.
- the identification accuracy value calculated by Eq. 3 is large and the identification accuracy is low.
- FIG. 10 is a diagram for explaining a specific example of the pass / fail judgment of the pattern information data according to the present embodiment.
- FIG. 10 a specific example of pass / fail judgment for two pattern information data (1) and (2) is shown.
- the area S is set to 100mm2, and the radius R is set to lmm.
- the number of feature points included in pattern information data (1) is 10, and the number of feature points included in pattern information data (2) is 8.
- the threshold of the discrimination accuracy value for pass / fail judgment is set to 1 ⁇ 10 ⁇ 6.
- the identification accuracy values calculated by Equation 3 are 9.3 X 10-7 for pattern information data (1) and 1.6 X for pattern information data (2), respectively.
- pattern information data (1) can be registered in the judgment step, pattern information data.
- pattern information data (2) Is determined not to be registered.
- the calculation of the identification accuracy value described in the present embodiment is a concept that is not unique. As described above, the identification system value is calculated with the probability that it matches the pattern information data to be judged when the same number of feature points as the pattern information data to be judged are randomly placed. In order to do this, it is necessary to change the arrangement of the same number of feature points by reciprocal times of the probability of matching.For example, if the identification accuracy value is 0.01, it is necessary to change the arrangement 100 times.
- the identification accuracy value There will be one piece of data that matches the pattern information data to be judged in the 100 pieces of pattern information that will be realized. Therefore, when the identification accuracy value is 0.01, it can be distinguished from about 100 pieces of data. Similarly, when the identification accuracy value is 0.0.001, it can be identified as about 100.000 items. Thus, the reciprocal of the identification accuracy value gives the degree of the number of data that can be identified. In other words, the identification accuracy value can be considered to indicate the identification capability of the pattern information data.
- the identification accuracy value described in this embodiment is a unique concept, and can be applied in combination with other methods as follows.
- the identification accuracy value can also be determined by other empirical methods.
- the other method is an empirical method using test data. Therefore, if the data of the method described in this embodiment is compared with the data of other empirical methods in advance, the same result as the other empirical methods will be obtained next time, and the pattern information display of the judgment target This can be achieved simply by entering the evening.
- the data of the method described in this embodiment with the data of other empirical methods, it is possible to match the other data by adjusting the radius R in Equation 1. Easy.
- taking advantage of the ability to calculate the identification accuracy value simply by inputting the pattern information data to be judged it can be applied as a standardization means for determining the criteria for the identification accuracy of pattern information data.
- the feature points are the end points and the branch points of the fingerprint.
- the present embodiment can be similarly applied to other feature points.
- the reason is to determine whether the identification accuracy of the pattern information data is good or not by calculating the identification accuracy value of the pattern information data.
- Another advantage is that pattern information data suitable for pattern matching can be selected regardless of the amount of pattern information data (for example, the number of feature points).
- the reason is that the identification accuracy value is calculated according to the amount of pattern information data.
- FIG. 11 is a block diagram showing the configuration of the pattern information registration apparatus 15 according to the present embodiment.
- the pattern information registration device 15 includes a common feature extraction unit 3 in the configuration of the data processing unit 30 of the pattern information registration device 10 according to the first embodiment of the present invention. 5 is added.
- the input unit 20 inputs the pattern information a plurality of times, and provides the input plurality of pattern information data and its feature data to the data processing unit 30.
- the common feature extraction means 35 can extract features included in common in a plurality of pattern information data captured by the input unit 20.
- the common feature extraction means 35 selects two pattern information data having a high commonality of information when the features included in all of the inputted plurality of pattern information data are extracted.
- the number of input pattern information data is two, two pattern information data are selected.
- the number of input data is 3, two or three pattern information data with high information commonness among the three data are selected.
- the number of pattern information data to be selected may be three or more when the number of input data is three or more.
- the two pattern information items selected in this way are the two pattern information items.
- the discrimination accuracy value calculation means 31 calculates the discrimination accuracy value.
- FIG. 12 is a flowchart for explaining the operation of the pattern information registration apparatus 15 according to the present embodiment.
- ⁇ Refer to the main part of Fig. 11 as necessary.
- pattern information data is input by the input means 21 (step 1 0 0 1), and the feature quantity is extracted by the feature quantity extraction means 2 2 (step 1 0 0 2).
- the common feature extraction means 35 extracts the common feature amount included in the plurality of pattern information data and provides it to the identification accuracy value calculation means 31 (step 1003). ⁇
- the identification accuracy value calculation means 31 calculates the identification accuracy value of the pattern information data using the common feature amount extracted by the common feature extraction means 35 (step 1004).
- the pass / fail judgment means 32 determines that the pattern information data is good when the discrimination accuracy value calculated by the discrimination accuracy value calculation means 31 is within the threshold value, and the pattern information data is not good when the threshold value is exceeded. (Step 1 0 0 5).
- the output unit 40 When it is determined that the pattern information data is good (step 100 0 6), the output unit 40 outputs a determination result indicating that the pattern information data is good (step 100 0 7).
- the registration unit 50 performs pattern information data registration processing (step 1 0 0 8), and data necessary for registration is stored in the database 50.
- the output unit 40 outputs a determination result indicating that the pattern information data is not good (step 1 0 0 9).
- the common feature extraction means 35 selects a plurality of highly common data from the inputted plurality of pattern information items, and extracts the feature values of the plurality of data. This makes it possible to reduce the influence of data variation, and to judge the quality more accurately than the case of using the feature value of one pattern information data.
- fingerprint data is input twice.
- FIG. 13 is a flowchart for explaining an operation example of the pattern information registration apparatus 15 according to the present embodiment. If necessary, refer to the main part of Fig. 11.
- the input unit 20 inputs the fingerprint data twice (step 1 1 0 1), and the data processing unit 3 Provides feature quantity to 0.
- the data processing unit 30 compares the two fingerprint data with the common feature extraction means 35 (step 1 1 0 2).
- FIG. 14 is a diagram showing fingerprint data input for the first time according to the present embodiment. It can be seen that there is a fifth feature point 1 3 1 in the fingerprint area 1 0 2.
- FIG. 15 is a diagram showing fingerprint data input a second time according to the present embodiment. It can be seen that there are five feature points 1 3 2 in the fingerprint area 1 0 3.
- FIG. 16 is a diagram showing an example in which common features are extracted from two fingerprint data by the common feature extraction means 35 of the present embodiment.
- the fingerprint area 1 0 2 of the fingerprint data input the first time and the fingerprint area 1 0 3 of the fingerprint data input the second time are the common areas (the following is the common area) (Abbreviated as 1 0 5).
- the common area 105 four common feature points (hereinafter abbreviated as common feature points) are arranged.
- the common feature extraction means 35 extracts the common area 1 0 5 and the common feature point 1 3 5 of the two fingerprint data (step 1 1 0 3).
- FIG. 17 is a diagram showing a common area 10 5 and common feature points 1 3 5 of two fingerprint data extracted by the common feature extraction means 35 according to this embodiment.
- the identification accuracy value calculation means 31 calculates the identification accuracy value of the fingerprint data composed of the common area 10 5 of the two fingerprint data and the common feature point 1 35 (step 110 4).
- the discrimination accuracy value calculated by the discrimination accuracy value calculation means 31 is compared with a predetermined threshold value (step 1105).
- the identification accuracy value calculated by the identification accuracy value calculation means 3 1 is smaller than the threshold, that is, the identification accuracy value is smaller than the threshold. It is determined that the fingerprint data is highly accurate and good (step 1 1 0 7).
- the output unit 40 For the fingerprint data determined to have high identification accuracy, the output unit 40 outputs a determination result indicating that the identification data is good (step 1 1 0 8).
- Step 1 1 1 0 9 registration processing is performed in the registration unit 50 (step 1 1 0 9).
- the pass / fail judgment means 3 2 determines that the fingerprint data is not good and has low identification accuracy.
- the output unit 40 For fingerprint data determined to have low identification accuracy, the output unit 40 outputs a determination result indicating that the identification accuracy is not good (step 1 1 1 1).
- fingerprint data is input twice, but the number of times of input may be three or more.
- the common feature extraction means 3 5 selects a combination of two highly common pattern information data from a plurality of fingerprint data, and calculates an identification accuracy value. Even if it is judged that the fingerprint data only by two times is not good by the pass / fail judgment means 32, the fingerprint data may be judged to be good after inputting three times or more. For this reason, it is possible to calculate a highly reliable identification accuracy value by sequentially inputting until fingerprint data is determined to be good.
- the identification accuracy value can be accurately calculated.
- the reason is that by selecting a combination of a plurality of pattern information data having high commonality, it is possible to reduce the influence of data variation.
- the feature point positions are used to explain whether the feature points match or do not match, but other feature quantities such as the direction of the ridges that contact the feature points are added. It is also possible to calculate the identification accuracy in more detail.
- the fingerprint pattern is used as the pattern information, but a palm pattern may be used. It is also possible to use other pattern information that can identify an individual, such as iris, face, palm shape, and vein pattern.
- a specific part for example, the edge of an eye or the edge of a lip can be used as a feature point.
- FIG. 18 is a diagram showing an example of feature points when a face is used as pattern information according to the present embodiment. As shown in the figure, the positions of the edges of the eyes and lips are evenly distributed throughout the face. However, it is not possible to calculate the identification accuracy value by obtaining the probability of matching with arbitrary data when the positions of feature points are randomly arranged like fingerprint data.
- FIG. 19 is a diagram illustrating an example of feature points when the position of the end of the eye is used as pattern information according to the present embodiment.
- the point of the edge part of a figure has shown the distribution (it abbreviates as appearance distribution below) of the position where the edge part of an eye appears.
- This appearance distribution (appearance frequency) may be based on past data, or may be based on the probability (occurrence probability) that data is expected to occur.
- the position where the edge of the eye appears is not uniform throughout the face. If the distribution of the feature point positions is not uniform, as shown in Fig. 19, it is assumed that the edge of the eye is placed according to the frequency of the position where the specific part appears. What is necessary is just to calculate the probability that the data matches any pattern information data.
- FIG. 20 is a block diagram showing the configuration of the pattern matching system 70 according to this embodiment. In the following description, reference will be made to the main parts of Figure 1 as necessary.
- the pattern matching system 70 includes a pattern information registration device 10, a database 60, and a pattern matching device 80.
- the pattern information registration device 10 and the pattern matching device 80 are connected via a data pace 60.
- the configuration of the pattern information registration device 10 is the same as that shown in FIG. However, the database
- the pattern matching device 80 stores therein a plurality of registered pattern information data for matching, and when the pattern information data to be matched is input, the pattern matching device 80 Specify which of the multiple pieces of pattern information data is stored in the information data.
- the pattern information registration device 10 determines whether the pattern information data input for registration is good or bad, and registers the pattern information data that is determined to be good with high identification accuracy. And store the registered data (hereinafter abbreviated as “registered data”) in database 60.
- the pattern matching device 80 reads the registered data stored in the database 60 and stores it in the pattern matching device 80 as matching data.
- the pattern matching device 80 uses the matching data to specify which of the plurality of pattern information data stored therein is the pattern information input to the pattern matching device 80.
- the fingerprint matching device 80 When the pattern matching device 80 is a fingerprint matching device, when fingerprint data is input, the fingerprint matching device checks the fingerprint data stored in the device, identifies the data corresponding to the fingerprint, and Judge whether there is.
- the pattern matching system 70 is described as one device.
- the pattern information registration device 10 and the pattern matching device 80 can be arranged at different locations.
- the database 60 may be installed inside the pattern information registration device 10 as shown in FIG. 1, and the pattern information registration device 10 and the pattern matching device 80 may be connected via a network.
- the pattern information registration device 1 0, 1 5 of the present invention its operation as a matter of course be a hardware manner realized, the pattern information registration Purodara beam (application) 3 0 0 to actual fi 1 each means described above
- the pattern information registration devices 10 and 15 which are computer processing devices, it can be realized as software.
- the pattern information registration program 300 is stored in a magnetic disk, a semiconductor memory, or other recording medium, and is read from the recording medium to the pattern information registration devices 10 and 15, and by controlling its operation, Each function described above is realized.
- the present invention it is possible to select pattern information suitable for pattern matching.
- the reason is to determine whether the identification accuracy of the pattern information data is good or bad by calculating the identification accuracy value of the pattern information data.
- the reason is that the identification accuracy value is calculated according to the amount of pattern information data.
Landscapes
- Engineering & Computer Science (AREA)
- Quality & Reliability (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Human Computer Interaction (AREA)
- Collating Specific Patterns (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
Description
Claims
Priority Applications (4)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US11/814,362 US8358814B2 (en) | 2005-01-19 | 2006-01-19 | Pattern information registration device, pattern information registration method, pattern information registration program and pattern collation system |
EP06701485A EP1840832A4 (en) | 2005-01-19 | 2006-01-19 | PRINTING INFORMATION RECORDING APPARATUS, METHOD AND PROGRAM THEREOF, FILTERING SYSTEM |
JP2006554005A JP4911300B2 (ja) | 2005-01-19 | 2006-01-19 | パターン情報登録装置、パターン情報登録方法、パターン情報登録プログラム及びパターン照合システム |
CN2006800027493A CN101107629B (zh) | 2005-01-19 | 2006-01-19 | 图样信息注册装置、图样信息注册方法和图样匹配系统 |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
JP2005-011372 | 2005-01-19 | ||
JP2005011372 | 2005-01-19 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2006078054A1 true WO2006078054A1 (ja) | 2006-07-27 |
Family
ID=36692420
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/JP2006/301164 WO2006078054A1 (ja) | 2005-01-19 | 2006-01-19 | パターン情報登録装置、パターン情報登録方法、パターン情報登録プログラム及びパターン照合システム |
Country Status (5)
Country | Link |
---|---|
US (1) | US8358814B2 (ja) |
EP (1) | EP1840832A4 (ja) |
JP (1) | JP4911300B2 (ja) |
CN (1) | CN101107629B (ja) |
WO (1) | WO2006078054A1 (ja) |
Families Citing this family (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8638815B2 (en) * | 2010-01-08 | 2014-01-28 | Blackberry Limited | Method and apparatus for logical channel prioritization for uplink carrier aggregation |
US8437517B2 (en) * | 2010-11-03 | 2013-05-07 | Lockheed Martin Corporation | Latent fingerprint detectors and fingerprint scanners therefrom |
JP5954712B2 (ja) * | 2011-01-13 | 2016-07-20 | パナソニックIpマネジメント株式会社 | 画像処理装置、画像処理方法、及びそのプログラム |
US20150241350A1 (en) | 2011-08-26 | 2015-08-27 | Edward J. Miesak | Latent fingerprint detection |
TW201342253A (zh) * | 2012-04-13 | 2013-10-16 | Hon Hai Prec Ind Co Ltd | 用於電子裝置中的指紋驗證方法及系統 |
US9400914B2 (en) * | 2014-10-24 | 2016-07-26 | Egis Technology Inc. | Method and electronic device for generating fingerprint enrollment data |
US9804096B1 (en) | 2015-01-14 | 2017-10-31 | Leidos Innovations Technology, Inc. | System and method for detecting latent images on a thermal dye printer film |
CN105447454B (zh) | 2015-11-13 | 2018-05-01 | 广东欧珀移动通信有限公司 | 指纹模板完善方法、装置和终端设备 |
CN105426835B (zh) * | 2015-11-13 | 2019-03-05 | Oppo广东移动通信有限公司 | 指纹注册方法、装置及移动终端 |
CN106682618A (zh) * | 2016-12-27 | 2017-05-17 | 努比亚技术有限公司 | 指纹识别方法及移动终端 |
CN107122116A (zh) * | 2017-04-28 | 2017-09-01 | 广东欧珀移动通信有限公司 | 指纹识别区域显示方法及相关产品 |
US10789449B2 (en) * | 2018-01-25 | 2020-09-29 | Egis Technology Inc. | Electronic device for distinguishing between fingerprint feature points and non-fingerprint feature points and method for the same |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH06149980A (ja) * | 1992-11-09 | 1994-05-31 | Fujitsu Ltd | 指紋辞書登録処理方式 |
JPH0991434A (ja) * | 1995-09-28 | 1997-04-04 | Hamamatsu Photonics Kk | 人物照合装置 |
JP2002288667A (ja) * | 2001-03-28 | 2002-10-04 | Nec Corp | パターン照合装置とそのパターン照合方法、及びパターン照合プログラム |
Family Cites Families (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPS634381A (ja) | 1986-06-24 | 1988-01-09 | Mitsubishi Electric Corp | 指紋照合装置 |
JPH01131978A (ja) | 1987-08-26 | 1989-05-24 | Komatsu Ltd | 指紋の同一性判定方法および同一性判定装置 |
JP2710434B2 (ja) | 1990-01-24 | 1998-02-10 | 日本電気セキュリティシステム株式会社 | 特徴抽出装置 |
JP2990491B2 (ja) | 1995-03-20 | 1999-12-13 | 富士通電装株式会社 | 指紋登録方法及び指紋照合装置 |
US5933516A (en) * | 1997-08-07 | 1999-08-03 | Lockheed Martin Corp. | Fingerprint matching by estimation of a maximum clique |
JP3858391B2 (ja) | 1997-11-20 | 2006-12-13 | ソニー株式会社 | 画像照合装置及び画像照合方法 |
JP4055974B2 (ja) | 1999-05-26 | 2008-03-05 | 株式会社リコー | パターン辞書作成方法及び装置、並びに記録媒体 |
JP4177598B2 (ja) * | 2001-05-25 | 2008-11-05 | 株式会社東芝 | 顔画像記録装置、情報管理システム、顔画像記録方法、及び情報管理方法 |
TWI299471B (en) * | 2001-08-24 | 2008-08-01 | Toshiba Kk | Person recognition apparatus |
-
2006
- 2006-01-19 EP EP06701485A patent/EP1840832A4/en not_active Withdrawn
- 2006-01-19 CN CN2006800027493A patent/CN101107629B/zh active Active
- 2006-01-19 US US11/814,362 patent/US8358814B2/en active Active
- 2006-01-19 JP JP2006554005A patent/JP4911300B2/ja active Active
- 2006-01-19 WO PCT/JP2006/301164 patent/WO2006078054A1/ja active Application Filing
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH06149980A (ja) * | 1992-11-09 | 1994-05-31 | Fujitsu Ltd | 指紋辞書登録処理方式 |
JPH0991434A (ja) * | 1995-09-28 | 1997-04-04 | Hamamatsu Photonics Kk | 人物照合装置 |
JP2002288667A (ja) * | 2001-03-28 | 2002-10-04 | Nec Corp | パターン照合装置とそのパターン照合方法、及びパターン照合プログラム |
Non-Patent Citations (1)
Title |
---|
See also references of EP1840832A4 * |
Also Published As
Publication number | Publication date |
---|---|
US8358814B2 (en) | 2013-01-22 |
US20090052752A1 (en) | 2009-02-26 |
CN101107629A (zh) | 2008-01-16 |
EP1840832A4 (en) | 2013-02-06 |
JPWO2006078054A1 (ja) | 2008-08-07 |
EP1840832A1 (en) | 2007-10-03 |
CN101107629B (zh) | 2011-07-27 |
JP4911300B2 (ja) | 2012-04-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2006078054A1 (ja) | パターン情報登録装置、パターン情報登録方法、パターン情報登録プログラム及びパターン照合システム | |
Jain et al. | A multimodal biometric system using fingerprint, face and speech | |
US6961452B2 (en) | Fingerprint data synthesis method, fingerprint data synthesis apparatus and computer-readable recording medium on which fingerprint data synthesis program is recorded as well as biometric information synthesis method | |
JP2821348B2 (ja) | 指紋照合装置 | |
US7257241B2 (en) | Dynamic thresholding for a fingerprint matching system | |
CN100414558C (zh) | 基于模板学习的自动指纹识别系统和方法 | |
JP5531963B2 (ja) | 認証システム、装置、認証方法及びプログラム | |
EP2523149A2 (en) | A method and system for association and decision fusion of multimodal inputs | |
Azzini et al. | A fuzzy approach to multimodal biometric continuous authentication | |
US20080013803A1 (en) | Method and apparatus for determining print image quality | |
US11062120B2 (en) | High speed reference point independent database filtering for fingerprint identification | |
JP2010061528A (ja) | 生体認証装置、生体認証プログラム及び生体認証方法 | |
US20080273767A1 (en) | Iterative print matching method and system | |
WO2008054940A2 (en) | Print matching method and apparatus using pseudo-ridges | |
Jain et al. | Biometric template selection: a case study in fingerprints | |
JP2006085268A (ja) | 生体認証システムおよび生体認証方法 | |
TWI325568B (en) | A method for face varification | |
CN114519898A (zh) | 生物特征多模态融合识别方法、装置、存储介质及设备 | |
CN114926865A (zh) | 指纹识别方法、装置、电子设备及存储介质 | |
JP2866461B2 (ja) | 指紋照合装置 | |
JP3425092B2 (ja) | パターン認識方式 | |
CN112001233A (zh) | 生物特征的辨识系统及辨识方法 | |
CN115668311A (zh) | 认证方法、认证程序以及信息处理装置 | |
Shende et al. | Soft computing approach for feature extraction of palm biometric | |
Qureshi et al. | Effectiveness of assigning confidence levels to classifiers and a novel feature in fingerprint matching |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application | ||
DPE1 | Request for preliminary examination filed after expiration of 19th month from priority date (pct application filed from 20040101) | ||
WWE | Wipo information: entry into national phase |
Ref document number: 2006554005 Country of ref document: JP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2006701485 Country of ref document: EP |
|
WWE | Wipo information: entry into national phase |
Ref document number: 11814362 Country of ref document: US Ref document number: 200680002749.3 Country of ref document: CN |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
WWP | Wipo information: published in national office |
Ref document number: 2006701485 Country of ref document: EP |