JP2007286970A - Data collation device, data collation method, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a data collation device for achieving efficient collation by reducing probability that any unnecessary collation degree table is retrieved. <P>SOLUTION: A correlation value between a collation degree, being the matching degree between collation data and registration data, and a collation degree between registration data in a collation degree table is calculated as the retrieval priority of the collation degree table, and the collation table is selected based on the retrieval priority, and the collation degree of the registration data and collation data retrieved based on the content of the collation degree table is compared with a prescribed threshold so that the registration data to which the collation data are equivalent can be determined. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a data collating apparatus and a data collating method for collating which of a plurality of registered data items to be collated is data, and in particular extracted from image data and image data The present invention relates to a data collation apparatus, a data collation method, and a program that causes a computer to function as the apparatus.

  As a conventional data collation method for performing one-to-multiple data collation, for example, there is one disclosed in Patent Document 1. In Patent Document 1, a matching degree that is a degree of coincidence calculated by comparing registration data read from a plurality of registration data and data to be matched (collation data), and two of the plurality of registration data. A cross-correlation table (matching degree table) composed of matching degrees for all combinations to be selected is used for matching. In other words, using this matching degree table, while controlling the order of registration data to be read and checked next, the matching degree between the registration data and the matching data is compared with a threshold value, and the matching data is out of a plurality of registered data. Decide which is the case. Thereby, collation data can be searched and collated at high speed.

  Patent Document 2 classifies registered data into a plurality of groups, and prepares a matching degree table (total matching degree table) in which a matching degree that is a degree of matching between registered data in a plurality of groups is described. This collation degree table is divided into collation degree tables used for collation. Thereby, at the time of collation, it can refer in a collation degree table unit, and the memory capacity required for the reference can be reduced. Furthermore, the configuration of each matching degree table can be changed as needed.

Japanese Patent No. 3053607 JP 2002-163652 A

  In the conventional data matching methods disclosed in Patent Literature 1 and Patent Literature 2, in order to perform data matching, only a plurality of matching degree tables are selected and processed in order. For this reason, there is a high probability that a matching degree table that does not need to be searched is originally searched, and there is a problem that efficient matching cannot be performed.

  For example, when collating data related to a human having a search scale of 100,000 to 200,000, the number of classifications for each group in the collation degree table reaches about 100 to 200. However, there is only one corresponding data as a search result in the above-described 100 to 200 matching degree tables.

  The present invention has been made to solve the above-described problems, and by assigning a search priority to the collation degree table, the probability that an unnecessary collation degree table is searched is reduced and efficient. It is an object of the present invention to obtain a data collating apparatus capable of performing collation, a data collating method, and a program for causing a computer to function as the apparatus.

  A data collating apparatus according to the present invention includes a storage unit that stores a collation degree table in which a collation degree indicating a degree of coincidence between registration data in a plurality of registration data is stored, and a registration searched based on the contents of the collation degree table A collation processing unit that compares the data with the collation data to be collated and calculates a collation degree that is a degree of coincidence between them, a collation degree calculated by the collation processing unit, and a collation degree between registered data in the collation degree table Is calculated by a control unit that calculates the correlation value of the search as a search priority of the matching degree table, a selection unit that selects a matching degree table to be searched for registered data based on the search priority from the storage unit, and a matching processing unit And a matching determination unit that compares the registered matching level with a predetermined threshold and determines which registered data the matching data corresponds to.

  According to this invention, the correlation value between the matching degree, which is the degree of matching between the matching data and the registered data, and the matching degree between the registered data in the matching degree table is calculated as the search priority of the matching degree table, and the matching degree table When searching for registration data that is a candidate for matching based on the contents of the data, the matching degree table is selected according to the search priority, and the matching degree of the matching data and the registered data is compared with a predetermined threshold value. Since it is determined whether the data corresponds to the registered data, it is possible to efficiently select the collation degree table in which the corresponding data will be registered, and there is an effect that collation can be performed at a higher speed than in the past. In addition, since the corresponding data is listed as a candidate at an early stage, there is an effect that the probability that data other than the corresponding data is determined to be the corresponding date becomes smaller than that in the past.

Embodiment 1 FIG.
1 is a block diagram showing a configuration of a data collating apparatus according to Embodiment 1 of the present invention. As shown in the figure, the data collating apparatus according to the first embodiment includes a collation degree table / registered data storage unit (storage unit) 1, a collation degree table selection unit (selection unit) 2, a collation processing unit 3, and a collation determination unit 4. , A registration data read control unit (control unit) 5 and a priority data storage unit 6 are provided.

  The collation degree table / registered data storage unit 1 collates a plurality of registration data to be compared with data to be collated (collation data) and a collation degree between the registration data in each group classified into a plurality of groups. The degree table is stored. Here, the collation degree is a value indicating the degree of matching between two data, and for example, a correlation value between the data is used.

  The collation degree table includes combination information of the collation degrees between the registration data calculated for all combinations in which two are selected from the registration data belonging to each group and the registration data obtained from the collation degrees. Moreover, the collation degree table is divided | segmented into the predetermined number, The collation degree about the combination of a predetermined number of registration data is stored in the collation degree table corresponding to each divided | segmented part.

Here, an example is given of collation degree tables X ₁ , X ₂ ,..., X _d divided into d pieces. In the k-th collation degree table X _k (k = 1, 2,..., D), for example, registration data E _i defined on the i-row side and the j-column side in the i-row / j-column matrix. The degree of matching with the defined registration data E _j is stored as an element of the matrix.

  The collation degree table selection unit 2 searches for one from the collation degree table stored in the collation degree table / registered data storage unit 1 based on the search priority of each collation degree table read from the priority data storage unit 6. Select as target. The collation degree table selection unit 2 reads out the collation candidate registration data from the collation degree table / registration data storage unit 1 based on the registration data number input from the registration data read control unit 5 and outputs the data to the collation processing unit 3. To do.

  The collation processing unit 3 calculates a collation degree indicating the degree of coincidence between the collation candidate registration data input via the collation degree table selecting unit 2 and the input unknown collation data, and outputs the collation degree to the collation determining unit 4. . The collation determination unit 4 compares the collation degree calculated by the collation processing unit 3 with a predetermined threshold value and determines whether or not the inputted unknown collation data corresponds to the registered data. For example, if the threshold value is equal to or greater than a predetermined threshold, a verification result indicating that the verification data corresponds to the registered data is output. If the threshold value is less than the threshold value, a verification result indicating that there is no registration data corresponding to the verification data is output. .

  The registered data reading control unit 5 uses the matching degree calculated by the matching processing unit 3 and the matching degree table selected by the matching degree table selecting unit 2 to use the matching degree data when the registered data thus verified does not correspond to the matching data. The order of registration data to be read out is controlled. For example, a search number counter (not shown) is provided, and the number of search processes for registered data corresponding to collation data is controlled according to the count value of the search number counter.

  The registered data reading control unit 5 is selected by the matching degree table selection unit 2 and the matching degree of each registered data and matching data calculated up to the current matching by the matching processing unit 3 in the matching process by the matching determination unit 4. The partial correlation value is calculated using the matching degree in the matching degree table, and the maximum partial correlation value until the current matching is output to the priority data storage unit 6 as the search priority of the matching degree table. Further, the priority data storage unit 6 stores the partial correlation value calculated by the registered data read control unit 5 as the search priority of the collation table.

  The collation degree table selection unit 2, the collation processing unit 3, the collation determining unit 4 and the registered data reading control unit 5 described above are, for example, a data collation processing program according to the gist of the present invention and an arithmetic processing of a computer constituting the data collation device. By causing the apparatus to execute and control the operation, it can be realized as a specific means in which software and hardware cooperate.

  The collation degree table / registration data storage unit 1 and the priority data storage unit 6 are, for example, a hard disk device mounted on the computer, an external storage medium that can be played back by a drive device, an external storage device that can perform data communication, and the like. Can be configured on the storage area.

Next, the operation will be described.
FIG. 2 is a flowchart showing a flow of operations performed by the data collating apparatus according to the first embodiment, and shows a series of flows of search processing and collation processing. Thereafter, the collation degree table is divided into d pieces of collation degree tables X ₁ , X ₂ ,..., X _d , and the registered data E is k-th (k = 1, 2,..., D) collation. An example in which the degree table X _k is divided into n _k pieces will be described.
First, when the unknown collation data V _u is input to the collation processing unit 3, the registered data reading control unit 5 sets the current search times m _k = 0 and the search priority p (k) = 0 in the search number counter. Initialization is performed (step ST1).

Next, the registered data reading control unit 5 sets j ^k (k = 1, j) as the registered data number r _k (m _k ) of the current m _k (first) matching candidate in the matching degree table X _{k to} be read. = 1) is set, and the registration data number is output to the collation degree table selection unit 2 (step ST2). Here, j ^k is the registered data number of the j th column in the k th collation degree table X _k .

The collation degree table selection unit 2 searches the collation degree table / registered data storage unit 1 based on the registration data number r _k (m _k ) (= 1) of the collation candidate input from the registered data read control unit 5, The corresponding registration data E _{rk (mk)} in the k-th matching degree table X _k is read out and output to the matching processing unit 3.

Collation processing section 3, a verification data V _u inputted, matching degree y _u of the current matching candidates registered data E _{rk (mk), rk} the _(mk), calculated on the basis of the following equation (1) ( Step ST3). In the following formula (1), f (A, B) is a collation function for obtaining the collation degree between data A and data B. The collation degree _{yu, rk (mk)} of the calculation result is output to the collation determining unit 4.

The collation determining unit 4 compares the collation degree _{yu, rk (mk)} input from the collation processing unit 3 with a preset threshold T relating to the collation degree, and outputs a collation result according to the comparison result (step ST4). ). Here, if the matching degree _{yu, rk (mk)} is equal to or greater than the threshold value T, the process proceeds to step ST11, where it is determined that the current matching candidate registration data E _{rk (mk)} corresponds to the matching data V _u. Is output as a verification result.

For example, if the user identified by the temporary registration data number u inputs the collation data V _u, registered matching candidate data E _{rk (mk)} is determined to correspond to the collation data V _u, registration of the user A collation result that the data number u is the registered data number r _k (m _k ) of the collation candidate is obtained.

On the other hand, matching degree y _u, when _{rk (mk)} is less than the threshold T, the matching determination unit 4 determines that the registered data E _rk current matching candidate _(mk) does not correspond to the verification data V _u, the A message to the effect is output to the registered data read control unit 5, and the process proceeds to step ST5.

In step ST5, upon receiving a notification that the current collation candidate registration data E _{rk (mk)} does not correspond to the collation data V _u , the registration data reading control unit 5 accumulates the search count m _k of the search count counter. It is determined whether m (m = Σ _k m _k ) has reached a preset number D of search terminations. Here, if the search abort count D has been reached, it is determined that there is no registered data corresponding to the current collation data V _u , and the search processing is terminated.

On the other hand, if the search abort count D has not been reached, the registration data read control unit 5 proceeds to step ST6 to determine registration data for the next collation candidate.
First, the registration data reading control unit 5 inputs the collation degree up to the current m _k -th collation for the registration data of the current collation candidate from the collation determination unit 4, and the current data represented by the following equation (2). A matching degree vector _yu ^k (m _k ) composed of the matching degrees up to the m _kth matching is calculated. However, k = 1,2, ···, is _{d, y u, rk (1} ) is a matching degree of the search first collation data V _u and the registration data E _{rk (1),} y _{u , rk (mk)} is the matching degree between the matching data V _u for the search m _k and the registered data E _{rk (mk)} .

Further, the registration data read control unit 5 excludes i rows in which the registration data E _{rk (mk)} of the current collation candidate is defined in the collation degree table X _k (k = 1, 2,..., D). The matching degree up to the current m _kth matching corresponding to the other i-th row and j = r _k (l) -th column component (l = 1, 2,..., M _k ) A partial vector x _i ^k (m _k ) is input from the matching degree table / registered data storage unit 1 via the selection unit 2 and expressed by the following equation (3).

Next, the registered data reading control unit 5 uses the matching degree vector _yu ^k (m _k ) and the partial vector x _i ^k (m _k ), and among the matching degrees of the i-th row in the matching degree table X _k , The degree of matching for all registered data i excluding registered data that has already been checked for collation is extracted, and the correlation value of the degree of matching for both vectors is calculated as a partial correlation value z _i ^k (m _k ) according to the following equation (4). To do. Thus, partial correlation values z _i ^k (1), z _i ^k (2),..., Z _i ^k (m _k ) are obtained by the current m _k -th collation.

In step ST7, the registered data read control unit 5 compares the partial correlation values z _i ^k (m _k ) calculated as described above up to the current m _k -th collation with each other, and among them, the maximum partial correlation is obtained. The value is stored in the priority data storage unit 6 as the search priority p (m _k ) of the matching degree table X _k . In addition, i row of the registered data from which the maximum partial correlation value in the matching degree table X _k is obtained is set to i _max .

Subsequently, the registration data read control unit 5 updates the search count of the search count counter to m _k = m _k +1 (step ST8), and sets the registration data number r _k (m _k ) of the next collation candidate to i _max . It is set and output to the matching degree table selection unit 2 (step ST9).

In step ST10, the matching degree table selection unit 2 uses the matching candidate registration data number r _k (m _k ) (= i _max ) input from the registered data read control unit 5 to check the matching degree table / registered data storage unit. 1 is retrieved, and the corresponding registration data E _{rk (mk)} is read and output to the verification processing unit 3. The verification processing unit 3 executes the processes after step ST3.

At this time, if all the matching degree tables X _k are the current search times m _k , the matching degree table selection unit 2 has the highest matching priority p (k) stored in the priority data storage unit 6. The degree table X _k is selected as a search target, and the registered data defined in the i _max row is output to the matching processing unit 3 as the next matching candidate. Thereafter, the process proceeds to step ST3, where the collation processing unit 3 executes collation processing.

On the other hand, if there is a collation degree table X _k that has not reached the current retrieval time m _k , k is updated (k = k + 1) as the next retrieval object, and the collation degree table corresponding to the updated k. the registration data in X _k, and outputs the collation processing section 3 as the next matching candidate. Thereafter, the process proceeds to step ST3, and the above-described collation process is executed by the collation processing unit 3.

As described above, according to the first embodiment, when selecting registered data to be collated with the input unknown collation data, a search is performed once for all collation degree tables X _k , and each collation degree table is searched. Search priority sought p (k) of X _k, the search priority p in the future matching process (k) is selected as the search target of the registration data matching degree table k is the maximum. Further, the search priority p (k) is updated according to the result of each matching process. By doing this, the search priority of the matching degree table is updated at any time during the search for the registered data of the matching candidate. Therefore, the matching degree table having a high probability that the registered data corresponding to the matching data exists. You can select at any time.

Embodiment 2. FIG.
It is a block diagram which shows the structure of the data collation apparatus by Embodiment 2 of this invention, and attaches | subjects the same code | symbol to the component which is the same as that of FIG. 1, or is equivalent to this. In the data collating apparatus according to the second embodiment, a collation degree table priority calculation unit (priority calculation unit) 7 and a search history storage unit 8 are newly provided in the configuration of FIG. 1 shown in the first embodiment. Yes. As shown in the first embodiment, the registered data read control unit 5 does not calculate the partial correlation value as the search priority of the collation degree table, and the collation degree table priority calculation unit 7 performs the search priority of the collation degree table. Calculate the degree.

  That is, in the collation degree table priority calculation unit 7, when the collation determination unit 4 determines that the registration data corresponds to the collation data, the registration data number of the corresponding registration data and the collation degree table / registration data storage Using the collation degree table read from the unit 1 and the search history of the registered data stored in the search history storage unit 8, the search priority of the collation degree table in which the corresponding data is registered is calculated.

Further, in the search history storage unit 8, the search history calculated during the search priority calculation by the matching degree table priority calculating unit 7 (how many times the registered data registered in each matching degree table becomes the corresponding data). Taka) is memorized. Hereinafter, among the data registered in each of the collation degree tables X _k (k = 1, 2,..., D) divided into d pieces constituting the collation degree table, the corresponding data is identified by the collation determining unit 4. It is assumed that the number of times determined to be is stored as the search history h (k).

  In addition, the collation degree table priority calculation part 7 mentioned above also makes the arithmetic processing unit of the computer which comprises a data collation apparatus execute the data collation processing program according to the meaning of this invention similarly to the said Embodiment 1, for example. By controlling the operation, it can be realized as a specific means in which software and hardware cooperate. The search history storage unit 8 can be configured on a storage area such as a hard disk device mounted on the computer, an external storage medium that can be played back by a drive device, or an external storage device that can perform data communication.

Next, the operation will be described.
FIG. 4 is a flowchart showing the flow of operations performed by the data collating apparatus according to the second embodiment, and the priority calculation process performed by the collation degree table priority calculating unit 7 will be described with reference to this figure.
First, when the collation determination unit 4 obtains a collation result that the collation candidate registration data is the corresponding data, the collation degree table priority calculation unit 7 obtains the registered data number r _k ( m _k ) is read (step ST1a).

Next, the collation degree table priority calculation unit 7 reads out all the collation degree tables divided from the collation degree table / registered data storage unit 1 (step ST3a), and the registered data number r _k obtained from the collation determination unit 4. Based on (m _k ), a matching degree table X _k for registering data of the registered data number is searched (step ST3a).

Subsequently, the collation degree table priority calculation unit 7 reads the search history h (k) for the collation table X _k searched in step ST3a from the search history storage unit 8 (step ST4a), and the collation degree table X _k. The search history h (k) is updated (step ST5a). Specifically, the search history h (k) is the number of times that the data registered in the matching degree table X _k (k = 1, 2,..., D) is determined to be the corresponding data. The number of times is updated as h (k) = h (k) +1.

Thereafter, the collation degree table priority calculation unit 7 reads the search history h (k) of all the collation degree tables X _k constituting the collation degree table, and calculates the search priority of each collation degree table (step ST6a). ). For example, the matching degree table priority calculation unit 7 calculates the sum sum of the search histories h (k) of all the matching degree tables X _k including the search history updated in step ST5a according to the following formula (5). The search priority p (k) of each matching degree table _Xk is calculated by (6). In other words, in the second embodiment, the probability that the registered data of any collation degree table can be the corresponding data for all the collation degree tables is set as the search priority.
sum = Σ _k h (k) (5)
p (k) = h (k) / sum (6)

Updating the search priority p (k) of the matching degree table X _k in step ST6a, matching degree table priority calculation unit 7, the updated search history h (k) and the search priority of each matching degree table X _k The degree p (k) is stored in the search history storage unit 8 (step ST7a). Using the search priority p (k) created in this way, the matching degree table selection unit 2 selects a matching degree table from the matching degree table / registered data storage unit 1.

Next, a collation degree table selection process using the search priority p (k) by the collation degree table selection unit 2 according to the second embodiment will be described.
For the selection of the matching degree table using the priority p (k), for example, a process of selecting the matching degree table in descending order of priority and a process of selecting the matching degree table with a probability based on the priority can be considered. In the collation degree table selection unit 2 according to the second embodiment, both processes can be appropriately selected.

  Here, in the process of selecting the matching degree table in descending order of priority as in the first embodiment, the matching degree table is always selected in the order determined by the priority value. On the other hand, since the process of selecting the matching degree table with the probability based on the priority is based on the priority, the matching degree table with a high priority may be selected many times, but the priority is low. A matching degree table may also be selected stochastically.

  Therefore, in the second embodiment, even when the matching degree table is selected with the probability based on the priority, a device is devised so that the matching degree table having a higher priority is selected as a search target with priority. Yes. For example, in step ST10 in FIG. 2 described in the first embodiment, the matching degree table selection unit 2 according to the second embodiment creates a selection table as shown in FIG. 5 in descending order of the search priority p (k). Then, a random number value from 0 to 1 is generated, and a matching degree table having a priority p (k) corresponding to the random value is selected as the next search target.

  In the example shown in FIG. 5, the selection table is represented by a pie chart corresponding to the sizes of the priorities p (1) to p (4). When the random number value is in the range of 0 to p (1) in the selection table shown in FIG. 5, the matching degree table selection unit 2 selects the matching degree table having the priority p (1) as a search target. . When the random number value is within the range of values from p (1) to {p (1) + p (2)}, the matching degree table having the priority p (2) is selected as a search target.

  Similarly, if the random number value is within the range of values from {p (1) + p (2)} to {p (1) + p (2) + p (3)}, the matching degree having the priority p (3) If the table is selected as a search target and the random number value is within the range of values from {p (1) + p (2) + p (3)} to 1, the matching degree table having the priority p (4) is searched. Selected as. By doing in this way, the collation degree table with a higher priority can be selected as a search object mainly.

  As described above, according to the second embodiment, the search priority is determined according to the number of times the data registered in the matching degree table is determined as the corresponding data. Can be selected as a search target at an early stage. Therefore, it is possible to shorten the time required for the search / collation process compared to the conventional case. In addition, since the probability that the relevant data is selected as a search target at an early stage increases, the number of times data other than the relevant data is collated is reduced, and it is possible to reduce the possibility that the data that is not the relevant data is erroneously designated as the relevant data.

It is a block diagram which shows the structure of the data collation apparatus by Embodiment 1 of this invention. 3 is a flowchart showing a flow of operations performed by the data collating apparatus according to the first embodiment. It is a block diagram which shows the structure of the data collation apparatus by Embodiment 2 of this invention. 6 is a flowchart showing a flow of operations performed by the data collating apparatus according to the second embodiment. It is a figure which shows the selection table by Embodiment 2.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 Collation degree table / registration data storage part (storage part) 2 Collation degree table selection part (selection part) 3 Collation processing part 4 Collation determination part 5 Registration data read-out control part (control part), 6 Priority data Storage unit, 7 collation degree table priority calculation unit (priority calculation unit), 8 search history storage unit.

Claims (6)

A storage unit for storing a matching degree table in which a matching degree indicating a matching degree between registered data in a plurality of registered data is described;
A collation processing unit that compares the registered data searched based on the content of the collation degree table with collation data to be collated, and calculates a collation degree that is a degree of coincidence between them;
A control unit that calculates a correlation value between a matching degree calculated by the matching processing unit and a matching degree between registered data in the matching degree table as a search priority of the matching degree table;
A selection unit that selects, from the storage unit, a matching degree table to be searched for registered data based on the search priority;
A data collation apparatus comprising: a collation determination unit that compares the collation degree calculated by the collation processing unit with a predetermined threshold value and determines which registered data the collation data corresponds to. A storage unit for storing a matching degree table in which a matching degree indicating a matching degree between registered data in a plurality of registered data is described;
A collation processing unit that compares the registered data searched based on the content of the collation degree table with collation data to be collated, and calculates a collation degree that is a degree of coincidence between them;
A collation determination unit that compares the collation degree calculated by the collation processing unit with a predetermined threshold and determines which registered data the collation data corresponds to;
A priority calculation unit that calculates a search priority of the matching degree table used for searching the registration data based on the number of times that the matching determination unit determines that the registration data corresponds to the matching data;
A data collating apparatus comprising: a selection unit that selects, from the storage unit, a collation degree table to be searched for registered data based on the search priority. A storage unit for storing a matching degree table in which a matching degree indicating a matching degree between registered data in a plurality of registered data is stored, and which of the plurality of registered data corresponds to the matching data to be matched In the data verification method of the data verification device to be determined,
Calculating a correlation value between a matching degree that is a matching degree between the matching data and the registered data and a matching degree between registered data in the matching degree table as a search priority of the matching degree table;
Selecting the matching degree table based on the search priority;
Comparing the registration data searched based on the contents of the matching degree table with the matching degree of the matching data with a predetermined threshold to determine which registered data the matching data corresponds to. A data matching method characterized by that. A storage unit for storing a matching degree table in which a matching degree indicating a matching degree between registered data in a plurality of registered data is stored, and which of the plurality of registered data corresponds to the matching data to be matched In the data verification method of the data verification device to be determined,
Calculating the search priority of the matching degree table used for searching the registered data based on the number of times that the registered data is determined to correspond to the matching data;
Selecting the matching degree table from the storage unit according to the search priority;
Comparing the registration data searched based on the contents of the matching degree table with the matching degree of the matching data with a predetermined threshold to determine which registered data the matching data corresponds to. A data matching method characterized by that. A storage unit for storing a matching degree table in which a matching degree indicating a matching degree between registered data in a plurality of registered data is described;
A collation processing unit that compares the registered data retrieved based on the content of the collation degree table with the collation data to be collated and calculates a collation degree that is a degree of coincidence between them;
A control unit that calculates a correlation value between a matching degree calculated by the matching processing unit and a matching degree between registered data in the matching degree table as a search priority of the matching degree table;
A selection unit that selects, from the storage unit, a matching degree table to be searched for registered data based on the search priority;
A program that causes a computer to function as a collation determination unit that compares the collation degree calculated by the collation processing unit with a predetermined threshold and determines which registration data the collation data corresponds to. A storage unit for storing a matching degree table in which a matching degree indicating a matching degree between registered data in a plurality of registered data is described;
A collation processing unit that compares the registered data retrieved based on the content of the collation degree table with collation data to be collated and calculates a collation degree that is a degree of coincidence between them;
A collation determination unit that compares the collation degree calculated by the collation processing unit with a predetermined threshold and determines which registered data the collation data corresponds to;
A priority calculation unit that calculates a search priority of the matching degree table used for searching for the registered data based on the number of times that the matching determination unit determines that the registered data corresponds to the matching data;
A program that causes a computer to function as a selection unit that selects, from the storage unit, a matching degree table to be searched for registered data based on the search priority.

Images