KR20230030179A - Driving method of fingerprint high speed matching system applicable to distributed environment - Google Patents

Abstract

The present invention relates to a method for driving a high-speed fingerprint matching system applicable to a distributed environment. The method for driving a high-speed fingerprint matching system comprises: a matching request signal registration step; an MPU matching request step; a matching step; and an MPU-specific candidate list selection step. According to the present invention, the identity of a subject can be quickly confirmed.

Description

Driving method of fingerprint high speed matching system applicable to distributed environment

The present invention obtains data from a plurality of matching servers at high speed and quickly confirms the identity of a subject through a high-speed 1vsN (one to many) fingerprint matching method, enabling a large amount of fingerprint identification and distributing It relates to a driving method of a fingerprint high-speed matching system applicable to the environment.

It takes a lot of time to process a fingerprint search for a large number of fingerprints, such as millions of fingerprints or tens of millions of fingerprints. For example, if 10,000 fingerprints are searched per second for 12 million fingerprints, it takes 1200 seconds (20 minutes) to search all the fingerprints. . In addition, it takes a lot of time to obtain a large amount of fingerprint data. Obtaining data from a file or DB that stores fingerprint characteristic point data requires considerable time and memory and disk space in terms of content search time and network time, and the time to obtain data is longer than the processing time of the fingerprint matching algorithm. It is common to pay more.

Therefore, in order to confirm the identity of the subject within 60 seconds in the matching system after requesting a fingerprint search, fingerprint data for search is quickly secured and multiple fingerprints are searched on multiple servers through an algorithm that processes fingerprint search at high speed. A way to do this is desired.

Therefore, the present invention obtains data from a plurality of matching servers at high speed and identifies a subject in a short time through a high-speed 1vsN (one to many) fingerprint matching method, thereby enabling a large amount of fingerprint identification, We propose a driving method for a high-speed fingerprint matching system applicable to a distributed environment.

In the present invention, a matching processing unit (MPU, Matching Processing Unit) that performs a matching task in order to perform a fingerprint search task requested by a plurality of servers, and a matching control unit that collects and manages matching request tasks and results, and manages the MPUs ( It consists of MCU, Matching Control Unit).

As a prior art, Japanese Unexamined Patent Publication No. 2003-30661 relates to a fingerprint combination device that matches one input fingerprint with a plurality of registered fingerprints, but the matching method uses the existing method as it is, so the fingerprint matching speed is not fast.

The problem to be achieved by the present invention is to quickly obtain data from a plurality of matching servers and quickly identify a subject through a high-speed 1vsN (one to many) fingerprint matching method, thereby identifying a large amount of fingerprints. This paper relates to a driving method of a fingerprint high-speed matching system applicable to a distributed environment.

Another task to be achieved by the present invention is to acquire and manage fingerprint data at high speed for a large number of target fingerprints, and to have a structure of a Matching Control Unit (MCU) and a plurality of Matching Processing Units (MPUs) operating in a distributed environment. It is to provide a driving method of a high-speed fingerprint high-speed matching system applicable to a distributed environment, which collects matching results from a plurality of MPUs and registers the upper result from the collected results.

In order to solve the above problems, the fingerprint matching system driving method of the present invention is a terminal for identifying an identity in relation to one of police, immigration, military, finance, medical, and social welfare, from an identification detection device to the web. To the server, when an identity verification request is input together with the input fingerprint, the web server transmits matching request information including the input fingerprint to the API (application programming interface) server, and the matching request information is registered in the matching DB by the API server. Matching request signal registration step; After the matching request signal registration step, the MCU (Matching Control Unit) reads the matching request information registered in the matching DB, creates a result data set for the read matching request, and performs matching on multiple MPUs (Matching Processing Units) Sending the request information, matching request step to the MPU; Based on the received matching request information, the MPU records a bundle of requested feature point information and target feature point information in a matching queue, and the matching thread obtains the corresponding matching information from the matching queue to proceed with matching and record the result. , matching step; When the processing of the matching request is completed in the matching step, the MPU collects the matching results, sorts the results in ascending order of scores, and transmits the results as many as the preset number of candidates from the top to the MCU as a candidate list for each MPU. It is characterized in that it includes; a step of selecting a list of candidates for each MPU.

In the fingerprint matching system driving method of the present invention, the MCU receives from the MPU the candidate list for each MPU, which is the matching result generated in the step of selecting the candidate list for each MPU, records it in the matching result data set, and records it in the matching result data set from all MPUs. Upon receiving the matching result for each MPU of , a final candidate list selection step of generating a candidate list by sorting the matching results for all received MPUs and recording them in a matching DB; further includes.

In the fingerprint matching system operating method of the present invention, a web server requests a matching result from a matching DB, searches a list of candidates selected in the final candidate list selection step, and transmits the selected candidate list to an identification detection device. The step of transmitting the list of candidates; is further included.

The MCU includes a network manager in charge of network processing, a job manager that processes received requests, and a request manager that checks whether there is a new fingerprint search request and sends a matching request to MPUs. ), Result Set that collects result data and records results in DB (Database) when all results are received, Matching DAO (Data Access Object), and MPU that are in charge of DB (Database) processing It includes an MPU client (Client) that sends requests to them.

The MPU consists of a Network Manager that handles network processing, a Job Manager that handles received requests, a Matching Thread that performs matching tasks in the form of threads, and that manages all target fingerprint data. It includes a matching data set and an MCU client that transmits the result to the MCU.

In the direct management of matching system feature points, the matching server unit including MCU and MPU processes feature point data in units processed by 1vsN matching algorithm and 1vs1 matching algorithm, and creates a memory manager that processes locations in Key/Value form , After loading basic data in the memory manager, reading and processing the fingerprint data is done.

The driving method of the high-speed fingerprint high-speed matching system applicable to the distributed environment of the present invention obtains data at high speed from a plurality of matching servers and uses a high-speed 1vsN (one to many) fingerprint matching method. It verifies the identity of a person, can be applied to a distributed environment, and can identify a large number of fingerprints.

That is, the present invention is capable of obtaining and managing fingerprint data at high speed for a large number of target fingerprints, and has an algorithm for performing 1 vs. multiple fingerprint matching at high speed for a large number of target fingerprints, in a distributed environment. It is designed with a structure of an operating MCU (Matching Control Unit) and multiple MPUs (Matching Processing Units), and is configured to collect matching results from multiple MPUs and register the upper result from the collected results.

Therefore, the fingerprint high-speed matching system of the present invention quickly secures fingerprint data for search and searches for a plurality of fingerprints in a plurality of servers through an algorithm that processes fingerprint search at high speed. The identity of the subject can be confirmed within 60 seconds in the matching system.

In addition, the present invention can be applied to the development of a prototype test bed environment.

1 is an explanatory diagram for schematically explaining a driving method of a high-speed fingerprint high-speed matching system applicable to a distributed environment according to the present invention.
FIG. 2 is a configuration diagram of a matching server unit of the fingerprint high-speed matching system of FIG. 1 .
FIG. 3 is a flowchart illustrating the operation of the fingerprint high-speed matching system of FIG. 1 .

A driving method of a high-speed fingerprint high-speed matching system applicable to a distributed environment according to the present invention will be described in detail below with reference to the accompanying drawings.

<Configuration of high-speed fingerprint matching system>

1 is an explanatory diagram schematically illustrating a fingerprint high-speed matching system for identifying a large number of fingerprints according to the present invention, and FIG. 2 is a configuration diagram of a matching server unit of the fingerprint high-speed matching system of FIG. 1 .

Identity from various identification detection devices (10) When a confirmation request signal is requested to the web server (or main server or application server) 20 through the Internet network, the identification detection device 10 performs security authentication through the web server 20 to the application service unit. It is linked with the application programming interface (API) server 51 of (50), and the identification detection device 10 performs identification of a specific application (eg, a local police-related identification application) through the web server 20. A matching request signal is transmitted to the API server 51, and the API server 51 registers the received matching request signal in the matching DB of the application DB 52. The API server 51 transmits the matching request signals registered in the matching DB to the matching server unit (matching server group) 100 in order, and the matched result from the matching server unit (matching server group) 100 is returned to the API The server 51 receives it and transmits it to the identification detection device 10 through the web server 20 .

Here, the identification detection device 10 is related to a local police-related identification unit, an immigration identification unit, a military security-related identification unit, a financial service-related application unit, a medical-related patient identification unit, and a social welfare-related identification unit. It may be the person in charge's terminal (smart phone in some cases).

When a matching request signal is received through the web server 20 and the like and the matching request signal is registered in the matching DB by the API server 51, the MCU 110 automatically reads new matching request information registered in the matching DB. After generating a result data set for a matching request corresponding to and transmits a matching request to the MPU (120). Based on the received matching request information, the MPU 120 records a bundle of requested feature point information and target feature point information in a matching queue, and the matching thread obtains corresponding matching information from the matching queue to perform matching and record the result. When the processing of one matching request is completed, the MPU 120 collects the matching results, sorts the results in order of high score, and then transfers the results of the top N cases (eg, 100 cases) to the MCU 110 . When the MCU 110 receives a matching result from the MPU 120, it is recorded in a matching result data set, and when the matching result data set receives a matching result from all MPUs 120, the result is rearranged and then recorded in the matching DB.

The matching server unit 100 is composed of a matching control unit (MCU) 110 and a plurality of matching processing units (MPUs) 120 . Here, since the core function of the fingerprint high-speed matching system is performed in the matching server unit 100, the fingerprint high-speed matching system largely consists of a single matching control unit (MCU) 110 and a plurality of matching processing units (MPU, Matching Processing Unit) (120).

The MCU 110 includes a network manager in charge of network processing, a job manager for processing received requests, and a request manager that checks whether there is a new fingerprint search request and sends a matching request to MPUs. Manager), Result Set that collects result data and records results in DB (Database) when all results are received, Matching DAO (Data Access Object), MPU responsible for DB (Database) processing (120) and includes an MPU client (Client) that transmits a request.

The MPU 120 includes a network manager in charge of network processing, a job manager for processing received requests, a matching thread for performing matching tasks in the form of threads, and all target fingerprint data. It includes a matching data set (Matching Data Set) that manages, and an MCU client (Client) that transmits the result to the MCU.

FIG. 3 is a flowchart illustrating the operation of the fingerprint high-speed matching system of FIG. 1 .

In the matching request step, when a matching request signal is received through the web server 20 and the like, and the matching request signal is registered in the matching DB of the application DB 52 by the API server 51 (S110), the API server 51 ), the MCU 110 reads new matching request information registered in the matching DB (S120), generates a result data set for the corresponding matching request (S130), and then requests matching to the MPUs 120 is transmitted (S140).

In the matching step, the MPU 120 records a bundle of requested feature point information and target feature point information in a matching queue based on the received matching request information (S150), and the matching thread performs the corresponding matching information uploaded to the matching queue. (ie, requested feature point information and target feature point information) are obtained, matching is performed, and the result is recorded (S160). Here, the matching thread performs matching based on the requested characteristic points.

In the MPU-specific candidate list selection step, when processing of one matching request is completed in the matching step, the MPU 120 collects matching results and sorts the results in order of high scores (S170), , The top N cases (eg, 100 cases)) are passed to the MCU 110 as a candidate list for each MPU (S180).

In the final candidate list selection step, when the MCU 110 receives the candidate list for each MPU as a matching result from the MPU 120, it is recorded in the matching result data set, and the matching result data set is selected for each MPU in all MPUs 120. When a matching result (ie, a candidate list for each MPU) is received, a candidate list is generated by rearranging the result (S200), and recorded in the matching DB (S210).

The web server 20 requests a matching result from the matching DB, searches the candidate list selected in the final candidate list selection step (S220), transmits it to the identification detection device 10, and displays it.

<Data Fetch>

In general, as a type of data fetch method, generally, fingerprint data is stored in a DB, and the stored fingerprint data is directly obtained through a DB query, and the obtained data is stored in a file and then processed based on the file. method, how to use the obtained data as a memory cache by using an in-memory store such as Redis, and how to use the obtained data by stacking it inside the development program. The method of directly acquiring data for fingerprint matching through a DB query is the most intuitive and general method, but has a disadvantage in that it takes a considerable amount of time to fetch the DB. In addition, in the case of methods such as file, in-memory store, and program direct management, the time to fetch data through DB query is relatively short, but it is confirmed that fingerprint data stored in the DB and fingerprint data handled in the file are completely consistent. The downside is that it is difficult to do. In this case, first of all, a procedure of obtaining the first data from the DB and loading it into a file is required, and additional processing is required to match the sync with the DB data. In the case of direct management of feature points in the matching system, the feature point data is bundled and processed in units processed by the matching algorithm, a memory manager is created to process the location in the form of Key/Value, basic data is loaded into the memory manager, and fingerprint data is read and processed. configured in such a way. However, when the feature points are directly managed in this form, the size of all data of the feature points that can be read is limited according to the size of the entire physical memory. In the present invention, as a result of testing each fetch method, the general method obtained from DB fetch was too slow to be adopted, and the method using a file or in-memory store was relatively fast compared to DB, but it was difficult to achieve sufficient speed. In the matching system of the present invention, that is, in the data patch in the matching request step, a direct memory manager is basically created and feature point data is directly managed in the memory manager.

<Processing Matching Algorithm>

In the case of the existing 1vs1 matching algorithm, the accuracy is very high, but there is a disadvantage in that it takes a lot of time. In the case of the 1vsN matching algorithm, firstly, the entire data is filtered with the highest candidate group, and 1vs1 is performed for only the highest candidate group to perform matching for the same target at a relatively high speed. can happen

In the case of processing with the 1vsN matching algorithm, the highest candidate group is first created based on the index of one search request feature point and a plurality of matching target feature points (information generated by indexing the correlation inside the fingerprint feature point), and then only the top candidate group is targeted to perform a one-to-one comparison. In this way, it is possible to perform the matching operation at a much faster speed by reducing the time-consuming one-to-one comparison. As a verification of this, as a result of testing the matching algorithm of the present invention in multi-thread or multi-process processing, it was confirmed that matching proceeds faster when multi-thread or multi-process is used. In general, it is desirable to create multi-threads or multi-processes as many as the number of CPU cores of the test computer.

The results of comparing the matching speed of the system of the present invention and the conventional AFIS system are shown in Table 1.

Here, the AFIS format is the result of matching data by 1vs1 algorithm after fetching data from the DB, and the format of the present invention is the result of matching by 1vsN algorithm based on data already fetched and loaded into memory.

As a result of the comparison, it can be confirmed that the fingerprint search task was processed about 800 times faster compared to performing the fingerprint search task in the same way as AFIS. Even assuming that one MPU server handles approximately 10 million fingerprints, the result can be calculated in approximately 12.5 seconds. Even assuming various other additional time-consuming factors, the result can be produced within the targeted 60 seconds, and if a faster speed is required, further improvement is possible through parameter adjustment or tuning of the algorithm. However, in case of processing in this form, there is a disadvantage that one device cannot process more than the total amount of physical memory of the device in which the MPU operates because the fingerprint data to be searched is loaded into the actual memory and processed. About 40.55 GiB per case) However, if sufficient memory is secured or a distributed environment with a sufficiently large number is established, it can be expected to process fingerprint search at a considerable speed.

<MCU Configuration>

MCU is based on Global that is in charge of initializing and accessing major objects such as JobManager, MatchingThread, and ResultSet that handle overall MCU tasks. There is a JobManager that handles JobManager creates JobThread for the received request and performs processing of feature point addition, removal, update and matching result processing. MatchingThread periodically checks whether there is a new matching job. If there is a new matching job, MpuResulSet is created to prepare the result to be received from the MPU in advance and transfers the matching request to the MPU. When MPU delivers matching results, JobThread records the results in MpuResultSet, and when results arrive from all MPUs, the results are sorted and then records matching results in DB using Dao.

< MPU configuration >

The MPU is based on Global, which is in charge of initialization and access to major objects such as JobManager, MatchingThread, etc., which handles the entire MPU business. There is a JobManager that handles it. JobManager creates JobThread for the received request and registers matching data in MtxQueue for processing of feature point addition, removal, and update tasks and matching request so that MatchingThread handles matching. MpuDataSet has a structure that stores all target fingerprint data in memory, and MpuDataUnit manages through memory at the bottom in order to optimize the transfer speed to the matching algorithm without using additional memory due to tasks such as memory copying during 1vsN (multiple) matching. to optimize performance for matching. When a specific matching task is completed, the corresponding result is registered in the MCU using McuClient.

In the fingerprint high-speed matching system of the present invention, in the case of direct management of keypoints in the matching system, the keypoint data is bundled and processed in units processed by the matching algorithm, and a memory manager is created that processes the location in the form of Key/Value, and basic data is stored in the memory manager. is loaded, and then the fingerprint data is read and processed.

As a result of testing for each fetch method, the general method obtained from DB fetch was too slow to adopt, and the method using a file or in-memory store was relatively fast compared to DB, but it was difficult to achieve sufficient speed. Therefore, the matching system of the present invention In , basically, a direct memory manager is created and feature point data is directly managed in the memory manager.

In the present invention, in the case of processing by the 1vsN matching algorithm, the highest candidate group is first created based on the index of one search request feature point and a plurality of matching target feature points (information generated by indexing the correlation inside the fingerprint feature point), and then As a result, a one-to-one comparison was performed targeting only the top candidates.

As a result of testing the matching algorithm of the present invention in multi-thread or multi-process processing, it was confirmed that matching proceeded faster when using multi-thread or multi-process. It was confirmed that performing matching using threads is optimal.

In the present invention, the MCU is basically in charge of global handling the initialization and access of major objects such as JobManager, MatchingThread, and ResultSet, which handle the entire task of the MCU, and the MPU is responsible for the JobManager, MatchingThread, etc. Based on Global, which is in charge of initialization and access of main objects, Main is in charge of network processing in the server for initialization of the whole. In addition, because MCU and MPU use multiple languages and have complex parts in maintenance, etc., they can be developed in the form of a socket server after removing unnecessary parts as a whole, simplifying them for convenient maintenance, integrating functions into C++, and so on.

As described above, although the present invention has been described by the limited embodiments and drawings, the present invention is not limited to the above embodiments, and those skilled in the art in the field to which the present invention belongs can make various modifications and transformation is possible Therefore, the spirit of the present invention should be grasped only by the claims described below, and all equivalent or equivalent modifications thereof will be said to belong to the scope of the spirit of the present invention.

10: identification detection device 20: web server
50: application service unit 51: API server
52: application DB 53: matching server unit
110: MCU 120: MPU

Claims (6)

When an identity verification request is input along with an input fingerprint from an identity detection device, which is a terminal for identity verification in relation to one of the police, immigration, military, finance, medical, and social welfare tasks, to the web server, the web server a matching request signal registration step of transmitting matching request information including the input fingerprint to an application programming interface (API) server, and registering the matching request information in a matching DB by the API server;
After the matching request signal registration step, the MCU (Matching Control Unit) reads the matching request information registered in the matching DB, creates a result data set for the read matching request, and performs matching on multiple MPUs (Matching Processing Units) Sending the request information, matching request step to the MPU;
Based on the received matching request information, the MPU records a bundle of requested feature point information and target feature point information in a matching queue, and the matching thread obtains the corresponding matching information from the matching queue to proceed with matching and record the result. , matching step;
When the processing of the matching request is completed in the matching step, the MPU collects the matching results, sorts the results in the order of highest score, and transmits the results as many as the preset number of candidates from the top to the MCU as a candidate list for each MPU. To, MPU-specific candidate list selection step;
Characterized in that it comprises a, fingerprint matching system driving method. According to claim 1,
The MCU receives the candidate list by MPU, which is the matching result generated in the candidate list selection step by MPU, from the MPU, records it in the matching result data set, and receives matching results by MPU from all MPUs in the matching result data set. A final candidate list selection step of generating a candidate list by sorting the matching results by all MPUs and recording them in a matching DB;
Characterized in that it further comprises a, fingerprint matching system driving method. According to claim 2,
a selected candidate list transmission step in which the web server requests a matching result from a matching DB, searches the selected candidate list in the final candidate list selection step, and transmits the selected candidate list to an identification detection device;
Characterized in that it further comprises a, fingerprint matching system driving method. According to claim 3,
The MCU includes a network manager in charge of network processing, a job manager that processes received requests, and a request manager that checks whether there is a new fingerprint search request and sends a matching request to MPUs. ), Result Set that collects result data and records results in DB (Database) when all results are received, Matching DAO (Data Access Object), and MPU that are in charge of DB (Database) processing Characterized in that it comprises an MPU client (Client) for sending a request to the, fingerprint matching system driving method. According to claim 3,
The MPU consists of a Network Manager that handles network processing, a Job Manager that handles received requests, a Matching Thread that performs matching tasks in the form of threads, and that manages all target fingerprint data. A method for driving a fingerprint matching system, characterized by comprising a matching data set and an MCU client that transmits a result to an MCU. According to claim 3,
In the direct management of matching system feature points in the matching step, the MPU bundles and processes feature point data in units processed by the 1vsN matching algorithm, creates a memory manager that processes locations in the form of Key/Value, and loads basic data into the memory manager. A method for driving a fingerprint matching system, characterized in that the next fingerprint data is read and processed.

Images