CN111368120A

CN111368120A - Target fingerprint database construction method and device, electronic equipment and storage medium

Info

Publication number: CN111368120A
Application number: CN202010464713.9A
Authority: CN
Inventors: 潘鸿裕; 王克成; 黄韦维
Original assignee: Guangdong Bozhilin Robot Co Ltd
Current assignee: Guangdong Bozhilin Robot Co Ltd
Priority date: 2020-05-28
Filing date: 2020-05-28
Publication date: 2020-07-03
Anticipated expiration: 2040-05-28
Also published as: CN111368120B

Abstract

The application discloses a target fingerprint database construction method and device, electronic equipment and a storage medium. The method comprises the following steps: acquiring a first fingerprint database matched with a target space; migrating data in the first fingerprint database to a second fingerprint database through a migration learning algorithm, wherein the data in the first fingerprint database comprises data corresponding to a reference point, and the second fingerprint database is a fingerprint database corresponding to a target space; selecting part of the reference points to calculate the positioning accuracy of the second fingerprint library; and when the positioning accuracy of the second fingerprint library does not reach the preset positioning accuracy, increasing the positioning accuracy of the second fingerprint library by using the rest parts of the reference points and an enhanced learning algorithm to obtain the target fingerprint library. The target fingerprint database is quickly constructed through the transfer learning algorithm and the reinforcement learning algorithm, and meanwhile, the constructed target fingerprint database is guaranteed to have higher positioning accuracy.

Description

Target fingerprint database construction method and device, electronic equipment and storage medium

Technical Field

The present application relates to the field of positioning technologies, and in particular, to a method and an apparatus for constructing a target fingerprint database, an electronic device, and a storage medium.

Background

With the development of science and technology, more and more sensors are provided for people to collect various kinds of information. As time goes on, more and more data are collected by the sensors, and the data are more and more difficult to process, so that the fingerprint method has come up, the fingerprint method abstracts the collected data into high-dimensional vectors as individual unique fingerprints, establishes a fingerprint library, and utilizes the fingerprint library to perform matching when data processing is needed, thereby realizing rapid processing of the data.

However, the establishment of the fingerprint database requires a large amount of data to be collected, the complexity of the fingerprint database is high, a large amount of time is required to establish the fingerprint database, and it is difficult to establish the fingerprint database quickly.

Disclosure of Invention

The application provides a method and a device for constructing a target fingerprint database, an electronic device and a storage medium, so as to solve the problems.

In a first aspect, an embodiment of the present application provides a method for constructing a target fingerprint library, where the method includes: acquiring a first fingerprint database matched with a target space; migrating data in the first fingerprint database to a second fingerprint database through a migration learning algorithm, wherein the data in the first fingerprint database comprises data corresponding to a reference point, and the second fingerprint database is a fingerprint database corresponding to a target space; selecting part of the reference points to calculate the positioning accuracy of the second fingerprint library; and when the positioning accuracy of the second fingerprint library does not reach the preset positioning accuracy, increasing the positioning accuracy of the second fingerprint library by using the rest parts of the reference points and an enhanced learning algorithm to obtain the target fingerprint library.

Further, actual data of the reference points in the target space are stored in advance, and selecting a part of the reference points to calculate the positioning accuracy of the second fingerprint database includes: calculating part of measurement data of the reference point by using a discrimination strategy corresponding to a current second fingerprint library, wherein the discrimination strategy is a parameter of the second fingerprint library; and taking the difference between the measured data and the actual data as the positioning precision of the second fingerprint database. Therefore, whether the preset positioning precision is reached or not can be determined according to the second fingerprint database positioning obtained through calculation.

Further, after selecting a part of the reference points to calculate the positioning accuracy of the second fingerprint database, the method further includes: judging whether the positioning precision of the second fingerprint database reaches a preset positioning precision or not; if so, outputting the second fingerprint database as the target fingerprint database; and if not, executing the step of increasing the positioning precision of the second fingerprint database by using the rest parts of the reference points and the reinforcement learning algorithm to obtain the target fingerprint database. If the positioning accuracy of the second fingerprint database reaches the preset positioning accuracy, the second fingerprint database can be directly used as a target fingerprint database, and if the positioning accuracy of the second fingerprint database does not reach the preset positioning accuracy, the second fingerprint database is used after the positioning accuracy is improved.

Further, the step of obtaining the target fingerprint database by increasing the positioning accuracy of the second fingerprint database by using the rest of the reference points and an enhanced learning algorithm includes: initializing a discrimination strategy corresponding to the current second fingerprint library to be an optimal discrimination strategy; iteratively updating the optimal discrimination strategy by using the rest reference points; and taking the second fingerprint library corresponding to the optimal discrimination strategy after iterative updating as the target fingerprint library. And updating the optimal judgment strategy for improving the positioning accuracy by using an enhanced learning algorithm so as to obtain a second fingerprint database with increased positioning accuracy as a target fingerprint database.

Further, pre-storing and randomly generating a plurality of other discrimination strategies different from the optimal discrimination strategy, and iteratively updating the optimal discrimination strategy by using the rest reference points, including: acquiring a target discrimination strategy in a mode of adopting the optimal discrimination strategy with a first probability and adopting other discrimination strategies with a second probability, wherein the sum of the first probability and the second probability is 1; respectively calculating a loss function corresponding to the target discrimination strategy and a loss function corresponding to the optimal discrimination strategy by using the rest reference points; judging whether the positioning precision of the second fingerprint library is improved by the target discrimination strategy relative to the optimal discrimination strategy according to the loss function corresponding to the target discrimination strategy and the loss function corresponding to the optimal discrimination strategy; and if so, taking the target judgment strategy as the optimal judgment strategy. And setting a corresponding probability selection discrimination strategy, and performing iterative updating on the optimal discrimination strategy by using the selected discrimination strategy, so that the optimal discrimination strategy is obtained by calculation conveniently.

Further, after the target discrimination policy is taken as an optimal discrimination policy, or after it is determined that the target discrimination policy cannot improve the positioning accuracy of the second fingerprint library relative to the optimal discrimination policy, the method further includes: judging whether a preset iteration condition is met; if so, reducing the first probability, outputting the optimal judgment strategy, and continuously executing the step of acquiring a target judgment strategy in a mode of adopting the optimal judgment strategy with the first probability and adopting other judgment strategies with the second probability; if not, the step of acquiring the target discrimination strategy by adopting the optimal discrimination strategy with the first probability and adopting other discrimination strategies with the second probability is executed. And outputting the optimal discrimination strategy when the iteration condition is met, and reducing the first probability so that the reinforcement learning algorithm can converge to obtain a good result.

Further, after outputting the optimal discrimination strategy, the method further includes: and taking the output optimal discrimination strategy as a discrimination strategy corresponding to the current second fingerprint library, and executing the step of calculating the measurement data of the part of the reference points by using the discrimination strategy corresponding to the current second fingerprint library. And the output result of the reinforcement learning algorithm is used for calculating the measurement data of the part of the reference points, so that the positioning precision of the second fingerprint library can be calculated according to the measurement data, and whether the second fingerprint library under the optimal judgment strategy can be used as the target fingerprint library or not is determined.

In a second aspect, an embodiment of the present application provides an apparatus for constructing a target fingerprint library, where the apparatus includes: the acquisition module is used for acquiring a first fingerprint library matched with the target space; the migration module is used for migrating the data in the first fingerprint database to a second fingerprint database through a migration learning algorithm, wherein the data in the first fingerprint database comprises data corresponding to a reference point, and the second fingerprint database is a fingerprint database corresponding to a target space; the calculation module is used for selecting part of the reference points to calculate the positioning precision of the second fingerprint library; and the enhancement module is used for increasing the positioning precision of the second fingerprint library by using the rest reference points and an enhanced learning algorithm to obtain the target fingerprint library when the precision of the second fingerprint library does not reach the positioning precision.

In a third aspect, an embodiment of the present application provides an electronic device, which includes one or more processors; a memory; one or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the one or more processors, the one or more programs configured to perform the method as applied to an electronic device, as described above.

In a fourth aspect, the present application provides a computer-readable storage medium having a program code stored therein, wherein the program code performs the above method when running.

According to the target fingerprint database construction method and device, the electronic equipment and the storage medium, data in a first fingerprint database matched with a target space are migrated to a second fingerprint database by means of migration learning, and the positioning accuracy of the second fingerprint database is calculated by means of partial reference points; and judging whether the preset positioning accuracy is achieved or not, and when the positioning accuracy of the second fingerprint database does not achieve the preset positioning accuracy, enhancing the positioning accuracy of the second fingerprint database by using the rest reference points and reinforcement learning to obtain a target fingerprint database so as to ensure that the constructed target fingerprint database has higher positioning accuracy while quickly constructing the target fingerprint database.

These and other aspects of the present application will be more readily apparent from the following description of the embodiments.

Drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings based on these drawings without creative efforts.

Fig. 1 shows a flowchart of a target fingerprint database construction method according to an embodiment of the present application.

Fig. 2 shows a flowchart of a target fingerprint database construction method according to another embodiment of the present application.

Fig. 3 shows a flow chart of part of the steps of the target fingerprint library construction method provided on the basis of the embodiment provided in fig. 2.

Fig. 4 shows a flowchart of a target fingerprint database construction method according to another embodiment of the present application.

Fig. 5 shows a flow chart of part of the steps of the target fingerprint library construction method provided on the basis of the embodiment provided in fig. 4.

Fig. 6 shows a flowchart of a target fingerprint database construction method according to still another embodiment of the present application.

Fig. 7 is a functional block diagram of a target fingerprint library construction apparatus according to an embodiment of the present application.

Fig. 8 shows a block diagram of an electronic device for executing a target fingerprint database construction method according to an embodiment of the present application.

Fig. 9 illustrates a storage medium provided in an embodiment of the present application and used for storing or carrying program code for implementing a target fingerprint library construction method according to an embodiment of the present application.

Detailed Description

In order to make the technical solutions better understood by those skilled in the art, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application.

Along with the development of science and technology, smart machine has walked into people's life, and smart machine can be based on sensor or internet and collect a large amount of data to satisfy people's diversified demand based on a large amount of data. Data is more and more in the process of time, and great difficulty is brought to data processing. In order to process a large amount of data, a fingerprint method is developed, which is to abstract the collected data into high-dimensional vectors, regard the vectors as unique fingerprints, then establish a fingerprint library, and perform matching when data processing is required, thereby rapidly processing a large amount of data.

The inventor finds in research that, due to the very high complexity of the fingerprint library, if a fingerprint library is reconstructed from the data acquisition, the time complexity is very high, and the rapid construction of the fingerprint library is difficult to realize. In real life, the difference between a plurality of fingerprint libraries is not very large, so that the existing fingerprint libraries can be copied to quickly construct a target fingerprint library.

However, the fingerprint library is very sensitive to fine variations, and if the existing fingerprint library which is copied is directly used as the target fingerprint library, very serious distortion is generated. In general, the method can be optimized by an error compensation method based on linear algebra, however, the slight change is usually a nonlinear change, and the error compensation method based on linear algebra has difficulty in achieving the expected effect.

Therefore, the inventor proposes a target fingerprint library construction method of the application, which includes acquiring a first fingerprint library matched with a target space, migrating data in the first fingerprint library to a second fingerprint library by means of migration learning, and calculating positioning accuracy of the second fingerprint library by means of part of reference points; and judging whether the preset positioning accuracy is achieved or not, and when the positioning accuracy of the second fingerprint database does not achieve the preset positioning accuracy, enhancing the positioning accuracy of the second fingerprint database by using the rest reference points and reinforcement learning to obtain a target fingerprint database so as to ensure that the constructed target fingerprint database has higher positioning accuracy while quickly constructing the target fingerprint database.

The following will describe embodiments of the present application in detail.

Referring to fig. 1, an embodiment of the present application provides a method for constructing a target fingerprint library, which may be applied to an electronic device, where the electronic device may be an intelligent device, or a local service, a cloud server, and the method may specifically include:

step S110, a first fingerprint database matching the target space is obtained.

The target space is a space to be positioned, the first fingerprint database is a fingerprint database matched with the target space in the existing fingerprint databases, and the fingerprint database comprises a plurality of data.

When the first fingerprint library matched with the target space is obtained, the features in the target space can be extracted, and the existing fingerprint library is screened as the first fingerprint library according to the feature matching.

The features in the target space may be the geometry of various components of the target space, e.g. signal generators, occlusions, shape of the space, etc. After the geometry is acquired, a fingerprint library matching the geometry of the target space may be used as the first fingerprint library. Of course, the features in the target space may also be data distribution, etc., and may be determined according to actual needs.

The geometric configuration corresponding to the existing fingerprint library may be obtained, the geometric configuration of the target space and the geometric configuration of the existing fingerprint library are matched, and a matching result is obtained, where the higher the value of the matching degree is, the more similar the geometric configurations are indicated, and the lower the value of the matching degree is, the larger the difference between the geometric configurations is indicated. In some embodiments, when the geometry of the target space is matched with the geometry corresponding to the existing fingerprint library, the fingerprint library corresponding to the geometry with the highest matching degree value may be used as the first fingerprint library.

For example, the existing fingerprint libraries are a fingerprint library a, a fingerprint library B, and a fingerprint library C, where the geometric configuration corresponding to the fingerprint library a is a geometric configuration a, the geometric configuration corresponding to the fingerprint library B is a geometric configuration B, and the geometric configuration corresponding to the fingerprint library C is a geometric configuration C; the geometry of the acquired target space is the target geometry.

Matching degrees obtained by matching the target geometric configuration with the geometric configuration a, the geometric configuration B and the geometric configuration c are respectively 30%, 80% and 75%, and then the fingerprint library B corresponding to the geometric configuration B can be used as the first fingerprint library.

In other embodiments, a preset matching degree may also be preset, and a fingerprint library corresponding to a geometric configuration whose matching degree value is greater than the preset matching degree may be used as the first fingerprint library; if a plurality of fingerprint libraries can be used as the first fingerprint library, one of the fingerprint libraries is arbitrarily selected as the first fingerprint library.

As in the foregoing example, if the preset matching degree is 70%, then the fingerprint database B corresponding to the geometric configuration B and the fingerprint database C corresponding to the geometric configuration C may both be used as the first fingerprint database, and thus, any one of the fingerprint databases B and C may be selected as the first fingerprint database.

And step S120, migrating the data in the first fingerprint database to a second fingerprint database through a migration learning algorithm.

After the first fingerprint library is obtained, the first fingerprint library is a fingerprint library matched with the target space, and therefore, if the fingerprint library corresponding to the target space, namely the second fingerprint library, is established, the difference between the target fingerprint library and the first fingerprint library is small, and therefore when the second fingerprint library is established, the second fingerprint library can be established quickly by directly utilizing the first fingerprint library.

The data in the first fingerprint may be migrated into the second fingerprint library through a migration learning algorithm, where the migration learning algorithm may be direct copy, post-filtering copy, or the like. After passing through the migration learning algorithm, the data in the first fingerprint database can be migrated into the second fingerprint database for subsequent use.

Before the transfer learning algorithm is utilized, a plurality of reference points can be randomly selected in advance, the plurality of reference points can be selected to be distributed in the whole target space as uniformly as possible so as to ensure the optimal effect of the algorithm, after the plurality of reference points are selected, the plurality of reference points can be accurately marked in a first fingerprint library and a second fingerprint library as far as possible, therefore, data in the first fingerprint library comprises data corresponding to the reference points, a part of the reference points are selected from the plurality of reference points and led into the transfer learning algorithm, and the data in the first fingerprint library is transferred to the second fingerprint library.

Step S130, selecting part of the reference points to calculate the positioning accuracy of the second fingerprint database.

After the second fingerprint library is obtained, the positioning accuracy of the second fingerprint library can be calculated to determine that the second fingerprint library is applicable to the target space. Thus, a selected portion of the reference points may be obtained, and the positioning accuracy of the second fingerprint library may be calculated based on the portion of the reference points.

And step S140, when the positioning accuracy of the second fingerprint database does not reach the preset positioning accuracy, increasing the positioning accuracy of the second fingerprint database by using the rest of the reference points and an enhanced learning algorithm to obtain the target fingerprint database.

Because the second fingerprint library is obtained by migrating the first fingerprint library, although the first fingerprint library is matched with the target space, slight changes still exist, and therefore after the positioning accuracy of the second fingerprint library is obtained through calculation, if the positioning accuracy of the second fingerprint library does not reach the preset positioning accuracy, the second fingerprint library cannot be well applicable to the target space, and therefore the second fingerprint library needs to be optimized to improve the positioning accuracy of the second fingerprint library to obtain the target fingerprint library, and therefore the target fingerprint library can be more applicable to the target space. When the second fingerprint library is optimized, the positioning accuracy of the second fingerprint library can be increased by using the rest reference points and an enhanced learning algorithm to obtain a target fingerprint library.

The target fingerprint database construction method provided by the application comprises the steps of obtaining a first fingerprint database matched with a target space; migrating the data in the first fingerprint database to a second fingerprint database through a migration learning algorithm; and selecting part of the reference points to calculate the positioning accuracy of the second fingerprint database, and increasing the positioning accuracy of the second fingerprint database by using the rest of the reference points and an enhanced learning algorithm to obtain a target fingerprint database when the positioning accuracy of the second fingerprint database does not reach the preset positioning accuracy. The second fingerprint library can be quickly constructed through the transfer learning algorithm, the positioning accuracy of the second fingerprint library is increased by utilizing the reinforcement learning algorithm to obtain the target fingerprint library, and the constructed target fingerprint library is ensured to have higher positioning accuracy while the target fingerprint library is quickly constructed.

Referring to fig. 2, another embodiment of the present application provides a method for constructing a target fingerprint database, and the process of calculating the positioning accuracy of the second fingerprint database is described in detail based on the previous embodiment. Specifically, the method may include:

step S210, a first fingerprint database matching the target space is obtained.

Step S220, migrating the data in the first fingerprint database to a second fingerprint database through a migration learning algorithm.

Step S210 and step S220 refer to corresponding parts of the foregoing embodiments, and are not described herein again.

Step S230, selecting a part of the reference points to calculate the positioning accuracy of the second fingerprint database.

The method comprises the steps of selecting a plurality of reference points in advance, selecting partial reference points from the reference points, importing the partial reference points into a migration learning algorithm, and migrating data in the first fingerprint library into the second fingerprint library, wherein however, whether the second fingerprint library can be applied to a target space or not needs to calculate the positioning accuracy of the second fingerprint library for determination.

Since the second fingerprint library is obtained by migrating the first fingerprint library, in order to ensure that the calculated positioning accuracy of the second fingerprint library is accurate, whether migration of the first fingerprint library is completed or not can be judged before the positioning accuracy of the second fingerprint library is calculated, and if yes, the positioning accuracy of the second fingerprint library can be calculated by using part of the reference points; if not, the first fingerprint library can be continuously migrated until the migration of the first fingerprint library is completed.

As an embodiment, when the first fingerprint library is migrated by using a migration learning algorithm, parameters of the migration learning algorithm may be changed, and after the parameters of the migration learning algorithm are changed, if the positioning accuracy of the second fingerprint library cannot be significantly improved, it may be considered that the migration of the first fingerprint library has been completed.

Specifically, referring to fig. 3, a process of selecting a part of the reference points to calculate the positioning accuracy of the second fingerprint library is shown, and may include the following steps:

step S231, calculating a part of the measurement data of the reference point by using the discrimination policy corresponding to the current second fingerprint library.

After it is determined that the migration of the second fingerprint library is completed, a discrimination strategy corresponding to the current second fingerprint library may be obtained, where the discrimination strategy corresponding to the current second fingerprint library is a parameter of the second fingerprint library after the migration learning is completed. That is, the determination policy is a parameter related to the second fingerprint library. In order to determine whether the positioning accuracy of the second fingerprint database can reach the preset positioning accuracy, the current second fingerprint database is used for identifying a reference point to obtain a corresponding identification result, i.e. measurement data.

Under the condition of determining the second fingerprint library distinguishing strategy, identifying selected partial reference points according to the distinguishing strategy, and obtaining measurement data corresponding to each reference point.

Step S232, using the difference between the measured data and the actual data as the positioning accuracy of the second fingerprint database.

The reference points are pre-marked in the second fingerprint database and the first fingerprint database, and when the reference points are marked, marked results can be correspondingly stored, so that the marked results are actual data, and therefore the actual data of all the reference points can be acquired. When the measurement data corresponding to a part of the reference points is obtained through calculation, the prestored actual data corresponding to the part of the reference points can be obtained, the difference between the measurement data and the actual data is obtained, and the difference is used as the positioning accuracy of the second fingerprint database.

For example, the measurement data and the actual data are three-dimensional coordinates of a reference point in the target space, the measurement data is the three-dimensional coordinates obtained by identifying and positioning the reference point by using the discrimination policy of the current second fingerprint library, the actual data is the three-dimensional coordinates recorded when the reference point is labeled, and one reference point may correspond to one measurement data and one actual data. When the positioning accuracy of the second fingerprint database is calculated, the difference between two three-dimensional coordinates corresponding to the same reference point may be calculated to obtain the positioning accuracy of the second fingerprint database. The expression form of the measurement data and the actual data is not limited to the three-dimensional coordinates, and can be selected according to actual needs.

Step 240, judging whether the positioning precision of the second fingerprint database reaches a preset positioning precision, if so, executing step 250; if not, go to step S260.

The method comprises the steps that preset positioning accuracy is preset, wherein the preset positioning accuracy indicates that the accuracy of a fingerprint library reaching the preset positioning accuracy is high, large errors cannot occur, and the preset positioning accuracy can be set according to actual needs. It is understood that the positioning accuracy is the difference between the measured data and the actual data, and then the corresponding preset positioning accuracy is also the difference between the measured data and the actual data under the condition that the accuracy of the fingerprint database is higher. After the positioning accuracy of the second fingerprint database is obtained through calculation, the size relationship between the positioning accuracy and the preset positioning accuracy can be judged; since the positioning accuracy is the difference between the measured data and the actual data, the smaller the value of the positioning accuracy, the higher the accuracy of the fingerprint library. Therefore, when the positioning accuracy is less than or equal to the preset positioning accuracy, the positioning accuracy of the second fingerprint library may be considered to reach the preset positioning accuracy, so that the step S250 may be continuously performed, and if the positioning accuracy is greater than the preset positioning accuracy, the positioning accuracy of the second fingerprint library may be considered to not reach the preset positioning accuracy, so that the step S260 may be continuously performed.

And step S250, outputting the second fingerprint database as the target fingerprint database.

If the positioning precision of the second fingerprint database is judged to reach the preset positioning precision, the accuracy of the second fingerprint database relative to the target space is high, and the second fingerprint database can be directly used, so that the second fingerprint database can be directly output as the target fingerprint database.

Step S260, the step of obtaining the target fingerprint database by using the remaining reference points and the reinforcement learning algorithm to increase the positioning accuracy of the second fingerprint database is performed.

If the positioning precision of the second fingerprint library is judged not to reach the preset positioning precision, the fact that a larger error is caused if the second fingerprint library is directly used in the target space is shown, the accuracy is low, the second fingerprint library can be continuously used only by improving the positioning precision of the second fingerprint library, and therefore the step that the positioning precision of the second fingerprint library is increased by using the rest of the reference points and the reinforcement learning algorithm to obtain the target fingerprint library can be executed.

According to the target fingerprint database construction method, the positioning accuracy of the second fingerprint database after migration is completed is calculated; judging whether the positioning precision of the second fingerprint database reaches a preset positioning precision or not, and if so, directly using the second fingerprint database as a target fingerprint database; if not, the positioning accuracy of the second fingerprint database is increased by using the rest parts of the reference points and the reinforcement learning algorithm to obtain the target fingerprint database, so that the constructed target fingerprint database is ensured to have higher positioning accuracy while the target fingerprint database is constructed rapidly.

Referring to fig. 4, another embodiment of the present application provides a method for constructing a target fingerprint library, and on the basis of the previous embodiment, a process of increasing the positioning accuracy of the second fingerprint library by using an reinforcement learning algorithm is described in detail. Specifically, the method may include:

in step S310, a first fingerprint database matching the target space is obtained.

And step S320, migrating the data in the first fingerprint database to a second fingerprint database through a migration learning algorithm.

And step S330, selecting part of the reference points to calculate the positioning accuracy of the second fingerprint database.

The steps S310 to S330 can refer to the corresponding parts of the foregoing embodiments, and are not described herein again.

Step S340, if the positioning accuracy of the second fingerprint library does not reach the preset positioning accuracy, initializing the discrimination policy corresponding to the current second fingerprint library to be the optimal discrimination policy.

And step S350, iteratively updating the optimal judgment strategy by using the rest reference points.

And if the positioning accuracy of the second fingerprint database does not reach the preset positioning accuracy, the positioning accuracy of the second fingerprint database needs to be enhanced, so that the method is better suitable for the target space. Specifically, when the positioning accuracy of the second fingerprint library is enhanced, the discrimination strategy corresponding to the second fingerprint library may be iteratively updated by using an enhanced learning algorithm to obtain an optimal discrimination strategy. The optimal discrimination strategy can be understood as a parameter corresponding to the second fingerprint library when the positioning accuracy of the second fingerprint library is highest. The positioning accuracy of the second fingerprint database corresponding to the optimal discrimination strategy is higher than that of the second fingerprint databases corresponding to other discrimination strategies. Therefore, we need to determine the best discriminant strategy to maximize the positioning accuracy of the second fingerprint library.

When the optimal discrimination strategy is iteratively updated, an optimal discrimination strategy needs to be determined to initialize the optimal discrimination strategy, that is, a discrimination strategy needs to be designated as the current optimal discrimination strategy. As an embodiment, when the second fingerprint library is obtained after the completion of the migration learning, the discrimination policy corresponding to the second fingerprint library obtained after the completion of the migration learning may be initialized to the optimal discrimination policy. That is, the current optimal discrimination strategy is the discrimination strategy corresponding to the second fingerprint library after the migration learning is completed. After the optimal discrimination strategy is initialized, the optimal discrimination strategy can be iteratively updated by using the rest of the reference points. Among the preselected reference points, part of the reference points are selected in the migration learning, and the rest of the reference points are the rest of the reference points.

Before the enhancement processing algorithm is used for enhancing the positioning accuracy of the second fingerprint library, a suitable loss function may be selected to measure the positioning accuracy, wherein the loss function may be designed from multiple dimensions, for example, the loss function may be a maximum positioning error, an average error, and the like, and specifically, the loss function may be selected according to actual needs. The enhancement processing algorithm can be a countermeasure generation network and the like, and can also select or train a corresponding model for enhancement processing according to needs.

Referring to fig. 5, a specific process for iteratively updating the optimal decision strategy using the remaining reference points is shown, which may include the following steps.

Step S351 is to acquire a target discrimination policy by adopting the optimal discrimination policy with the first probability and adopting another discrimination policy with the second probability.

And the judgment strategy is a parameter corresponding to the second fingerprint library, and the judgment strategy corresponding to the second fingerprint library obtained after the transfer learning is finished is the current optimal judgment strategy. Then, the parameters of the second fingerprint library may be randomly changed to obtain a plurality of other discrimination strategies, and when the plurality of other discrimination strategies are randomly obtained, the other discrimination strategies may be stored for use.

In order to find the optimal discrimination strategy from the plurality of discrimination strategies, the optimal discrimination strategy may be adopted at a first probability, and the other discrimination strategies may be adopted at a second probability to obtain the target discrimination strategy. Wherein the target discrimination policy is any one selected from the optimal discrimination policy and other discrimination policies. That is, when the discrimination strategy is selected as the target discrimination strategy, there is a first probability that the best discrimination strategy that has been confirmed is selected, and there is a second probability that the other discrimination strategies are selected. For example, if the first probability is 0.3 and the second probability is 0.7, in one selection process, the best discrimination strategy determined at present is selected as the target discrimination strategy with a probability of 0.3, and other discrimination strategies are selected as the target discrimination strategies with a probability of 0.7.

Step S352, calculating a loss function corresponding to the target discrimination policy and a loss function corresponding to the optimal discrimination policy respectively using the remaining reference points.

Step S353, judging whether the target discrimination strategy improves the positioning precision of the second fingerprint library relative to the optimal discrimination strategy; if yes, go to step S354; if not, go to step S351.

After the target discrimination strategy is obtained, the loss function under the target discrimination strategy can be calculated by using the rest of the reference points, and the loss function corresponding to the optimal discrimination strategy is calculated. Since the loss function can be used to measure the positioning accuracy of the second fingerprint library, it can be determined whether the positioning accuracy of the second fingerprint library is improved by the target discrimination policy relative to the optimal discrimination policy according to the loss function corresponding to the target discrimination policy and the loss function corresponding to the optimal discrimination policy.

As described above, if the loss function is designed with the maximum positioning error or the average error, the smaller the value of the loss function, the higher the positioning accuracy. Thus, the magnitude of the value of the loss function corresponding to the optimal discrimination policy and the magnitude of the value of the loss function corresponding to the target discrimination policy may be compared to determine whether the target discrimination policy improves the positioning accuracy of the second fingerprint library with respect to the optimal discrimination policy.

If the value of the loss function corresponding to the target discrimination policy is smaller than the value of the loss function corresponding to the optimal discrimination policy, it indicates that the positioning accuracy of the second fingerprint library when the target discrimination policy is adopted is better than the positioning accuracy when the optimal discrimination policy is adopted, that is, the target discrimination policy improves the positioning accuracy of the second fingerprint library relative to the optimal discrimination policy, so that step S354 may be executed.

If the value of the loss function corresponding to the target discrimination policy is greater than or equal to the value of the loss function corresponding to the optimal discrimination policy, it indicates that the positioning accuracy of the second fingerprint library when the target discrimination policy is adopted is worse than the positioning accuracy when the optimal discrimination policy is adopted, that is, the positioning accuracy of the second fingerprint library is not improved by the target discrimination policy with respect to the optimal discrimination policy, the previously determined optimal discrimination policy may be maintained, and step S351 may be continuously performed.

After step S351 is executed, step S352 and step S353 are executed in sequence, and step S354 is executed until it is determined that the target discrimination policy improves the positioning accuracy of the second fingerprint database with respect to the optimal discrimination policy.

Step S354, the target discrimination policy is taken as the optimal discrimination policy.

If it is determined that the target discrimination strategy improves the positioning accuracy of the second fingerprint library relative to the optimal discrimination strategy, the target discrimination strategy may be taken as the optimal discrimination strategy, which indicates that the target discrimination strategy is the discrimination strategy with the best positioning accuracy at present, so as to update the optimal discrimination strategy.

Step S360, if the positioning accuracy of the second fingerprint database corresponding to the optimal judgment strategy after the iterative updating reaches the preset positioning accuracy, taking the second fingerprint database corresponding to the optimal judgment strategy after the iterative updating as the target fingerprint database.

After the optimal discrimination strategy is iteratively updated, the optimal discrimination strategy may be output, the positioning accuracy of the second fingerprint library may be recalculated according to the optimal discrimination strategy, and if the positioning of the second fingerprint library reaches a preset positioning accuracy, the second fingerprint library corresponding to the optimal discrimination strategy may be used as the target fingerprint library.

According to the target fingerprint database construction method, the optimal judgment strategy is adopted according to the first probability, and other judgment strategies are adopted according to the second probability to obtain a target judgment strategy; respectively calculating a loss function corresponding to the target discrimination strategy and a loss function corresponding to the optimal discrimination strategy by using the rest reference points; and judging whether the positioning precision of the second fingerprint base is improved by the target discrimination strategy relative to the optimal discrimination strategy, and if so, taking the target discrimination strategy as the optimal discrimination strategy. And designing a loss function corresponding to the reinforcement learning algorithm, continuously updating a discrimination strategy of the second fingerprint library according to the loss function, and correcting nonlinear change in the discrimination strategy, so that the positioning precision and the stability of the second fingerprint library are improved.

Referring to fig. 6, a method for constructing a target fingerprint database is provided in yet another embodiment of the present application, and a process after the target discrimination policy is taken as the optimal discrimination policy is mainly described on the basis of the previous embodiment. As shown in fig. 6, the method may include:

step S401, a first fingerprint database matched with the target space is obtained.

And step S402, migrating the data in the first fingerprint database to a second fingerprint database through a migration learning algorithm.

And step S403, selecting part of the reference points to calculate the positioning accuracy of the second fingerprint database.

Step S404, if the positioning accuracy of the second fingerprint library does not reach the preset positioning accuracy, initializing the discrimination policy corresponding to the current second fingerprint library to an optimal discrimination policy.

Step S405, the target discrimination strategy is obtained by adopting the optimal discrimination strategy with the first probability and adopting other discrimination strategies with the second probability.

Step S406, calculating a loss function corresponding to the target discrimination policy and a loss function corresponding to the optimal discrimination policy respectively using the remaining reference points.

Step S407, judging whether the positioning accuracy of the second fingerprint library is improved by the target discrimination strategy relative to the optimal discrimination strategy; if yes, go to step S408; if not, go to step S409.

If it is determined that the target discrimination strategy improves the positioning accuracy of the second fingerprint library relative to the optimal discrimination strategy, step S408 may be executed; if it is determined that the target discrimination policy does not improve the second fingerprint library positioning accuracy relative to the optimal discrimination policy, step S409 may be performed.

Step S408, using the target discrimination policy as the optimal discrimination policy.

The remaining contents of step S401 to step S408 refer to the corresponding parts of the foregoing embodiments, and are not described herein again.

Step S409, judging whether preset iteration conditions are met; if yes, go to step S410; if not, go to step S405.

After the target discrimination strategy is taken as the optimal discrimination strategy, the optimal discrimination strategy needs to be output, so that the positioning accuracy of the second fingerprint library is calculated according to the optimal discrimination strategy, and whether the positioning accuracy of the second fingerprint library reaches a preset positioning accuracy is judged.

In some embodiments, before outputting the optimal decision policy, it may be determined whether a preset iteration condition is satisfied, if the preset iteration condition is satisfied, step S410 may be performed, and if the preset iteration condition is not satisfied, step S405 may be performed.

The preset iteration condition may be that whether a preset iteration number is reached is judged, before the positioning accuracy of the second fingerprint library is enhanced by using an enhanced learning algorithm, a counter may be initialized, and the counter is incremented by 1 every time iteration is performed, so that the corresponding iteration number can be obtained through the counter. And when the optimal judgment strategy is output, clearing the counter and counting again.

The preset iteration number may be set according to an actual use requirement, for example, if the optimal discrimination strategy needs to be input every 100 seconds, and the computer calculates 10,000 times per second, the preset iteration number is 1,000,000 times.

After the iteration times are obtained, whether the iteration times are greater than or equal to the preset iteration times or not can be judged, if yes, the preset iteration conditions can be judged to be met, and the step S411 can be continuously executed; if not, it may be determined that the preset iteration condition is not satisfied, then step S405 may be continuously performed.

And S410, reducing the first probability and outputting the optimal judgment strategy.

If the first probability is judged to be satisfied, the first probability can be reduced, and the optimal judgment strategy is output, wherein the output optimal judgment strategy is used for calculating the positioning accuracy of the second fingerprint library so as to judge whether the positioning accuracy of the second fingerprint library reaches the preset positioning accuracy.

In one embodiment, the first probability may be reduced at a predetermined ratio, and when the first probability is reduced at the predetermined ratio, the first probability may not be zero, and it usually takes a long time to enhance the positioning accuracy of the second fingerprint library by the reinforcement learning, so that after step S410, the loop of steps S405 to S410 may be continued, and the nonlinear distortion may be corrected as much as possible by the reinforcement learning.

After the optimal discrimination strategy is output, step S411 may also be performed. That is, after step S410 is executed, while the steps from step S405 to step S410 are executed in a loop, the optimal discrimination policy may be output while the iterative update of the optimal discrimination policy is continued, and step S411 may be executed based on the optimal discrimination policy.

Step S411, if the positioning accuracy of the second fingerprint library corresponding to the optimal decision policy after the iterative update reaches a preset positioning accuracy, taking the second fingerprint library corresponding to the optimal decision policy after the iterative update as the target fingerprint library.

After the optimal discrimination strategy is iteratively updated, the optimal discrimination strategy may be output, and the positioning accuracy of the second fingerprint library may be recalculated according to the optimal discrimination strategy, and if the positioning accuracy of the second fingerprint library reaches a preset positioning accuracy, the second fingerprint library corresponding to the optimal discrimination strategy may be used as the target fingerprint library.

For example, with the continuous update of reinforcement learning, the output optimal discrimination strategies are X and Y in sequence, and the preset positioning accuracy is z. When the received optimal judgment strategy is X, the positioning precision of the second fingerprint library is obtained by calculation according to X, and when the received optimal judgment strategy is Y, the positioning precision of the second fingerprint library is obtained by calculation according to Y, wherein X > z > Y, and the smaller the value of the positioning precision is, the higher the positioning precision of the second fingerprint library is. Then, when the received optimal discrimination strategy is X, the positioning accuracy of the corresponding second fingerprint library does not reach the preset positioning accuracy, so that the output optimal discrimination strategy can be continuously received as Y, and when the optimal discrimination strategy is Y, the positioning accuracy of the corresponding second fingerprint library reaches the preset positioning accuracy, so that the second fingerprint library corresponding to the optimal discrimination strategy as Y can be output as the target fingerprint library, and all the circulation processes are ended. Therefore, the target fingerprint database obtained at this time can be well suitable for the target space.

According to the target fingerprint library construction method, an optimal discrimination strategy of a second fingerprint library is continuously optimized by using an enhanced learning algorithm, the optimal discrimination strategy is output in stages, the positioning precision of the second fingerprint library is recalculated by using the optimal discrimination strategy output by the enhanced learning algorithm, and when the positioning precision of the second fingerprint library reaches a preset positioning precision, the second fingerprint library corresponding to the optimal discrimination strategy is output as the target fingerprint library. The method and the device realize that the constructed target fingerprint database has higher positioning precision while the target fingerprint database is constructed rapidly.

Referring to fig. 7, a target fingerprint library constructing apparatus 500 according to an embodiment of the present application is shown, where the target fingerprint library constructing apparatus 500 includes an obtaining module 510, a migrating module 520, a calculating module 530, and an enhancing module 540.

The obtaining module 510 is configured to obtain a first fingerprint library matching a target space; the migration module 520 is configured to migrate data in the first fingerprint database to a second fingerprint database through a migration learning algorithm, where the data in the first fingerprint database includes data corresponding to a reference point, and the second fingerprint database is a fingerprint database corresponding to a target space; the calculating module 530 is configured to select a part of the reference points to calculate the positioning accuracy of the second fingerprint library; and the enhancing module 540 is configured to, when the accuracy of the second fingerprint database does not reach the positioning accuracy, increase the positioning accuracy of the second fingerprint database by using the remaining reference points and an enhanced learning algorithm to obtain the target fingerprint database.

Further, actual data of the reference point in the target space is pre-stored, and the calculating module 530 is further configured to calculate part of the measurement data of the reference point by using a discrimination policy corresponding to the current second fingerprint library, where the discrimination policy is a parameter of the second fingerprint library; and taking the difference between the measured data and the actual data as the positioning precision of the second fingerprint database.

Further, the target fingerprint library constructing apparatus 500 further includes a determining module, after the positioning accuracy of the second fingerprint library is calculated by selecting a part of the reference points, the determining module is configured to determine whether the positioning accuracy of the second fingerprint library reaches a preset positioning accuracy; if so, outputting the second fingerprint database as the target fingerprint database; if not, the enhancement module 540 is instructed to execute the step of increasing the positioning precision of the second fingerprint database by using the rest of the reference points and the enhanced learning algorithm to obtain the target fingerprint database.

Further, the enhancing module 540 is further configured to initialize a discrimination policy corresponding to the current second fingerprint library to an optimal discrimination policy; iteratively updating the optimal discrimination strategy by using the rest reference points; and taking the second fingerprint library corresponding to the optimal discrimination strategy after iterative updating as the target fingerprint library.

Further, a plurality of other discrimination strategies different from the optimal discrimination strategy are pre-stored and randomly generated, the enhancing module 540 is further configured to obtain a target discrimination strategy by adopting the optimal discrimination strategy with a first probability and adopting the other discrimination strategies with a second probability, and a sum of the first probability and the second probability is 1; respectively calculating a loss function corresponding to the target discrimination strategy and a loss function corresponding to the optimal discrimination strategy by using the rest reference points; judging whether the positioning precision of the second fingerprint library is improved by the target discrimination strategy relative to the optimal discrimination strategy according to the loss function corresponding to the target discrimination strategy and the loss function corresponding to the optimal discrimination strategy; and if so, taking the target judgment strategy as the optimal judgment strategy.

Further, after the target discrimination policy is taken as an optimal discrimination policy, or after it is determined that the positioning accuracy of the second fingerprint library cannot be improved by the target discrimination policy relative to the optimal discrimination policy, the enhancing module 540 is further configured to determine whether a preset iteration condition is satisfied; if so, reducing the first probability, outputting the optimal judgment strategy, and continuously executing the step of acquiring a target judgment strategy in a mode of adopting the optimal judgment strategy with the first probability and adopting other judgment strategies with the second probability; if not, the step of acquiring the target discrimination strategy by adopting the optimal discrimination strategy with the first probability and adopting other discrimination strategies with the second probability is executed.

Further, after outputting the optimal distinguishing policy, the calculating module 530 is further configured to use the output optimal distinguishing policy as the distinguishing policy corresponding to the current second fingerprint library, and perform the step of calculating the measurement data of the part of the reference points by using the distinguishing policy corresponding to the current second fingerprint library.

It should be noted that, as will be clear to those skilled in the art, for convenience and brevity of description, the specific working processes of the above-described apparatuses and modules may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.

In summary, the target fingerprint database construction method provided by the application obtains a first fingerprint database matched with a target space; migrating the data in the first fingerprint database to a second fingerprint database through a migration learning algorithm; and selecting part of the reference points to calculate the positioning accuracy of the second fingerprint database, and increasing the positioning accuracy of the second fingerprint database by using the rest of the reference points and an enhanced learning algorithm to obtain a target fingerprint database when the positioning accuracy of the second fingerprint database does not reach the preset positioning accuracy. The second fingerprint library can be quickly constructed through the transfer learning algorithm, the positioning accuracy of the second fingerprint library is increased by utilizing the reinforcement learning algorithm to obtain the target fingerprint library, and the constructed target fingerprint library is ensured to have higher positioning accuracy while the target fingerprint library is quickly constructed.

In the several embodiments provided in the present application, the coupling or direct coupling or communication connection between the modules shown or discussed may be through some interfaces, and the indirect coupling or communication connection between the devices or modules may be in an electrical, mechanical or other form.

In addition, functional modules in the embodiments of the present application may be integrated into one processing module, or each of the modules may exist alone physically, or two or more modules are integrated into one module. The integrated module can be realized in a hardware mode, and can also be realized in a software functional module mode.

Referring to fig. 8, a block diagram of an electronic device according to an embodiment of the present disclosure is shown. The electronic device 600 may be a smart phone, a tablet computer, an electronic book, a server, or other electronic devices capable of running an application program, or a server. The electronic device 600 in the present application may include one or more of the following components: a processor 610, a memory 620, and one or more applications, wherein the one or more applications may be stored in the memory 620 and configured to be executed by the one or more processors 610, the one or more programs configured to perform the methods as described in the aforementioned method embodiments.

The processor 610 may include one or more processing cores. The processor 610 interfaces with various components throughout the electronic device 600 using various interfaces and circuitry to perform various functions of the electronic device 600 and process data by executing or executing instructions, programs, code sets, or instruction sets stored in the memory 620 and invoking data stored in the memory 620. Alternatively, the processor 610 may be implemented in hardware using at least one of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), and Programmable Logic Array (PLA). The processor 610 may integrate one or a combination of a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), a modem, and the like. Wherein, the CPU mainly processes an operating system, a user interface, an application program and the like; the GPU is used for rendering and drawing display content; the modem is used to handle wireless communications. It is understood that the modem may not be integrated into the processor 610, but may be implemented by a communication chip.

The Memory 620 may include a Random Access Memory (RAM) or a Read-Only Memory (Read-Only Memory). The memory 620 may be used to store instructions, programs, code sets, or instruction sets. The memory 620 may include a stored program area and a stored data area, wherein the stored program area may store instructions for implementing an operating system, instructions for implementing at least one function (such as a touch function, a sound playing function, an image playing function, etc.), instructions for implementing various method embodiments described below, and the like. The data storage area may also store data created during use by the electronic device 600 (e.g., phone books, audio-visual data, chat log data), and so forth.

Referring to fig. 9, a block diagram of a computer-readable storage medium according to an embodiment of the present application is shown. The computer-readable storage medium 700 has stored therein program code that can be called by a processor to execute the methods described in the above-described method embodiments.

The computer-readable storage medium 700 may be an electronic memory such as a flash memory, an EEPROM (electrically erasable programmable read only memory), an EPROM, a hard disk, or a ROM. Optionally, the computer-readable storage medium 700 includes a non-transitory computer-readable storage medium. The computer readable storage medium 700 has storage space for program code 710 to perform any of the method steps of the method described above. The program code can be read from or written to one or more computer program products. The program code 710 may be compressed, for example, in a suitable form.

Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit the same; although the present application has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; such modifications and substitutions do not necessarily depart from the spirit and scope of the corresponding technical solutions in the embodiments of the present application.

Claims

1. A method for constructing a target fingerprint database, the method comprising:

acquiring a first fingerprint database matched with a target space;

migrating data in the first fingerprint database to a second fingerprint database through a migration learning algorithm, wherein the data in the first fingerprint database comprises data corresponding to a reference point, and the second fingerprint database is a fingerprint database corresponding to a target space;

selecting part of the reference points to calculate the positioning accuracy of the second fingerprint library;

and when the positioning accuracy of the second fingerprint library does not reach the preset positioning accuracy, increasing the positioning accuracy of the second fingerprint library by using the rest parts of the reference points and an enhanced learning algorithm to obtain the target fingerprint library.

2. The method according to claim 1, wherein actual data of the reference point in the target space is pre-stored, and the selecting the portion of the reference point to calculate the positioning accuracy of the second fingerprint library comprises:

calculating part of measurement data of the reference point by using a discrimination strategy corresponding to the current second fingerprint library, wherein the discrimination strategy is a parameter of the second fingerprint library;

and taking the difference between the measured data and the actual data as the positioning precision of the second fingerprint database.

3. The method according to claim 1, wherein after the selecting the portion of the reference points to calculate the positioning accuracy of the second fingerprint library, further comprising:

judging whether the positioning precision of the second fingerprint database reaches a preset positioning precision or not;

if so, outputting the second fingerprint database as the target fingerprint database;

and if not, executing the step of increasing the positioning precision of the second fingerprint database by using the rest parts of the reference points and the reinforcement learning algorithm to obtain the target fingerprint database.

4. The method of claim 3, wherein using the remaining portion of the reference points and an reinforcement learning algorithm to increase the positioning accuracy of the second fingerprint library to obtain the target fingerprint library comprises:

initializing a discrimination strategy corresponding to the current second fingerprint library to be an optimal discrimination strategy;

iteratively updating the optimal discrimination strategy by using the rest reference points;

and taking the second fingerprint library corresponding to the optimal discrimination strategy after iterative updating as the target fingerprint library.

5. The method of claim 4, wherein pre-storing and randomly generating a plurality of other discriminant strategies different from the optimal discriminant strategy, and wherein iteratively updating the optimal discriminant strategy using remaining portions of the reference points comprises:

acquiring a target discrimination strategy in a mode of adopting the optimal discrimination strategy with a first probability and adopting other discrimination strategies with a second probability, wherein the sum of the first probability and the second probability is 1;

respectively calculating a loss function corresponding to the target discrimination strategy and a loss function corresponding to the optimal discrimination strategy by using the rest reference points;

judging whether the positioning precision of the second fingerprint library is improved by the target discrimination strategy relative to the optimal discrimination strategy according to the loss function corresponding to the target discrimination strategy and the loss function corresponding to the optimal discrimination strategy;

and if so, taking the target judgment strategy as the optimal judgment strategy.

6. The method of claim 5, wherein after the target discrimination strategy is identified as an optimal discrimination strategy or after determining that the target discrimination strategy does not improve the positioning accuracy of the second fingerprint library relative to the optimal discrimination strategy, the method further comprises:

judging whether a preset iteration condition is met;

if so, reducing the first probability, outputting the optimal judgment strategy, and continuously executing the step of acquiring a target judgment strategy in a mode of adopting the optimal judgment strategy with the first probability and adopting other judgment strategies with the second probability;

if not, the step of acquiring the target discrimination strategy by adopting the optimal discrimination strategy with the first probability and adopting other discrimination strategies with the second probability is executed.

7. The method of claim 6, wherein after outputting the optimal decision strategy, further comprising:

and taking the output optimal discrimination strategy as a discrimination strategy corresponding to the current second fingerprint library, and executing the step of calculating the measurement data of the part of the reference points by using the discrimination strategy corresponding to the current second fingerprint library.

8. An apparatus for constructing a target fingerprint library, the apparatus comprising:

the acquisition module is used for acquiring a first fingerprint library matched with the target space;

the migration module is used for migrating the data in the first fingerprint database to a second fingerprint database through a migration learning algorithm, wherein the data in the first fingerprint database comprises data corresponding to a reference point, and the second fingerprint database is a fingerprint database corresponding to a target space;

the calculation module is used for selecting part of the reference points to calculate the positioning precision of the second fingerprint library;

and the enhancement module is used for increasing the positioning precision of the second fingerprint library by using the rest reference points and an enhanced learning algorithm to obtain the target fingerprint library when the precision of the second fingerprint library does not reach the positioning precision.

9. An electronic device, characterized in that the electronic device comprises:

one or more processors;

a memory electrically connected with the one or more processors;

one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs configured to perform the method of any of claims 1-7.

10. A computer-readable storage medium, characterized in that a program code is stored in the computer-readable storage medium, which program code can be called by a processor to execute the method according to any one of claims 1 to 7.