Classifications

• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
  • G06T7/00—Image analysis
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
  • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
  • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
  • G06V40/12—Fingerprints or palmprints

Definitions

• This disclosure relates to the technical fields of a learning device, a fingerprint state estimation device, a learning method, a fingerprint state estimation method, and a recording medium.
• Patent Document 1 discloses technology related to matching of people whose fingerprints are difficult to match, such as the elderly, women, and children.
• Patent Document 2 discloses technology related to fingerprint authentication of children who are still growing.
• Patent Document 3 discloses technology for extracting features suitable for the intended use from a fingerprint image.
• Patent Document 4 discloses technology for enlarging a fingerprint image to accommodate the fingerprints of young people, whose fingerprints grow significantly in a short period of time.
• One aspect of the learning device disclosed herein includes a generation means for inputting a first image including a first fingerprint into an image generation model to generate a second image including a second fingerprint in a state different from that of the first fingerprint, a calculation means for matching the first image with the second image to calculate a matching score indicating the degree of matching, and a learning means for learning a machine learning model that expresses the relationship between the second image and an output value based on the matching score for the second image.
• One aspect of the fingerprint state estimation device disclosed herein is a fingerprint state estimation device that uses a machine learning model trained by a learning device that includes: a generation means for inputting a first image including a first fingerprint into an image generation model and generating a second image including a second fingerprint whose state is different from that of the first fingerprint; a calculation means for matching the first image with the second image and calculating a matching score indicating a degree of matching; and a learning means for learning a machine learning model that expresses a relationship between the second image and an output value based on the matching score for the second image.
• the fingerprint state estimation device includes: a fingerprint acquisition means for acquiring a fingerprint image of a target user; a state value calculation means for inputting the fingerprint image of the target user into the machine learning model and calculating a state value relating to the state of the fingerprint of the target user; and an output means for outputting the state value.
• One aspect of the learning method disclosed herein involves at least one computer inputting a first image including a first fingerprint into an image generation model, generating a second image including a second fingerprint in a state different from that of the first fingerprint, matching the first image with the second image, calculating a matching score indicating the degree of matching, and learning a machine learning model that represents the relationship between the second image and an output value based on the matching score for the second image.
• One aspect of the fingerprint state estimation method disclosed herein is a fingerprint state estimation method using a machine learning model learned by a learning method, which involves inputting a first image including a first fingerprint into an image generation model by at least one computer, generating a second image including a second fingerprint having a different state from the first fingerprint, matching the first image with the second image, calculating a matching score indicating the degree of matching, and learning a machine learning model that expresses the relationship between the second image and an output value based on the matching score for the second image.
• the method obtains a fingerprint image of a target user, inputs the fingerprint image of the target user into the machine learning model, calculates a state value related to the state of the fingerprint of the target user, and outputs the state value.
• a computer program is recorded on at least one computer to execute a learning method, which includes inputting a first image including a first fingerprint into an image generation model, generating a second image including a second fingerprint in a state different from that of the first fingerprint, matching the first image with the second image, calculating a matching score indicating the degree of matching, and learning a machine learning model that represents the relationship between the second image and an output value based on the matching score for the second image.
• Another aspect of the recording medium of this disclosure is a fingerprint state estimation method that uses a machine learning model learned by a learning method, which inputs a first image including a first fingerprint into an image generation model to generate a second image including a second fingerprint in a state different from that of the first fingerprint, compares the first image with the second image to calculate a matching score indicating the degree of match, and learns a machine learning model that expresses the relationship between the second image and an output value based on the matching score for the second image, and a computer program is recorded on the recording medium to execute the fingerprint state estimation method, which obtains a fingerprint image of a target user, inputs the fingerprint image of the target user into the machine learning model, calculates a state value related to the state of the fingerprint of the target user, and outputs the state value.
• FIG. 2 is a block diagram showing the hardware configuration of the learning device according to the first embodiment.
• FIG. 1 is a block diagram showing a functional configuration of a learning device according to a first embodiment.
• 4 is a flowchart showing a flow of operations of the learning device according to the first embodiment.
• FIG. 11 is a plan view (part 1) showing an example of a second image generated by the learning device according to the second embodiment.
• FIG. 11 is a plan view (part 2) showing an example of a second image generated by the learning device according to the second embodiment.
• FIG. 13 is a block diagram showing the functional configuration of a learning device according to a third embodiment. 13 is a flowchart showing the flow of operations of a determination unit in the learning device according to the third embodiment.
• FIG. 11 is a plan view (part 1) showing an example of a second image generated by the learning device according to the second embodiment.
• FIG. 11 is a plan view (part 2) showing an example of a second image generated by the learning device according to the second embodiment
• FIG. 13 is a conceptual diagram showing a learning method of the learning device according to the fourth embodiment.
• FIG. 13 is a conceptual diagram showing a learning method of the learning device according to the fifth embodiment.
• 13 is a heat map showing the cumulative value of the matching score calculated by the learning device according to the fifth embodiment.
• FIG. 13 is a block diagram showing the functional configuration of a fingerprint state estimating device according to a sixth embodiment. 20 is a flowchart showing the flow of operations of the fingerprint state estimating device according to the sixth embodiment.
• FIG. 1 A learning device according to a first embodiment will be described with reference to FIGS. 1 to 3.
• FIG. 1 A learning device according to a first embodiment will be described with reference to FIGS. 1 to 3.
• FIG. 1 A learning device according to a first embodiment will be described with reference to FIGS. 1 to 3.
• FIG. 1 A learning device according to a first embodiment will be described with reference to FIGS. 1 to 3.
• FIG. 1 A learning device according to a first embodiment will be described with reference to FIGS. 1 to 3.
• Fig. 1 is a block diagram showing the hardware configuration of the learning device according to the first embodiment.
• the learning device 10 includes a processor 11, a RAM (Random Access Memory) 12, and a ROM (Read Only Memory) 13.
• the learning device 10 may further include a storage device 14, an input device 15, and an output device 16.
• the above-mentioned processor 11, RAM 12, ROM 13, storage device 14, input device 15, and output device 16 are connected via a data bus 17.
• the processor 11 reads a computer program.
• the processor 11 is configured to read a computer program stored in at least one of the RAM 12, the ROM 13, and the storage device 14.
• the processor 11 may read a computer program stored in a computer-readable storage medium using a storage medium reading device (not shown).
• the processor 11 may obtain (i.e., read) a computer program from a device (not shown) disposed outside the learning device 10 via a network interface.
• the processor 11 controls the RAM 12, the storage device 14, the input device 15, and the output device 16 by executing the computer program that the processor 11 reads.
• a processing block for learning a machine learning model related to fingerprint matching is realized within the processor 11. That is, the processor 11 may function as a controller that executes each control in the learning device 10.
• the processor 11 may be configured as, for example, a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), an FPGA (field-programmable gate array), a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), or a quantum processor.
• the processor 11 may be configured as one of these, or may be configured to use multiple processors in parallel.
• RAM 12 temporarily stores computer programs executed by processor 11.
• RAM 12 temporarily stores data that processor 11 uses temporarily while processor 11 is executing a computer program.
• RAM 12 may be, for example, a D-RAM (Dynamic Random Access Memory) or an SRAM (Static Random Access Memory). Also, other types of volatile memory may be used instead of RAM 12.
• ROM 13 stores computer programs executed by processor 11. ROM 13 may also store other fixed data. ROM 13 may be, for example, a P-ROM (Programmable Read Only Memory) or an EPROM (Erasable Read Only Memory). Also, other types of non-volatile memory may be used instead of ROM 13.
• the storage device 14 stores data that the learning device 10 stores long-term.
• the storage device 14 may operate as a temporary storage device for the processor 11.
• the storage device 14 may include, for example, at least one of a hard disk device, a magneto-optical disk device, an SSD (Solid State Drive), and a disk array device.
• the input device 15 is a device that receives input instructions from a user of the learning device 10.
• the input device 15 may include, for example, at least one of a keyboard, a mouse, and a touch panel.
• the input device 15 may be configured as a mobile terminal such as a smartphone or a tablet.
• the input device 15 may be, for example, a device that includes a microphone and is capable of voice input.
• the output device 16 is a device that outputs information related to the learning device 10 to the outside.
• the output device 16 may be a display device (e.g., a display) that can display information related to the learning device 10.
• the output device 16 may be configured as a mobile terminal such as a smartphone or a tablet.
• the output device 16 may also be a device that outputs information in a format other than an image.
• the output device 16 may be a speaker that outputs information related to the learning device 10 as audio.
• the learning device 10 may be configured to include only the above-mentioned processor 11, RAM 12, and ROM 13, and the other components (i.e., the storage device 14, the input device 15, and the output device 16) may be provided by an external device connected to the learning device 10.
• the other components i.e., the storage device 14, the input device 15, and the output device 16
• some of the calculation functions of the learning device 10 may be realized by an external device (e.g., an external server or cloud, etc.).
• Fig. 2 is a block diagram showing the functional configuration of the learning device according to the first embodiment.
• the learning device 10 is configured to include, as components for realizing its functions, a fingerprint image acquisition unit 110, a fingerprint image generation unit 120, a matching score calculation unit 130, and a learning unit 140.
• Each of the fingerprint image acquisition unit 110, the fingerprint image generation unit 120, the matching score calculation unit 130, and the learning unit 140 may be a processing block realized, for example, by the above-mentioned processor 11 (see FIG. 1).
• the fingerprint image acquisition unit 110 is configured to be able to acquire a first image including a first fingerprint.
• the fingerprint image acquisition unit 110 may directly acquire a fingerprint image captured by a scanner or the like as the first image, or may acquire a fingerprint image stored in a database or the like (i.e., a fingerprint image captured in the past) as the first image.
• the first image acquired by the fingerprint image acquisition unit 110 is configured to be output to each of the fingerprint image generation unit 120, the matching score calculation unit 130, and the learning unit 140.
• the fingerprint image acquisition unit 110 is not an essential component of the learning device 10 according to the first embodiment, and the learning device 10 according to the first embodiment may be configured without including the fingerprint image acquisition unit 110.
• the first image may be directly input to each of the fingerprint image generation unit 120, the matching score calculation unit 130, and the learning unit 140.
• the fingerprint image generating unit 120 is configured to be able to generate a second image including a second fingerprint from the first image acquired by the fingerprint image acquiring unit 110. Specifically, the fingerprint image generating unit 120 inputs the first image into an image generation model to generate a second image.
• the second fingerprint included in the second image is a fingerprint in a different state from the first fingerprint included in the first image.
• the "state" refers to characteristics that affect fingerprint matching, such as image quality and image size.
• the fingerprint image generating unit 120 may generate multiple second images from one first image.
• the image generation model may be, for example, a model that is machine-learned by inputting training data.
• the image generation model may be a neural network that is trained by deep learning. Specific examples of the image generation model will be described in detail in other embodiments described later.
• the second image generated by the fingerprint image generation unit 120 is configured to be output to each of the matching score calculation unit 130 and the learning unit 140.
• the matching score calculation unit 130 is configured to be able to calculate a matching score by matching a first image with a second image generated from the first image.
• the matching score is a score indicating the degree of matching between the first image and the second image.
• the matching score calculation unit 130 may calculate a higher matching score as the degree of matching (similarity of the images) is higher.
• the method of calculating the matching score is not particularly limited.
• the matching score calculation unit 130 may calculate the matching score by matching a feature amount extracted from the first image with a feature amount extracted from the second image.
• the matching score calculated by the matching score calculation unit 130 is configured to be output to the learning unit 140.
• the learning unit 140 is configured to be able to learn a machine learning model.
• the machine learning model is a model that represents the relationship between the second image and an output value based on the matching score for the second image.
• the "output value" here is a value that is determined according to the matching score, and may be a value calculated from the matching score or may be the matching score itself. Specific examples of the output value will be described in other embodiments described later.
• the machine learning model may be configured, for example, as a model including a neural network.
• the learning unit 140 learns the machine learning model using the first image acquired by the fingerprint image acquisition unit 110, the second image generated by the fingerprint image generation unit 120, and the matching score calculated by the matching score calculation unit 130. That is, the learning unit 140 may learn the machine learning model using training data including the first image, the second image, and an output value based on the matching score. At this time, the learning unit 140 may learn the output value based on the matching score as ground truth data.
• Fig. 3 is a flowchart showing the flow of operations of the learning device according to the first embodiment.
• the fingerprint image acquisition unit 110 first acquires a first image (step S101). Then, the fingerprint image generation unit 120 inputs the first image acquired by the fingerprint image acquisition unit 110 into an image generation model to generate a second image (step S102).
• the matching score calculation unit 130 compares the first image acquired by the fingerprint image acquisition unit 110 with the second image generated by the fingerprint image generation unit 120 to calculate a matching score (step S103). Note that if multiple second images have been generated from the first image, the matching score calculation unit 130 may calculate a matching score for each of the multiple second images.
• the learning unit 140 learns a machine learning model based on the first image, the second image, and the output value based on the matching score (step S104). That is, the learning unit 140 learns the machine learning model as a model that represents the relationship between the second image and the output value based on the matching score for the second image.
• a machine learning model is trained based on a first image including a first fingerprint, a second image including a second fingerprint, and a matching score between the first image and the second image. In this way, learning can be performed according to changes in the state of the fingerprint. Therefore, it is possible to achieve highly accurate fingerprint matching using a trained machine learning model.
• Second Embodiment The learning device 10 according to the second embodiment will be described with reference to Figures 4 and 5.
• the second embodiment describes an example of generation of a second image by the fingerprint image generating unit 120, and the device configuration and operation flow may be the same as those of the first embodiment. Therefore, in the following, the parts that differ from the first embodiment already described will be described in detail, and the explanation of other overlapping parts will be omitted as appropriate.
• Fig. 4 is a plan view (part 1) showing an example of a second image generated by the learning device according to the second embodiment.
• Fig. 5 is a plan view (part 2) showing an example of a second image generated by the learning device according to the second embodiment.
• the fingerprint image generation unit 120 generates the second image using a Generative Adversarial Network (GAN).
• GAN Generative Adversarial Network
• the fingerprint image generation unit 120 may generate the second image by applying image transformation such as distortion or noise to the first image, for example.
• the second image shown in Figure 4 is a second image to which a slap/rolled transformation has been applied.
• the second image shown in Figure 5 is a second image to which a texture transformation has been applied.
• the fingerprint image generation unit 120 may generate multiple second images by changing parameters (+, 0, -) related to the transformation.
• the second image is generated using a GAN. In this way, it is possible to appropriately generate a second image that includes a second fingerprint that is in a different state from the first fingerprint.
• the second image is generated using the GAN.
• the second image may be generated using other methods.
• the fingerprint image generating unit 120 The second image may be generated using a variational auto encoder (VAE), a flow-based model, a diffusion model, etc. In this case, the above-mentioned technical effects are also achieved accordingly.
• VAE variational auto encoder
• the learning device 10 according to the third embodiment will be described with reference to Figures 6 and 7.
• the third embodiment differs from the first and second embodiments in some configurations and operations, and other parts may be the same as the first and second embodiments. Therefore, the following will describe in detail the parts that differ from the embodiments already described, and will omit descriptions of other overlapping parts as appropriate.
• Fig. 6 is a block diagram showing the functional configuration of the learning device according to the third embodiment.
• the same reference numerals are used to designate the same elements as those described in Fig. 2.
• the learning device 10 according to the third embodiment is configured to include, as components for realizing its functions, a fingerprint image acquisition unit 110, a fingerprint image generation unit 120, a matching score calculation unit 130, a learning unit 140, and a judgment unit 150. That is, the learning device 10 according to the third embodiment further includes a judgment unit 150 in addition to the configuration of the first embodiment (see FIG. 2).
• the judgment unit 150 may be, for example, a processing block realized by the processor 11 described above.
• the determination unit 150 determines whether a sufficient number of second images have been generated based on the matching score calculated by the matching score calculation unit 130. Specifically, the determination unit 150 determines whether second images corresponding to the states of multiple fingerprints used for learning the machine learning model have been generated.
• the states of multiple fingerprints used for learning may be set in advance based on a learning method or the like.
• the second images used for learning may be set to images with output values of 20, 40, 60, and 80 based on the matching score. In this case, the determination unit 150 may determine whether all second images with output values of 20, 40, 60, and 80 based on the matching score are available. If all second images are not available, the determination unit 150 is configured to output an instruction to the fingerprint image generation unit 120 to generate a new second image.
• Fig. 7 is a flowchart showing the flow of operations of the determination unit in the learning device according to the third embodiment.
• the determination unit 150 of the learning device 10 first obtains the matching score calculated by the matching score calculation unit 130 (step S201). At this time, the determination unit 150 obtains the matching score corresponding to each of the multiple second images that have already been generated.
• the determination unit 150 determines whether or not all second images corresponding to the states of the multiple fingerprints used to train the machine learning model have been generated based on the acquired matching score. In other words, the determination unit 150 determines whether or not there are insufficient second images corresponding to the states of the multiple fingerprints used to train the machine learning model (step S202).
• step S202 If it is determined that a second image is missing (step S202: YES), the determination unit 150 outputs an instruction to the fingerprint image generation unit 120 to generate a new second image (step S203).
• This instruction may simply instruct the generation of a new second image, or may include information about the specific missing image (e.g., a matching score corresponding to the missing second image).
• step S202 if it is determined that there is no shortage of second images (step S202: NO), the processing of step S203 described above may be omitted. In other words, the determination unit 150 does not output an instruction to the fingerprint image generation unit 120 to generate a new second image.
• the learning device 10 when there are insufficient second images to be used for learning, an instruction to generate a new second image is output. In this way, it is possible to have an adequate number of images to be used for learning, making it possible to carry out learning that assumes various state changes.
• the learning device 10 according to the fourth embodiment will be described with reference to Figures 8 and 9.
• the fourth embodiment differs from the first to third embodiments in some of its operations, and other parts may be the same as the first to third embodiments. Therefore, the following will describe in detail the parts that differ from the embodiments already described, and will omit descriptions of other overlapping parts as appropriate.
• Fig. 8 is a conceptual diagram showing a learning method of the learning device according to the fourth embodiment.
• the fingerprint image generating unit 120 generates a second image having a different quality from the first image.
• the fingerprint image generating unit 120 may generate the second image, for example, by adding distortion, noise, or the like to the first image.
• the fingerprint image generating unit 120 may generate a plurality of second images having different qualities from each other. In the example shown in FIG. 8, three second images having different qualities are generated for one first image.
• the learning unit 140 calculates a matching score by matching the first image and the second image, which are different in quality as described above. Then, a quality value indicating the quality of the second image is calculated as an output value based on the matching score.
• the quality value here is a specific example of a value indicating the "state" of a fingerprint. In the example shown in FIG. 8, the quality value is calculated to be 25 for the second image having a matching score of 500. The quality value is calculated to be 50 for the second image having a matching score of 1000. The quality value is calculated to be 50 for the second image having a matching score of 2000. Note that here, the matching score and the quality value are calculated to be proportional, but the matching score and the quality value do not have to be proportional.
• the learning unit 140 learns a machine learning model using the calculated quality value as a teacher signal (correct answer data). That is, the learning unit 140 learns a machine learning model using a pair of a second image and a quality value corresponding to the second image as training data.
• Fig. 9 is a conceptual diagram showing an example of the operation of the machine learning model trained by the learning device according to the fourth embodiment.
• a fingerprint image is input to the machine learning model trained by the learning device 10 according to the fourth embodiment.
• the machine learning model outputs a quality value corresponding to the fingerprint image.
• a second image having a different quality from the first image is generated, and learning is performed using the quality value of the second image as correct answer data.
• learning can be performed using training data that is not affected by the quality of the fingerprint image.
• a machine learning model that can accurately estimate the quality of the fingerprint image can be generated.
• the learning device 10 according to the fifth embodiment will be described with reference to Figures 10 to 12.
• the fifth embodiment differs from the first to fourth embodiments in some of its operations, and other parts may be the same as the first to fourth embodiments. Therefore, the following will describe in detail the parts that differ from the embodiments already described, and will omit descriptions of other overlapping parts as appropriate.
• Fig. 10 is a conceptual diagram showing the learning method of the learning device according to the fifth embodiment.
• Fig. 11 is a heat map showing the cumulative value of the matching score calculated by the learning device according to the fifth embodiment.
• the fingerprint image generating unit 120 generates a second image having a different size (reduction ratio) from the first image.
• the reduction ratio here is a specific example of a value indicating the "state" of the fingerprint.
• the fingerprint image generating unit 120 may generate a second image having a different size, for example, by using an image generation model that enlarges or reduces the size of the first image.
• the fingerprint image generating unit 120 may generate a second image having a different size by using a GAN.
• the fingerprint image generating unit 120 may generate a second image that assumes the growth of the subject whose fingerprint is to be taken (for example, an image that estimates the fingerprint several years later).
• the learning unit 140 calculates a matching score by matching the first image and the second image, which are different in size as described above. Specifically, the learning unit 140 calculates a matching score by matching the first image and the second image, which are different in reduction ratio. Then, the learning unit 140 learns the reduction ratio that maximizes the matching score as a teacher signal.
• a process T for reducing an image is performed on a first image Ip to generate an image Ip ⁇ with a different reduction ratio ⁇ .
• a process T for reducing an image is performed on a second image Ig to generate an image Ig ⁇ with a different reduction ratio ⁇ .
• ⁇ min which is the lower limit of the reduction ratio ⁇
• ⁇ max which is the upper limit
• ⁇ min which is the lower limit of the reduction ratio ⁇
• ⁇ max which is the upper limit
• the learning unit 140 uses p(argmax) that has the highest accumulated value in this heat map as a teacher signal.
• Fig. 12 is a conceptual diagram showing an example of the operation of the machine learning model trained by the learning device according to the fifth embodiment.
• a fingerprint image is input to the machine learning model trained by the learning device 10 according to the fifth embodiment.
• the machine learning model outputs a reduction ratio corresponding to the fingerprint image.
• a second image having a different size from the first image is generated, and learning is performed in which the reduction ratio of the second image that maximizes the matching score is used as the correct answer data.
• the reduction ratio that maximizes the matching score can be constructed as training data. Therefore, it is possible to generate a model that estimates the reduction ratio of the second image that is suitable for matching.
• a fingerprint state estimating device will be described with reference to Fig. 13 and Fig. 14. Note that the fingerprint state estimating device according to the sixth embodiment may have the same hardware configuration (see Fig. 1) as the learning device 10 according to the first to fifth embodiments described above.
• Fig. 13 is a block diagram showing the functional configuration of a fingerprint state estimating device according to the sixth embodiment.
• the fingerprint state estimation device 50 is configured to include, as components for realizing its functions, a fingerprint acquisition unit 210, a state value calculation unit 220, and an output unit 230.
• a fingerprint acquisition unit 210 the fingerprint acquisition unit 210
• a state value calculation unit 220 the state value calculation unit 220
• an output unit 230 the output unit 230 may be realized, for example, by the processor 11 described above.
• the fingerprint acquisition unit 210 is configured to be able to acquire a fingerprint image of the target user.
• the fingerprint acquisition unit 210 acquires a fingerprint image scanned by, for example, a fingerprint scanner.
• the fingerprint acquisition unit 210 may have the same functions as the fingerprint image acquisition unit 110 in the learning device 10 already described.
• the fingerprint image acquired by the fingerprint acquisition unit 210 is configured to be output to the state calculation unit 220.
• the fingerprint acquisition unit 210 is not an essential component of the fingerprint state estimation device 50 according to the sixth embodiment, and the fingerprint state estimation device 50 according to the sixth embodiment may be configured without including the fingerprint acquisition unit 210. In this case, the fingerprint image may be directly input to the state calculation unit 220.
• the state calculation unit 220 includes a machine learning model 225 that has been trained by the learning device 10 described above.
• the state calculation unit 220 inputs the fingerprint image acquired by the fingerprint acquisition unit 210 into the machine learning model 225, and calculates a state value of the fingerprint of the target user.
• the "state value” here is a value that indicates the state of the fingerprint, and examples include the quality value described in the fourth embodiment and the reduction ratio described in the fifth embodiment.
• the output unit 230 is configured to be able to output the state value calculated by the state calculation unit 220.
• the output unit 230 may be configured to output the state value to an output device such as a display. That is, the output unit 230 may have a function of presenting the state value to a user.
• the output unit 230 may be configured to output the state value to an external fingerprint matching device. In this case, the fingerprint matching device may perform fingerprint matching according to the state of the fingerprint image.
• Fig. 14 is a flowchart showing the flow of operations of the fingerprint state estimating device according to the sixth embodiment.
• the fingerprint acquisition unit 210 first acquires a fingerprint image of the target user (step S601). After that, the state calculation unit 220 inputs the fingerprint image acquired by the fingerprint acquisition unit 210 to the machine learning model 225 and calculates a state value of the fingerprint of the target user (step S602). Next, the output unit 230 outputs the state value calculated by the state calculation unit 220 (step S603).
• a state value indicating the state of a fingerprint image is calculated using machine learning learned by the learning device 10. In this way, information regarding the state of a fingerprint image can be appropriately estimated from the fingerprint image of the target user.
• each embodiment also includes a processing method in which a program that operates the configuration of each embodiment to realize the functions of the above-mentioned embodiments is recorded on a recording medium, the program recorded on the recording medium is read as code, and executed on a computer.
• computer-readable recording media are also included in the scope of each embodiment.
• each embodiment includes not only the recording medium on which the above-mentioned program is recorded, but also the program itself.
• the recording medium may be, for example, a floppy disk, hard disk, optical disk, magneto-optical disk, CD-ROM, magnetic tape, non-volatile memory card, or ROM.
• the scope of each embodiment is not limited to programs recorded on the recording medium that execute processes by themselves, but also includes programs that operate on an OS in conjunction with other software or the functions of an expansion board to execute processes.
• the program itself may be stored on a server, and part or all of the program may be made downloadable from the server to a user terminal.
• the program may be provided to the user in, for example, a SaaS (Software as a Service) format.
• the learning device described in Appendix 1 is a learning device that includes a generation means that inputs a first image including a first fingerprint into an image generation model and generates a second image including a second fingerprint in a state different from that of the first fingerprint, a calculation means that compares the first image with the second image and calculates a matching score indicating a degree of matching, and a learning means that learns a machine learning model that expresses the relationship between the second image and an output value based on the matching score for the second image.
• the learning device according to Supplementary Note 2 is the learning device according to Supplementary Note 1, wherein the generating means generates the second image using a generative adversarial network.
• the learning device described in Appendix 3 is the learning device described in Appendix 1 or 2, further comprising a determination means for determining whether or not the second images corresponding to the states of multiple fingerprints used for training the machine learning model have been generated based on the matching score calculated by the calculation means, and the generation means generates new second images when there are insufficient second images corresponding to the states of the multiple fingerprints.
• the learning device described in Supplementary Note 4 is the learning device described in any one of Supplementary Notes 1 to 3, wherein the generation means generates the second image so that the fingerprint quality is different between the first fingerprint and the second fingerprint, and the learning means trains the machine learning model so as to output a quality score indicating the quality of the fingerprint as the output value.
• the learning device described in Supplementary Note 5 is the learning device described in any one of Supplementary Notes 1 to 4, wherein the generation means generates the second image so that the first fingerprint and the second fingerprint have different fingerprint sizes, and the learning means trains the machine learning model to output, as the output value, a reduction rate corresponding to the size of the second image at which the matching score is maximized.
• the fingerprint state estimation device described in Supplementary Note 6 is a fingerprint state estimation device that uses a machine learning model trained by a learning device that includes: a generation means for inputting a first image including a first fingerprint into an image generation model and generating a second image including a second fingerprint whose state is different from that of the first fingerprint; a calculation means for matching the first image with the second image and calculating a matching score indicating a degree of matching; and a learning means for learning a machine learning model that expresses a relationship between the second image and an output value based on the matching score for the second image.
• the fingerprint state estimation device includes: a fingerprint acquisition means for acquiring a fingerprint image of a target user; a state value calculation means for inputting the fingerprint image of the target user into the machine learning model and calculating a state value related to the state of the fingerprint of the target user; and an output means for outputting the state value.
• the learning method described in Supplementary Note 7 is a learning method that, by at least one computer, inputs a first image including a first fingerprint into an image generation model, generates a second image including a second fingerprint in a state different from that of the first fingerprint, matches the first image with the second image, calculates a matching score indicating a degree of matching, and learns a machine learning model that represents a relationship between the second image and an output value based on the matching score for the second image.
• the fingerprint state estimation method described in Appendix 8 is a fingerprint state estimation method using a machine learning model learned by a learning method, which includes at least one computer: inputting a first image including a first fingerprint into an image generation model; generating a second image including a second fingerprint having a state different from that of the first fingerprint; matching the first image with the second image; calculating a matching score indicating a degree of matching; and learning a machine learning model that expresses a relationship between the second image and an output value based on the matching score for the second image.
• the fingerprint state estimation method obtains a fingerprint image of a target user, inputs the fingerprint image of the target user into the machine learning model, calculates a state value related to the state of the fingerprint of the target user, and outputs the state value.
• the computer program described in Supplementary Note 9 is a computer program that causes at least one computer to execute a learning method of inputting a first image including a first fingerprint into an image generation model to generate a second image including a second fingerprint in a state different from that of the first fingerprint, matching the first image with the second image to calculate a matching score indicating a degree of matching, and learning a machine learning model that expresses a relationship between the second image and an output value based on the matching score for the second image.
• the computer program described in Supplementary Note 10 is a fingerprint state estimation method that uses a machine learning model learned by a learning method, the machine learning model learning a relationship between the second image and an output value based on the matching score for the second image, and causes at least one computer to execute the fingerprint state estimation method, the machine learning model learning a first image including a first fingerprint into an image generation model to generate a second image including a second fingerprint in a state different from that of the first fingerprint, matching the first image with the second image to calculate a matching score indicating a degree of matching, and learning a machine learning model that expresses a relationship between the second image and an output value based on the matching score for the second image, the computer program acquiring a fingerprint image of a target user, inputting the fingerprint image of the target user into the machine learning model, calculating a state value related to the state of the fingerprint of the target user, and outputting the state value.
• the recording medium described in Appendix 11 is a recording medium having recorded thereon a computer program for causing at least one computer to execute a learning method, the learning method including inputting a first image including a first fingerprint into an image generation model to generate a second image including a second fingerprint in a state different from that of the first fingerprint, matching the first image with the second image to calculate a matching score indicating a degree of matching, and learning a machine learning model that expresses a relationship between the second image and an output value based on the matching score for the second image.
• the recording medium described in Supplementary Note 12 is a recording medium on which a computer program is recorded to cause at least one computer to execute a fingerprint state estimation method using a machine learning model learned by a learning method, the machine learning model including: inputting a first image including a first fingerprint into an image generation model to generate a second image including a second fingerprint having a state different from that of the first fingerprint, matching the first image with the second image to calculate a matching score indicating a degree of matching, and learning a machine learning model that expresses a relationship between the second image and an output value based on the matching score for the second image, the recording medium having a computer program recorded thereon to cause the fingerprint state estimation method to execute the fingerprint state estimation method, the machine learning model including: acquiring a fingerprint image of a target user, inputting the fingerprint image of the target user into the machine learning model to calculate a state value related to the state of the fingerprint of the target user, and outputting the state value.

Landscapes

• Engineering & Computer Science (AREA)
• Physics & Mathematics (AREA)
• General Physics & Mathematics (AREA)
• Theoretical Computer Science (AREA)
• Human Computer Interaction (AREA)
• Multimedia (AREA)
• Computer Vision & Pattern Recognition (AREA)
• Collating Specific Patterns (AREA)
• Image Analysis (AREA)

Priority Applications (1)

Applications Claiming Priority (2)

Publications (1)

Family

ID=92674412

Family Applications (1)

Country Status (2)

Citations (2)

• 2024
  • 2024-01-24 JP JP2025505113A patent/JPWO2024185330A1/ja active Pending
  • 2024-01-24 WO PCT/JP2024/002051 patent/WO2024185330A1/ja not_active Ceased

Patent Citations (2)

Also Published As

Similar Documents

Legal Events