Hand shape recognition device, hand shape recognition method, and hand shape recognition program

Abstract

This hand shape identification device has a measuring wearable body unit, a signal generation / processing unit, a hand shape inference unit, and an identification unit. The measuring wearable body unit includes a vibration production / acquisition unit capable of producing and acquiring vibrations, and makes it possible for the vibration production / acquisition unit to be worn, with good positioning reproducibility, on the back of a measurement subject's hand. The signal generation / processing unit generates a vibration production signal for causing the vibration production / acquisition unit to produce vibrations, transmits the vibration production signal to the vibration production / acquisition unit, receives and processes a vibration signal indicating vibrations which have passed through the hand, said vibrations having been acquired by the vibration production / acquisition unit, and outputs the processed signal. The hand shape inference unit generates a trained model that uses machine learning to infer a hand shape from the processed signal. The identification unit uses the trained model to identify hand shapes.

Description

Technical Field

[0001] The present invention relates to a hand shape recognition device, a hand shape recognition method, and a hand shape recognition program.

Background Art

[0002] There is a method of using a pair of sensors attached to the back of the hand with double-sided tape, acquiring an acoustic signal irradiated from one side by the other sensor, and distinguishing the hand shape based on the displacement of the acquired signal (Non-Patent Document 1).

Prior Art Documents

Non-Patent Documents

[0003]

Non-Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, the method of Non-Patent Document 1 does not inhibit finger movement in order to attach the sensor to the back of the hand, but uses double-sided tape for the propagation of the acoustic signal between the sensor and the living body, and an error occurs in the attachment position every time it is attached, making it difficult for the user to easily attach it.

[0005] This invention has been made in view of the above circumstances, and its main objective is to provide a hand shape recognition device that does not interfere with the user's finger movements, allows the user to attach sensors to the same position even unintentionally, and identifies the shape of the hand. [Means for solving the problem]

[0006] The hand shape recognition device according to this embodiment comprises a measurement device, a signal generation / processing unit, a hand shape estimation unit, and an identification unit. The measurement device includes a vibration generation / acquisition unit capable of generating and acquiring vibrations, and enables the vibration generation / acquisition unit to be attached to the back of the hand being measured with good positional reproducibility. The signal generation / processing unit generates a vibration generation signal to generate vibrations in the vibration generation / acquisition unit, transmits the vibration generation signal to the vibration generation / acquisition unit, and receives and processes the vibration signal representing the vibrations that have passed through the hand acquired by the vibration generation / acquisition unit, and outputs a processed signal. The hand shape estimation unit generates a learning model that estimates the hand shape using machine learning from the processed signal. The identification unit identifies the hand shape using the learning model. The vibration generation and acquisition unit has two piezoelectric elements capable of generating and acquiring arbitrary vibrations. The measurement device unit, in addition to the vibration generation and acquisition unit, has a housing reinforcement unit, a bio-adhesive unit, an external force absorption unit, a device unit, and a device unit fixing unit. The housing reinforcement unit is fixed to the piezoelectric element and reinforces it. The bio-adhesive unit is fixed to the piezoelectric element on the opposite side of the housing reinforcement unit and is in close contact with the back of the hand. The external force absorption unit covers and protects the bio-adhesive unit, the piezoelectric element, and the housing reinforcement unit from the side of the housing reinforcement unit, and absorbs external forces applied to the bio-adhesive unit, the piezoelectric element, and the housing reinforcement unit. The device unit has a pocket that accommodates the bio-adhesive unit, the piezoelectric element, the housing reinforcement unit, and the external force absorption unit with the bio-adhesive unit exposed, and can be positioned to cover the back of the hand. The device unit fixing unit is attached to the device unit and fixes the device unit that covers the back of the hand to the hand. [Effects of the Invention]

[0007] According to the present invention, a hand shape recognition device is provided that does not interfere with the user's finger movements, allows the user to attach sensors to the same position even unintentionally, and identifies the shape of the hand. [Brief explanation of the drawing]

[0008] [Figure 1] Figure 1 shows an example of the configuration of a hand shape recognition device according to an embodiment. [Figure 2] Figure 2 is a functional block diagram of the measurement device unit, signal generation / analysis unit, finger shape estimation unit, and identification unit shown in Figure 1. [Figure 3] Figure 3 shows the hardware configuration of the information processing equipment that constitutes the signal generation / analysis unit, finger shape estimation unit, and identification unit in Figure 1. [Figure 4] Figure 4 is a perspective view showing the piezoelectric element and the housing reinforcement section. [Figure 5]Figure 5 is a plan view of the external force absorption section that covers and protects the bio-adhesion section, the piezoelectric element, and the housing reinforcement section. [Figure 6] Figure 6 schematically shows the cross-sectional structure of the integrated bio-adhesive portion, piezoelectric element, housing reinforcement portion, and external force absorption portion. [Figure 7] Figure 7 is a perspective view of the external force absorption section. [Figure 8] Figure 8 is a plan view of the device body and the device fixation part as seen from the outside. [Figure 9] Figure 9 is a plan view of the device body and the device fixation part as seen from the outside. [Figure 10] Figure 10 shows the device attached to the hand. [Figure 11] Figure 11 shows the part that secures the device to the device. [Figure 12] Figure 12 is a cross-sectional perspective view of the bio-adhesion area and the external force absorption area. [Figure 13] Figure 13 is a cross-sectional view of the bio-adhesion section, piezoelectric element, housing reinforcement section, and external force absorption section. [Figure 14] Figure 14 is a flowchart showing the operation flow of the hand shape recognition device. [Modes for carrying out the invention]

[0009] Embodiments of the present invention will be described below with reference to the drawings.

[0010] (Configuration of the hand shape recognition device) Figure 1 shows an example of the configuration of the hand shape recognition device 10 according to the embodiment. The hand shape recognition device 10 includes a measurement device unit 20, a signal generation / processing unit 30, a hand shape estimation unit 40, an identification unit 50, and a database 60. Figure 2 shows a functional block diagram of the measurement device unit 20, the signal generation / processing unit 30, the hand shape estimation unit 40, and the identification unit 50.

[0011] (Measurement device unit 20) The body part 20 for measurement is provided with a vibration generation / acquisition unit 21 that applies vibration to the hand of the measurement target and acquires the vibration that has passed through the hand of the measurement target, and is an element for mounting the vibration generation / acquisition unit 21 on the hand of the measurement target with good position reproducibility.

[0012] As shown in FIG. 2, the body part 20 for measurement has six functional blocks: a vibration generation / acquisition unit 21, a biological adhesion part 22, a housing reinforcement part 23, an external force absorption part 24, a mounting body part 25, and a mounting body fixing part 26.

[0013] The vibration generation / acquisition unit 21 can generate and acquire vibration. For example, as shown in FIGS. 8 to 10, the vibration generation / acquisition unit 21 has a pair of piezoelectric elements 21a capable of generating and acquiring arbitrary vibration. One of the first piezoelectric elements 21a receives a vibration generation signal from the signal generation / processing unit 30 and generates vibration having the same frequency characteristics as the vibration generation signal. The other second piezoelectric element 21a acquires the vibration generated from the first piezoelectric element 21a and passing through the hand of the measurement target. The second piezoelectric element 21a outputs a vibration signal representing the acquired vibration.

[0014] Here, an example of the piezoelectric element 21a is given as an element that receives a vibration generation signal and generates vibration, and an element that acquires vibration and outputs a vibration signal. However, the form is not limited as long as it has such a function. The form of connecting the vibration generation / acquisition unit 21 and the signal generation / processing unit 30 is not limited as long as it has a function of transmitting and receiving an electrical signal between the two.

[0015] The housing reinforcement part 23 is shown in FIGS. 4, 6, and 13. In order to use the piezoelectric element 21a continuously for a long time, the housing reinforcement part 23 is fixed to the piezoelectric element 21a. For example, the housing reinforcement part 23 has a recess 23a in which the piezoelectric element 21a just fits, and the piezoelectric element 21a is fixed in the recess 23a with an adhesive. The housing reinforcement part 23 reinforces the strength of the fixed piezoelectric element 21a. For example, the housing reinforcement part 23 is composed of acrylic, plastic parts printed by 3D printing, and the like.

[0016] The bio-adhesive portion 22 is shown in Figures 6 and 13. The bio-adhesive portion 22 is fixed to the piezoelectric element 21a on the opposite side of the housing reinforcement portion 23, for example, by adhesive. The bio-adhesive portion 22 is in close contact with the back of the hand being measured during measurement. Vibrations generated from the first piezoelectric element 21a of the vibration generation / acquisition unit 21 are transmitted to the back of the hand via the bio-adhesive portion 22, and vibrations that have passed through the back of the hand are transmitted to the second piezoelectric element 21a of the vibration generation / acquisition unit 21 via the bio-adhesive portion 22. The bio-adhesive portion 22 is made of a material with high frictional force and high elasticity. This prevents gaps from forming between the bio-adhesive portion 22 and the back of the hand, even if the shape of the back of the hand changes due to the movement of the hand and fingers. The bio-adhesive portion 22 is made of, for example, human skin gel. The bio-adhesive portion 22 has a thickness T suitable for vibration propagation. For example, the thickness T is 2 mm.

[0017] The external force absorbing section 24 is shown in Figures 5-7, 12, and 13. The external force absorbing section 24 covers and protects the bio-adhesive section 22, piezoelectric element 21a, and housing reinforcement section 23, which are fixed to each other, from the side of the housing reinforcement section 23, and absorbs external forces applied to the bio-adhesive section 22, piezoelectric element 21a, and housing reinforcement section 23. For this reason, the external force absorbing section 24 is made of an elastic material. The external force absorbing section 24 has a recess 24a that exposes the bio-adhesive section 22 and accommodates the bio-adhesive section 22, piezoelectric element 21a, and housing reinforcement section 23. The external force absorbing section 24 also has a protrusion 24b that presses against the housing reinforcement section 23. Furthermore, the external force absorbing section 24 has a hole 24c through which a conductor 21b extending from the piezoelectric element 21a passes.

[0018] The attachment part 25 is shown in Figures 8 to 10. The attachment part 25 is an element that can be positioned to cover the back of the hand. The attachment part 25 has a hole 25a for passing the thumb through. The attachment part 25 also has a pocket 25b that houses the integrated bio-adhesive part 22, piezoelectric element 21a, housing reinforcement part 23, and external force absorption part 24, with the bio-adhesive part 22 exposed. The external force absorption part 24 covers the bio-adhesive part 22, piezoelectric element 21a, and housing reinforcement part 23, and is the only component that touches the attachment part 25 within the pocket 25b. It prevents displacement of the piezoelectric element 21a and also plays a role in making the adhesion position of the bio-adhesive part 22 to the back of the hand more unique and accurate.

[0019] The device fixing part 26 is shown in Figures 8, 9, and 11. The device fixing part 26 is attached to the device body 25 and is an element for fixing the device body 25, which covers the back of the hand, to the hand. The device fixing part 26 has a hook-and-loop fastener 26b fixed to one end of the device body 25 and a slit member 26a fixed to the other end of the device body 25, which has a slit for the hook-and-loop fastener 26b to pass through. The device body 25 is placed on the back of the hand, and the hook-and-loop fastener 26b crosses the palm side, passes through the slit of the slit member 26a, and is folded back, thereby the device fixing part 26 fixes the device body 25 to the back of the hand. The folded position of the hook-and-loop fastener 26b is adjustable, so that the user can easily adjust the fit of the device body 25 to fit the size of their hand. In other words, the user can wear the device body 25 on their hand with an appropriate fit. Furthermore, the attachment body 25 and the attachment body fixing part 26 do not obstruct the movement of the fingers.

[0020] (Signal generation / processing unit 30) The signal generation and processing unit 30 generates a vibration generation signal that generates vibrations in the vibration generation and acquisition unit 21, and transmits the vibration generation signal to the vibration generation and acquisition unit 21. The signal generation and processing unit 30 also receives and processes the vibration signal representing the vibrations that have passed through the hand, which has been acquired by the vibration generation and acquisition unit 21, and outputs a processed signal. For example, the vibration generation signal is a signal having arbitrary frequency characteristics in the ultrasonic band. The signal generation and processing unit 30 is composed of an information processing device such as a personal computer. The signal generation and processing unit 30 has four functional blocks: a signal generation unit 31, a signal receiving unit 32, a signal amplification unit 33, and a signal extraction unit 34.

[0021] The signal generation unit 31 generates a vibration generation signal that causes vibration to be generated in the vibration generation / acquisition unit 21, based on arbitrarily set parameters. The arbitrarily set parameters are stored in the database 60. The signal generation unit 31 transmits the generated vibration generation signal to the vibration generation / acquisition unit 21.

[0022] The signal receiving unit 32 receives a vibration signal representing the vibration that has passed through the hand, which has been acquired by the vibration generation / acquisition unit 21. The signal receiving unit 32 outputs the received vibration signal to the signal amplification unit 33.

[0023] The signal amplification unit 33 amplifies the vibration signal input from the signal amplification unit 33. The signal amplification unit 33 outputs the amplified vibration signal to the signal extraction unit 34.

[0024] The signal extraction unit 34 extracts the amplified vibration signal input from the signal amplification unit 33 at regular intervals and outputs this as a processed signal to the hand shape estimation unit 40 and the database 60.

[0025] (Hand shape estimation unit 40) The hand shape estimation unit 40 acquires a processing signal from the signal generation / processing unit 30 and generates a learning model that estimates the hand shape using machine learning based on the processing signal. Here, the hand shape consists of the state of the fingers, hand, and wrist. The hand shape estimation unit 40 is composed of an information processing device such as a personal computer. The hand shape estimation unit 40 has two functional blocks: a feature generation unit 41 and a learning unit 42.

[0026] The feature generation unit 41 acquires the processed signal output from the signal extraction unit 34 and generates features for identification based on the processed signal (for example, its waveform). The feature generation unit 41 also generates training data consisting of a pair of features and the ID of the hand shape label assigned to the features.

[0027] The learning unit 42 generates and trains an analysis model using training data consisting of pairs of features generated by the feature generation unit 41 and IDs of hand shape labels.

[0028] The identification unit 50 identifies the hand shape using the analysis model learned by the hand shape estimation unit 40. The identification unit 50 is composed of an information processing device such as a personal computer. The identification unit 50 has two functional blocks: an identification determination unit 51 and a determination result evaluation unit 52.

[0029] The identification determination unit 51 acquires the feature quantities obtained by the hand shape estimation unit 40 and the learned analysis model, and uses the feature quantities and the learned analysis model to determine a numerical value for determining the hand shape.

[0030] The judgment result evaluation unit 52 identifies the hand shape based on the numerical value obtained by the identification judgment unit 51.

[0031] (Database 60) The database 60 stores data required by the signal generation / processing unit 30, the hand shape estimation unit 40, and the identification unit 50, as well as data generated by the signal generation / processing unit 30, the hand shape estimation unit 40, and the identification unit 50. For example, the database 60 stores parameters of the signal generation unit 31, processed signals, feature quantities, IDs of hand shape labels, models, model parameters, numerical values ​​for determining hand shapes, and hand shape identification results.

[0032] (Hardware configuration of information processing equipment) Next, we will describe the hardware configuration of the information processing device 70, which comprises the signal generation / processing unit 30, the hand shape estimation unit 40, and the identification unit 50.

[0033] Figure 3 shows the hardware configuration of the information processing device 70. The information processing device 70 includes a processor 71, a ROM (Read Only Memory) 72, a RAM (Random Access Memory) 73, an auxiliary storage device 74, and an input / output interface 75.

[0034] The processor 71, ROM 72, RAM 73, auxiliary storage device 74, and input / output interface 75 are electrically connected to each other via a bus 76, and data is exchanged via the bus 76.

[0035] The processor 71 is composed of a general-purpose hardware processor, such as a CPU (Central Processing Unit) or a GPU (Graphical Processing Unit). The processor 71 controls the entirety of the ROM 72, RAM 73, auxiliary storage device 74, and input / output interface 75.

[0036] ROM72 is a non-volatile memory that constitutes part of the main memory. ROM72 non-temporarily stores the startup program required when the processor 71 starts up. The processor 71 starts up by executing the program in ROM72. ROM72 is composed of, for example, EPROM (Erasable Programmable Read Only Memory) and stores various startup settings in addition to the startup program.

[0037] RAM73 is a volatile memory that constitutes part of the main memory. RAM73 temporarily stores the program necessary for processing by the processor 71 and the data necessary for executing the program. The processor 71 executes the program in RAM73, performs calculations on the data in RAM73, and stores the calculation results in RAM73.

[0038] The auxiliary storage device 74 consists of non-volatile memory such as an HDD (Hard Disk Drive) or SSD (Solid State Drive). The auxiliary storage device 74 non-temporarily stores programs executed by the processor 71 and the data necessary for program execution. The processor 71 reads the programs and data from the auxiliary storage device 74 into the RAM 73 and executes various functions by running the programs. The auxiliary storage device 74 (and RAM 73) also constitute the database 60.

[0039] The input / output interface 75 is connected to an external input device 81 and an output device 82, etc., enabling the input of information from the input device 81 and the output of information to the output device 82. For example, the input / output interface 75 may be a wired interface or a wireless interface. A wired interface includes a port to which the device is connected. A wireless interface includes Bluetooth®, WiFi®, etc.

[0040] The input device 81 may include a keyboard, mouse, touch panel, receiver, disk drive, etc. However, the input device 81 is not limited to these and may include any other input device. The output device 82 may include a display, transmitter, disk drive, etc. However, the output device 82 is not limited to these and may include any other output device. The input device 81 and the output device 82 may be configured as an input / output device 83 that has the functions of both.

[0041] Programs stored non-temporarily in the auxiliary storage device 74 are provided to the information processing device 70, for example, via a recording medium 84 that is readable by the information processing device 70 on which the program is stored non-temporarily. Such a recording medium 84 is called a non-temporarily computer-readable recording medium. Non-temporarily computer-readable recording media include disks such as flexible disks, optical disks (CD-ROM, CD-R, DVD-ROM, DVD-R, etc.), magneto-optical disks (MO, etc.), and semiconductor memory.

[0042] The program stored non-temporarily in the auxiliary storage device 74 includes a hand shape identification program. The hand shape identification program is a program that causes the information processing device 70 to execute at least some of the functions of the signal generation / processing unit 30, the hand shape estimation unit 40, and the identification unit 50.

[0043] A program stored non-temporarily in the auxiliary storage device 74 is read into and stored non-temporarily via the input device 81, which is a disk drive, and the input / output interface 75, if the recording medium 84 is a disk, or via the input / output interface 75, which is a port, if the recording medium 84 is semiconductor memory. Alternatively, the program may be stored on a server on a network, downloaded from the server, and stored non-temporarily in the auxiliary storage device 74.

[0044] When the information processing device 70 is started, the processor 71 executes a program in the ROM 72 and loads the OS into the RAM 73 to start up. Under the control of the OS, the processor 71 monitors instruction inputs and the connection of external devices. Also, under the control of the OS, the processor 71 sets up a program area and a data area in the RAM 73. In response to the instruction input to start up the information processing device 70, the processor 71 loads the hand shape recognition program from the auxiliary storage device 74 into the program area of ​​the RAM 73, and loads the data necessary for executing the hand shape recognition program from the auxiliary storage device 74 into the data area of ​​the RAM 73. The processor 71 calculates the data in the data area according to the hand shape recognition program and writes the calculation result to the data area. Through these operations, the processor 71, RAM 73, auxiliary storage device 74, input / output interface 75, and bus 76 work together to perform at least some of the functions of the components of the signal generation / processing unit 30, the hand shape estimation unit 40, and the identification unit 50.

[0045] (Example of operation) Next, the operation of the hand shape recognition device 10 will be explained. A flowchart showing the operation flow of the hand shape recognition device 10 is shown in Figure 14.

[0046] First, as a preparation step, the user places the integrated bio-adhesive part 22, piezoelectric element 21a, housing reinforcement part 23, and external force absorption part 24 into the pocket 25b and adjusts it so that the bio-adhesive part 22 adheres to the back of the hand when the device body 25 is attached to the hand.

[0047] Next, the user inserts their thumb through the thumb hole 25a of the device body 25, and then uses the device body fixing part 26 to adjust the length up to the fold of the hook fastener 26b to fit the size of their hand and secures it, thereby attaching the device body 25 to their hand.

[0048] In step S1, the signal generation unit 31 of the signal generation / processing unit 30 generates a vibration generation signal that generates vibrations in the vibration generation / acquisition unit 21 based on arbitrarily set parameters. For example, the signal generation unit 31 generates a signal in the ultrasonic frequency band that sweeps from 20kHz to 40kHz. The settings of the vibration generation signal, such as whether to sweep or use other frequency bands, are not specified. The signal generation unit 31 transmits the generated vibration generation signal to the vibration generation / acquisition unit 21.

[0049] In step S2, the first piezoelectric element 21a of the vibration generation / acquisition unit 21 receives the vibration generation signal and generates vibrations having the same frequency characteristics as the vibration generation signal. As a result, the vibration generation / acquisition unit 21 applies vibrations to the user's hand. The vibrations at this time may contain other frequencies, as long as they include frequencies included in the vibrations used to generate the feature quantities contained in the registered hand shape registered in the database 60.

[0050] In step S2, the second piezoelectric element 21a of the vibration generation / acquisition unit 21 acquires the vibration generated from the first piezoelectric element 21a and passing through the hand being measured. As the vibration applied from one of the first piezoelectric elements 21a propagates to the other pair of second piezoelectric elements 21a, the user's hand functions as a propagation path, and the frequency characteristics of the applied vibration change according to this propagation path. The second piezoelectric element 21a transmits a vibration signal representing the acquired vibration to the signal receiving unit 32 of the signal generation / processing unit 30.

[0051] In step S4, the signal receiving unit 32 receives the vibration signal transmitted from the second piezoelectric element 21a. The signal receiving unit 32 outputs the received vibration signal to the signal amplification unit 33.

[0052] In step S5, the signal amplification unit 33 amplifies the vibration signal input from the signal amplification unit 33. This is because vibrations that pass through the hand are attenuated, and therefore need to be amplified to a level that can be processed. The signal amplification unit 33 outputs the amplified vibration signal to the signal extraction unit 34.

[0053] In step S6, the amplified vibration signals input from the signal amplification unit 33 are extracted at regular intervals and output as processed signals to the hand shape estimation unit 40 and the database 60. The number of signal samples is not limited.

[0054] In step S7, the feature generation unit 41 of the hand shape estimation unit 40 acquires a processing signal and generates training data based on the processing signal (for example, its waveform), which consists of a pair of feature quantities representing the vibration frequency characteristics of the hand and an ID of the hand shape label assigned to the feature quantity. The training data may also be generated by extracting it from a registered database that has been created and stored in the database 60 in advance.

[0055] In step S8, the learning unit 42 generates and learns an analysis model in which the input is the features obtained from the feature generation unit 41 and the output is an ID representing the hand shape as a label. The classification model obtained through this learning process, or the model itself or the parameters of the model, are registered in the database 60.

[0056] The type of classification model and the library used for its training are not restricted, as long as it is possible to train the model to obtain the optimal output by performing parameter tuning on the training data. For example, a commonly known machine learning library may be used, and algorithms for generating classification models such as Support Vector Machines (SVMs) or neural networks may be trained to obtain the optimal output by performing parameter tuning on the training data.

[0057] For example, in the learning unit 42, if SVM is used as the algorithm for model generation, a score indicating the similarity of each label in the analysis model to the input is output. For example, if the similarity is normalized and expressed between 0 and 1, it can be output as "1-(similarity)".

[0058] For example, in the learning unit 42, if Random Forest is used as the algorithm for generating a classification model, data is randomly extracted from the training data and multiple decision trees are generated. The number of decisions made for each label in each decision tree for the input data is output. A higher number of decisions is considered better, so "(number of decisions) - (number of decisions)" is output as a reference value.

[0059] Furthermore, other classification algorithms such as DNNs may be used. In that case, the reference value may be obtained by subtracting the normalized similarity from 1, or by transforming the similarity using the reciprocal of the similarity, etc.

[0060] In step S9, the discrimination determination unit 51 uses the analysis model obtained from the learning unit 42 to input the features obtained from the feature generation unit 41 as test data into the analysis model and obtains a list of reference values.

[0061] In step S10, the judgment result evaluation unit 52 uses the list of obtained reference values ​​to output the ID of the (similar) hand shape with the smallest reference value. In the judgment process of the judgment result evaluation unit 52, a threshold for judgment may be set for the similarity, and a judgment may be made only if the reference value is smaller than the threshold.

[0062] (effect) According to the embodiment, a hand shape recognition device 10 is provided that does not hinder the user's finger movements, allows the user to attach the vibration generation / acquisition unit 21 to the same position even unintentionally, and identifies the shape of the hand. Furthermore, the hand shape recognition device 10 can also prevent the vibration generation / acquisition unit 21 from shifting position due to hand or finger movements.

[0063] Embodiments of the present invention have been described above with reference to the drawings. However, the above embodiments are merely examples of configurations that embody the present invention. In other words, it is clear that the present invention is not limited to the above embodiments. Therefore, additions, omissions, substitutions, and other modifications of components are permitted without departing from the technical spirit of the present invention.

[0064] In short, the present invention is not limited to the embodiments described above, and can be modified in various ways during implementation without departing from its essence. Furthermore, each embodiment may be combined as appropriate, and in that case, the combined effects can be obtained. Moreover, the above embodiments include various inventions, and various inventions can be extracted by selecting combinations from the multiple constituent elements disclosed. For example, if the problem can be solved and effects obtained even if some constituent elements are deleted from all the constituent elements shown in the embodiment, then the configuration with these deleted constituent elements can be extracted as an invention. [Explanation of Symbols]

[0065] 10…Hand shape recognition device 20... Measuring device 21…Vibration generation / acquisition unit 21a... Piezoelectric element 21b…Conducting wire 22… Bio-adhesion site 23… Enclosure reinforcement section 23a…recess 24… External force absorption section 24a…recess 24b... protruding part 24c…hole 25... Mounting body 25a...hole 25b...Pocket 26... Mounting part 26a... Slit member 26b...Folding fastener 30…Signal generation / processing unit 31... Signal generation unit 32...Signal receiving unit 33... Signal amplification section 34... Signal extraction unit 40...Hand shape estimation unit 41...Feature generation unit 42…Learning Department 50...Identification part 51... Identification and determination unit 52…Judgment Result Evaluation Department 60…Database 70… Information processing equipment 71… Processor 72...ROM 73...RAM 74…Auxiliary storage device 75… Input / Output Interface 76... Bus 81...Input device 82…Output device 83… Input / Output Devices 84…Recording media

Claims

1. A measurement device comprising a vibration generation and acquisition unit capable of generating and acquiring vibrations, and the vibration generation and acquisition unit being attached to the back of the hand of the object to be measured with good positional reproducibility, A signal generation and processing unit generates a vibration generation signal to generate vibrations in the vibration generation and acquisition unit, transmits the vibration generation signal to the vibration generation and acquisition unit, receives and processes the vibration signal representing the vibrations that have passed through the hand acquired by the vibration generation and acquisition unit, and outputs a processed signal. A hand shape estimation unit generates a learning model that estimates hand shape using machine learning from the aforementioned processed signal, It has an identification unit that identifies the hand shape using the learning model, The vibration generation and acquisition unit has two piezoelectric elements capable of generating and acquiring arbitrary vibrations. In addition to the vibration generation and acquisition unit, the aforementioned measuring device unit includes: A housing reinforcement portion fixed to the piezoelectric element and reinforcing the piezoelectric element, On the opposite side of the housing reinforcement portion, a bio-adhesive portion is fixed to the piezoelectric element and is in close contact with the back of the hand, The bio-adhesive portion, the piezoelectric element, and the housing reinforcement portion are covered and protected from the side of the housing reinforcement portion, and an external force absorbing portion is provided to absorb external forces applied to the bio-adhesive portion, the piezoelectric element, and the housing reinforcement portion. The bio-adhesive portion, the piezoelectric element, the housing reinforcement portion, and the external force absorbing portion are housed in a pocket that exposes the bio-adhesive portion, and the device portion can be positioned to cover the back of the hand, The device has a device fixing part that is attached to the device body and fixes the device body, which covers the back of the hand, to the hand. Hand shape recognition device.

2. The aforementioned mounting part has a hole for passing the thumb of the hand through, The hand shape recognition device according to claim 1.

3. The aforementioned device fixing portion includes a hook-and-loop fastener fixed to one end of the device portion and a slit member fixed to the other end of the device portion, which has a slit through which the hook-and-loop fastener passes. The hand shape recognition device according to claim 1.

4. The bio-adhesive portion has high frictional force and high elasticity, and prevents gaps from forming between the piezoelectric element and the back of the hand even when the shape of the back of the hand changes due to hand and finger movements. The hand shape recognition device according to claim 1.

5. The aforementioned signal generation and processing unit is A signal generation unit that generates the vibration generation signal, A signal receiving unit that receives the vibration signal, A signal amplification unit that amplifies the vibration signal input from the signal receiving unit, The system includes a signal extraction unit that extracts the amplified vibration signal input from the signal amplification unit at regular intervals, The hand shape estimation unit is A feature generation unit acquires the processing signal and generates feature quantities for identification based on the processing signal, The system includes a learning unit that acquires the aforementioned features and learns an analysis model using training data consisting of pairs of the aforementioned features and the labels of the hand shapes. The aforementioned identification unit is An identification determination unit that acquires the aforementioned feature quantities and the aforementioned analysis model, and determines a numerical value for determining the hand shape based on the aforementioned feature quantities and the aforementioned model, The system includes a determination result evaluation unit that acquires the aforementioned numerical value and identifies the hand shape based on the aforementioned numerical value, The hand shape recognition device according to claim 1.

6. A measurement device including a vibration generating and acquisition unit capable of generating and acquiring vibrations generates a vibration generation signal that generates vibrations in the vibration generating and acquisition unit which is attached to the back of the hand of the object to be measured with good positional reproducibility, transmits the vibration generation signal to the vibration generating and acquisition unit, and receives and processes a vibration signal representing the vibrations that have passed through the hand acquired by the vibration generating and acquisition unit, and outputs a processed signal. The steps include acquiring the processing signal and generating a learning model that estimates hand shape using machine learning from the processing signal, The process includes the steps of acquiring the learning model and identifying the hand shape using the learning model, The vibration generation and acquisition unit has two piezoelectric elements capable of generating and acquiring arbitrary vibrations. In addition to the vibration generation and acquisition unit, the aforementioned measuring device unit includes: A housing reinforcement portion fixed to the piezoelectric element and reinforcing the piezoelectric element, On the opposite side of the housing reinforcement portion, a bio-adhesive portion is fixed to the piezoelectric element and is in close contact with the back of the hand, The bio-adhesive portion, the piezoelectric element, and the housing reinforcement portion are covered and protected from the side of the housing reinforcement portion, and an external force absorbing portion is provided to absorb external forces applied to the bio-adhesive portion, the piezoelectric element, and the housing reinforcement portion. The bio-adhesive portion, the piezoelectric element, the housing reinforcement portion, and the external force absorbing portion are housed in a pocket that exposes the bio-adhesive portion, and the device portion can be positioned to cover the back of the hand, The device has a device fixing part that is attached to the device body and fixes the device body, which covers the back of the hand, to the hand. A method for recognizing hand shapes.

7. A computer having a processor and memory, To perform at least some of the functions of the signal generation / processing unit, the hand shape estimation unit, and the identification unit as described in claim 1, Hand shape recognition program.

Images