JP5660556B1 - Mastication number identification display device - Google Patents

Abstract

The present invention provides a mastication frequency identification and display device that accurately counts and displays the number of mastications even if noise is mixed in mastication detection data. An electric signal obtained by detecting a change in the shape of the ear canal by a sensor is sampled by an AD conversion circuit and converted into mastication detection data, and the mastication detection data is stored in a memory 4a of a memory reading means 4. The autocorrelation calculating means 5 calculates the normalized autocorrelation value R (t) by calculating the mastication detection data read from the memory 4a of the memory reading means 4. Only when the peak of the autocorrelation value R (t) is included in the assumed chewing interval, the discrimination counting means 6 discriminates and counts the number of chewing times and outputs an identification count value. The display means 7 displays the identification count value. [Selection] Figure 1

Description

  The present invention relates to a mastication frequency identification / counting display device that detects mastication by detecting deformation of the ear canal, and counts and displays the number of mastications.

  Conventionally, many devices for detecting jaw movement and counting the number of mastications during a meal have been proposed. However, in such a chewing frequency counting device, there is a problem that counting errors are large because jaw movements and the like caused by conversation during meals are counted together. Therefore, in chewing, focusing on the fact that jaw movement changes at a predetermined interval unlike conversation, a method has been proposed in which counting accuracy is increased by counting only the signal components of the assumed chewing interval.

  Patent Document 1 proposes a method of counting the state in which the jaw is moving at an assumed mastication interval by capturing and processing the jaw movement with an image. Patent Document 2 proposes a method of counting the number of occurrences of peaks only when the peak generation timing interval of the signal waveform obtained by converting the deformation of the ear canal into an electric signal is the assumed mastication interval. .

  However, the methods of Patent Document 1 and Patent Document 2 have the following problems. In the method of Patent Document 1, it is necessary to handle a large amount of image data. Therefore, in the method of Patent Document 1, not only complicated calculation means are required, but it is not easy to identify mastication only from jaw movement. Further, in the method of Patent Document 2, since noise components other than mastication are included in the electrical signal, the signal waveform is not a simple waveform, and it is not easy to extract only the peak based on mastication. Furthermore, in the method of Patent Document 2, if a peak due to a noise component is mixed, the assumed chewing interval is divided and is not distinguished from mastication, and an error in counting the number of mastication occurs.

JP 2004-242902 A JP 2011-10791 A

  Therefore, the present invention has been made in view of the above-described conventional drawbacks, and its purpose is less affected by noise and can be used to count the number of chewing times with high counting accuracy. To propose a device.

Chewing number recognition and counting display device of the present invention stores a sensor for converting the movement of the jaws into an electric signal, a mastication detection data obtained by sampling the electrical signal to the memory, predetermined the mastication detection data from said memory It reads in each period T as the reference data a, and the storage reading means for reading the mastication detection data reading start end is different by one sample interval at the same sampling number as a plurality of comparison data C _m from the memory and the reference data a, correlation calculation processing by autocorrelation calculating means for outputting the autocorrelation value, the timing of the peak as the drop level H above the autocorrelation value chewing between each of the reference data a and the plurality of comparison data C _m An identification counting means for discriminating and counting the number of mastication times and outputting an identification count value when present within an assumed interval; Characterized in that it comprises a Shimesuru display means.

Further, the present invention is characterized in that the autocorrelation value input to the identification counting means is a normalized autocorrelation value obtained by dividing the autocorrelation value within the predetermined period T by the maximum autocorrelation value within the predetermined period T. To do. Furthermore, the present invention is characterized in that the memory reading means, autocorrelation calculating means, and identification counting means are realized by software of a personal computer. Still further, the present invention is characterized in that the memory reading means reads the reference data A and the plurality of comparison data C _m from the memory while storing the mastication detection data in the memory, and the display means displays the updated state of the number of mastication times. To do.

  Therefore, according to the present invention, the number of mastications can be accurately counted by identifying the signal component resulting from mastication based on autocorrelation.

It is a block diagram which shows the whole structure of the chewing count meter identification number display apparatus of one Embodiment of this invention. It is a wave form diagram which shows the calculation principle of the autocorrelation function in one Embodiment of this invention. It is a figure which shows the relationship between the electric signal waveform and autocorrelation value waveform in one Embodiment of this invention. It is a flowchart figure of the personal computer which shows the procedure of the autocorrelation calculation in one embodiment of this invention, an identification coefficient, and a count value display.

  Hereinafter, a mastication frequency identification / counting display device according to an embodiment of the present invention will be described with reference to the drawings. In the present embodiment, the reference data section T as a unit for performing the autocorrelation calculation is set to 2 seconds, and D is set to 0.7 seconds and d is set to 0.2 seconds in the assumed chewing interval (D ± d). It is set.

  1 includes a sensor 1, an amplifier 2, an AD conversion circuit 3, and a memory reading means 4 in a personal computer. The storage / reading means 4 includes a memory 4a. The memory reading means 4 stores mastication detection data during the meal period in the memory 4a. The amplifier 2 and the AD conversion circuit 3 are configured integrally with the sensor 1. The AD conversion circuit 3 and the memory reading means 4 in the personal computer are connected by wire.

  The sensor 1 detects the mastication state and converts it into an electric signal F by paying attention to the fact that the ear canal is deformed as the jaw moves. The sensor 1 has a built-in microphone or barometer 12 that detects a change in atmospheric pressure in a soft resin container 11 filled with air and converts the atmospheric pressure change into an electrical signal. Therefore, the sensor 1 outputs an electrical signal that is linked to the movement of the jaw by inserting the container 11 into the ear canal. Note that the configuration of the sensor 1 is a well-known configuration disclosed in, for example, Japanese Patent Application Laid-Open No. 11-318862, and a detailed description thereof is omitted.

  The amplifier 2 amplifies the electrical signal F that is linked to the amount of deformation of the ear canal. The AD conversion circuit 3 samples the amplified electrical signal at 1 kHz and converts it into mastication detection data. The mastication detection data during the meal period is sequentially stored in the memory 4a by the memory reading means 4 in the personal computer.

After the meal period ends, the memory reading means 4 reads the mastication detection data in units of 2000 samples for 2 seconds stored in the memory 4a as the reference data A from the memory 4a. On the other hand, the memory reading means 4 shifts the reading start timing by 1 sample from the timing preceding the reference data A by 1 sample, and reads 2000 mastication detection data in units of 2000 samples. read from the memory 4a as comparative data _{C m.}

The autocorrelation calculation means 5 in the personal computer that inputs the reference data A and the m-th comparison data C _m that is sequentially updated performs 2000 autocorrelation calculation processing and normalization processing on the reference data A, A normalized autocorrelation value R (t) is output.

  The identification counting means 6 in the personal computer that inputs the normalized autocorrelation value R (t) forms a peak that is detected when the normalized autocorrelation value R (t) is equal to or higher than a certain drop level H within the estimated mastication interval. Only when it is determined that the mastication has been performed, the number of mastications is discriminated and counted, and an identification count value is output.

  The identification count value counted by the identification counting means 6 is displayed on the display means 7 also serving as a display panel of the personal computer. Note that the above-described storage / reading means 4, autocorrelation calculating means 5 and identification counting means 6 in the personal computer are all functional blocks realized by software.

As shown in FIG. 2, in the present embodiment, the storage / reading means 4 reads the reference data A in units of 2000 samples from the memory 4a by dividing the electric signal F at intervals of 2 seconds with the reference data interval being 2 seconds. . On the other hand, the memory reading means 4 reads from the memory 4a the _mth comparison data Cm having the same number of samples as the reference data A and different read start timing.

As shown in FIG. 2, m of the m-th comparison data C _m is an integer value that changes in a range of 0 ≦ m <2000, and the start end of the m-th comparison data C _m corresponds to the sampling timing. It is something to identify. The 0th comparison data C ₀ is the first comparison data C _m which means that the start of the 0th comparison data C ₀ is exactly 1 second ahead of the reference data A.

Further, the 1999 comparison data C ₁₉₉₉ is the 2000th last comparison data C _m , which means that the start of the comparison data C ₁₉₉₉ is a timing preceding the reference data A by exactly 1 second behind the timing of one sample period. is there.

In the present embodiment, the autocorrelation calculating unit 5, and a comparison data C _m of the m reference data A and 2000 samples 2000 samples in the corresponding relationship, and accumulated on the multiplied for each sample autocorrelation value X (T) is obtained. Autocorrelation value X (t) is a value that indicates the correlation of height between the comparative data C _m of the reference data A and the m.

  As apparent from comparison of the waveform change of the reference data A and the autocorrelation value X (t) in FIG. 2, the waveform of the autocorrelation value X (t) is gentler than the waveform of the reference data A. Therefore, in order to evaluate the peak interval, it is much more advantageous to evaluate the waveform of the autocorrelation value X (t) than the waveform of the reference data A. Therefore, in the present embodiment, the autocorrelation calculation process is intentionally performed.

Further, the relationship between the mth comparison data _Cm and the autocorrelation value X (t) is specified as follows. For example, the autocorrelation value between the reference data A and the 0th comparison data C ₀ is specified by X (−1), and the autocorrelation value between the reference data A and the 1000th comparison data C ₁₀₀₀ is X It is specified by (0).

  Since the autocorrelation value X (0) is an autocorrelation value between the same data, the value is always the maximum value in each reference data section. However, the maximum value varies greatly for each data section because the electric signal waveform F differs for each data section.

  If the maximum value is different for each data section, the head level H for identifying mastication must be changed in proportion to the maximum value level.

  Therefore, the autocorrelation calculation means 5 of this embodiment makes the head level of the mastication identification constant, so that the maximum value is always 1 in each data section, and the normalized autocorrelation value R (t) = X (T) / X (0) is calculated and output.

As shown in FIG. 2, the normalized autocorrelation value R (t) becomes 1 at the timing of t = 0, and the timing corresponds to the 453rd comparison data C ₄₅₃ and the 1549th comparison data C _1549. A peak showing a high correlation value is formed. At this time, since the drop level H from the immediately preceding valley to the peak is exceeded, it is regarded as a peak in this embodiment.

The peak of the 453rd comparison data C _{453 is} a peak 547 samples before the reference data A, that is, 0.547 seconds before. The peak of the 1549th comparative data C _{1549 is} a peak after 549 samples from the reference data A, that is, after 0.549 seconds.

In the present embodiment, the assumed mastication interval T is 0.5 seconds ≦ T ≦ 0.9 seconds. Since the peak corresponding to the 453rd comparison data C ₄₅₃ and the peak corresponding to the 1549th comparison data C ₁₅₄₉ belong to the assumed mastication range, the identification counting means 6 regards it as mastication, and the count value of the number of mastication times Update.

  Hereinafter, the number of chewing times of the identification counting means 6 together with the three types of reference data A shown in FIGS. 3A, 3B, and 3C and the waveform charts of the normalized autocorrelation values R (t) corresponding thereto, respectively. The principle of identification counting will be described in detail.

First, in the case of FIG. 3A, the normalized autocorrelation value R (t) corresponding to the reference data A has a high peak at the timing corresponding to the 432nd comparison data C ₄₃₂ and the 1586th comparison data C _1586. Is recognized.

  These peaks are data that is 0.568 seconds before the reference data A and data that is 0.586 seconds after the reference data A, both of which are within the assumed chewing interval. From this, the identification counting means 6 renews the number of mastication times by assuming that mastication has occurred twice.

In the case of FIG. 3B, the normalized autocorrelation value R (t) corresponding to the reference data A is high at the timing corresponding to the 252nd comparison data C ₂₅₂ and the 1732nd comparison data C _1732. A peak is recognized, and a low peak is recognized at a timing corresponding to the 630th comparative data C ₆₃₀ and the 1308th comparative data C ₁₃₀₈ .

The peaks corresponding to the 630th comparison data C ₆₃₀ and the 1308th comparison data C ₁₃₀₈ are data at timings 0.370 seconds before and 0.308 seconds after the reference data A, respectively, and the assumed chewing range Not. For this reason, the discriminating / counting means 6 exceeds the head level H from the immediately preceding valley to the peak, but does not regard it as a peak.

On the other hand, the peaks corresponding to the 252nd comparison data C ₂₅₂ and the 1732nd comparison data C ₁₇₃₂ are data at timings of 0.748 seconds before the reference data A and 0.732 seconds after the reference data, respectively. And both are within the assumed chewing interval. From this, the identification counting means 6 renews the number of mastications by assuming that mastication has occurred due to the presence of these peaks.

  Further, in the case of FIG. 3 (c), the normalized autocorrelation value R (t) corresponding to the reference data A has a peak with a head level H or higher from the immediately preceding valley to the peak other than R (0). I can't. Therefore, the identification counting means 6 considers that there was no mastication and does not update the number of mastications.

  As described above, since the identification counting means 6 in the present embodiment satisfies the assumed mastication interval 0.5 ≦ t ≦ 0.9, in the normalized autocorrelation value R (t), from the immediately preceding valley to the peak The number of mastications is counted assuming that mastication occurs only when peaks at a head level H or higher exist at −0.9 ≦ t ≦ −0.5 and 0.5 ≦ t ≦ 0.9.

  The estimated chewing interval is two places before and after the reference data A, that is, −0.9 ≦ t ≦ −0.5 and 0.5 ≦ t ≦ 0.9, and the peak is only in one range. It is also assumed that it exists. However, in that case, since it cannot be regarded as continuous mastication, the identification counting means 6 does not update the number of mastications.

  Hereinafter, according to the flowchart of FIG. 4, the operation of the mastication number identification counting display program of the present embodiment, which reads the mastication detection data stored in the memory 4a from the memory 4a and counts the number of mastications, will be described.

  First, in the first step S1, the personal computer sets all parameters to initial values. Subsequently, in the second step S2, the memory reading means 4 reads the reference data A for the first reference data interval of 2 seconds, that is, 2000 samples from the memory 4a. The reference data A is sequentially updated at intervals of 2 seconds after the autocorrelation value X (t) is calculated as described above.

Next, in the third step S3, the memory reading means 4 initializes the read start timing of the _mth comparison data Cm. That is, the storage / reading means 4 initializes m = 0, which is exactly one second ahead of the beginning of the reference data A.

Further, in the fourth step S4, the storage / reading means 4 reads the mth comparison data _Cm for 2 seconds from the memory 4a for 2000 samples.

In the fifth step S5, the autocorrelation calculation means 5 multiplies the read-out mth comparison data _Cm and the reference data A for each sampling data and integrates the multiplication values to obtain the autocorrelation value X ( t) is calculated and stored in a first memory (not shown) in the identification counting means 6.

Next, in a sixth step S6, the memory reading means 4 checks whether or not m has reached 1999 and the reading of the mth comparison data _Cm has been completed. If m has not reached 1999, the memory reading means 4 updates the reading start timing of the _mth comparison data Cm by one sample in the seventh step S7, and returns to the fourth step S4.

  The step returning to the fourth step S4 is repeated 1999 times, and when the autocorrelation calculating means 5 obtains the autocorrelation value X (t), m = 1999, and the process proceeds to the eighth step S8 via the sixth step S6. To do.

  In the eighth step S8, the autocorrelation calculating means 5 obtains a normalized autocorrelation value R (t) by dividing the autocorrelation value X (t) by the maximum autocorrelation value (0), and the autocorrelation value X Instead of (t), the normalized autocorrelation value R (t) is stored in the memory in the identification counting means 6.

  In the ninth step S9, the discrimination counting means 6 detects whether or not a peak having a drop level H or higher from the immediately preceding valley to the peak is formed in the normalized autocorrelation value R (t).

  In the tenth step S10, the discriminating counting means 6 discriminates whether or not the detected peak exists at −0.9 ≦ t ≦ −0.5, and when the peak exists, the eleventh step S11. Migrate to On the other hand, when the peak is not identified, the identification counting means 6 assumes that there is no mastication and does not proceed to the eleventh step S11, but proceeds to the thirteenth step S13.

  In the eleventh step S11, the identification counting means 6 identifies whether or not there is a peak of the normalized autocorrelation value R (t) in the subsequent estimated chewing interval 0.5 ≦ t ≦ 0.9. When the peak is not identified, the identification counting means 6 considers that there was no mastication and does not proceed to the twelfth step S12, but proceeds to the thirteenth step S13.

  If there is a peak, the identification counting means 6 assumes that chewing has occurred, updates the number of chewing times by 2 in the 12th step S12, and proceeds to the 13th step S13.

  In the thirteenth step S13, the personal computer checks whether or not the reference data section is the final end. If the final end has not been reached, the personal computer moves to the fourteenth step S14, updates the reference data section, returns to the second step S2, and repeats the steps described above.

  In the thirteenth step S13, when the reference data section is the final end, the personal computer finishes updating the number of mastications, and displays the identification count value counted by the identification counter 6 in the fifteenth step S15 as the number of mastications. To display.

In this embodiment, the number of chewing times is counted after the end of the meal period, but the present invention is not limited to this. For example, when the number of chewing times is updated, the reference data A is read from the memory immediately after the data for 2 seconds is stored in the memory of the memory reading unit, and the m-th comparison data C _m and the reference data A are subsequently read out. The number of chewing times may be updated by performing autocorrelation calculation. By configuring in this way, it is possible to display a state in which the count value of the number of mastications is updated with a delay of about 1 second, and the accuracy of counting the number of mastications while adjusting the drop level H for determining the peak of the autocorrelation value Can be up.

  Further, in the present invention, the sensor 1, the amplifier 2, the AD conversion circuit 3, the battery, and the wireless transmission circuit are integrated, and the mastication detection data is received by the personal computer and read from the memory 4a while being stored in the memory 4a of the memory reading means 4. In addition, the autocorrelation calculation process, the mastication count process, and the display can be performed in parallel in the personal computer.

  In the present embodiment, the configuration after the memory reading means 4 is configured by a personal computer. However, in the present invention, the number-of-mastication identification display device can be configured as a dedicated device.

  Furthermore, the sensor of the present invention is not limited to the type that detects pressure as described in the present embodiment, but may be a type that detects deformation of a container by attaching a piezoelectric sheet in the container. A type in which a microphone is brought into contact with the jaw and bone conduction sound is detected may be used.

  Furthermore, in the present embodiment, the assumed chewing interval is 0.5 to 0.9 seconds, but the counting accuracy can be further increased by adjusting the range according to individual differences and the like.

  Similarly, if the value of the head level H is adjusted as necessary, the counting accuracy can be further increased.

  Furthermore, in the present invention, the correlation value may be either an autocorrelation value X (t) obtained by directly integrating the multiplication values or a normalized autocorrelation value R (t) obtained by normalization. In this case, the head level H corresponding to the correlation value X (t) needs to be constantly changed according to the value of the correlation value X (0). On the other hand, once the head level H with respect to the normalized correlation value R (t) is set, it is not necessary to change the value.

A part or all of the above-described embodiment can be described as in the following supplementary notes, but is not limited thereto.
(Appendix 1) a sensor that converts jaw movement into an electrical signal;
Mastication detection data obtained by sampling the electrical signal is stored in a memory, and the mastication detection data is read from the memory as reference data A every predetermined period T, and reading starts with the same sampling number as the reference data A storage reading means for reading a plurality of comparison data C _m from the memory is different mastication detection data one sample interval ends,
Autocorrelation calculating means for outputting the autocorrelation value by correlation operation between each of the reference data A and the plurality of comparison data C _m,
Identification counting means for discriminating and counting the number of mastication times and outputting an identification count value when the timing of the peak at which the autocorrelation value falls above the drop level H exists within the assumed chewing interval;
Display means for displaying the identification count value;
A chewing frequency identification and counting display device comprising:
(Supplementary Note 2) The autocorrelation value input to the identification counting means is a normalized autocorrelation value obtained by dividing the autocorrelation value within the predetermined period T by the maximum autocorrelation value within the predetermined period T. The chewing frequency identification counting display device according to appendix 1.
(Supplementary Note 3) The memory reading means reads the reference data A and the plurality of comparison data C _m from the memory while storing the mastication detection data in the memory,
3. The mastication number identification count display device according to appendix 1 or 2, wherein the display means displays an updated state of the number of mastication times.
(Additional remark 4) The said memory | storage reading means, the said autocorrelation calculating means, and the said identification counting means are implement | achieved by the software of a personal computer, The mastication frequency identification of any one of Additional remark 1 thru | or 3 characterized by the above-mentioned Counting display device.
(Supplementary note 5) A method of processing mastication detection data obtained by sampling an electrical signal obtained by converting jaw movement with a sensor by a computer to identify and display the number of mastication times,
A storage step of storing the mastication detection data in a memory;
The mastication detection data is read from the memory as the reference data A every predetermined period T, and the mastication detection data having the same sampling number as the reference data A and the reading start end is different by one sample interval from the memory. A read step for reading as C _m ;
Autocorrelation calculating step of outputting the autocorrelation value by correlation operation between each of the reference data A and the plurality of comparison data C _m,
An identification counting step for identifying and counting the number of mastication times and outputting an identification count value when the peak timing at which the autocorrelation value falls above the drop level H exists within the assumed chewing interval;
A display step of displaying the identification count value on a display means;
Mastication frequency identification and counting display method.
(Supplementary Note 6) The autocorrelation value input to the identification counting step is a normalized autocorrelation value obtained by dividing the autocorrelation value within the predetermined period T by the maximum autocorrelation value within the predetermined period T. The method of identifying and displaying the number of chewing times according to appendix 5.
(Supplementary Note 7) The reading step reads the reference data A and the plurality of comparison data C _m the mastication detection data from said memory while stored in said memory,
The display step of displaying the number of chewing times according to appendix 5 or 6, wherein the display step displays the updated state of the number of chewing times on the display means.
(Supplementary note 8) A program for processing mastication detection data obtained by sampling an electric signal obtained by converting jaw movement with a sensor by a computer to identify and count the number of mastication times, the computer In addition,
A storage process for storing the mastication detection data in a memory;
The mastication detection data is read from the memory as the reference data A every predetermined period T, and the mastication detection data having the same sampling number as the reference data A and the reading start end is different by one sample interval from the memory. Read processing to read as C _m ,
And the self correlation operation of outputting the autocorrelation value by correlation operation between each of the reference data A and the plurality of comparison data C _m,
An identification counting process for discriminating and counting the number of mastications and outputting an identification count value when the peak timing at which the autocorrelation value falls above the drop level H exists within the assumed chewing interval;
Display processing for displaying the identification count value on a display means;
A program for identifying and displaying the number of chewing times to execute.
(Supplementary note 9) The autocorrelation value input to the identification counting process is a normalized autocorrelation value obtained by dividing the autocorrelation value in the predetermined period T by the maximum autocorrelation value in the predetermined period T. The program for identifying and displaying the number of times of chewing according to appendix 8.
(Supplementary Note 10) The reading process reads the reference data A and the plurality of comparison data C _m the mastication detection data from said memory while stored in said memory,
10. The mastication frequency identification count display program according to appendix 8 or 9, wherein the display process displays an updated state of the mastication frequency on the display means.

DESCRIPTION OF SYMBOLS 1 ... Sensor 2 ... Amplifier 3 ... AD conversion circuit 4 ... Memory reading means 5 ... Auto correlation Calculation means 6... Identification counting means 7... Display means

Claims (4)

A sensor that converts jaw movements into electrical signals;
Mastication detection data obtained by sampling the electrical signal is stored in a memory, and the mastication detection data is read from the memory as reference data A every predetermined period T, and reading starts with the same sampling number as the reference data A storage reading means for reading a plurality of comparison data C _m from the memory is different mastication detection data one sample interval ends,
Autocorrelation calculating means for outputting the autocorrelation value by correlation operation between each of the reference data A and the plurality of comparison data C _m,
Identification counting means for discriminating and counting the number of mastication times and outputting an identification count value when the timing of the peak at which the autocorrelation value falls above the drop level H exists within the assumed chewing interval;
Display means for displaying the identification count value;
A chewing frequency identification and counting display device comprising:   The autocorrelation value input to the identification counting means is a normalized autocorrelation value obtained by dividing the autocorrelation value within the predetermined period T by the maximum autocorrelation value within the predetermined period T. The chewing frequency identification and counting display device according to 1. Said storage reading means reads the reference data A and the plurality of comparison data C _m the mastication detection data from said memory while stored in said memory,
The mastication number identification count display device according to claim 1 or 2, wherein the display means displays an updated state of the number of mastication times.   4. The mastication number identification / counting display device according to claim 1, wherein the storage / reading unit, the autocorrelation calculating unit, and the identification / counting unit are realized by software of a personal computer. 5. .

Images