JP2013235012A - Data processing apparatus, operation recognition system, operation determination method, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a data processing apparatus, an operation recognition system, an operation determination method and a program in which operation of a target can be recognized, without depending on magnitude of acceleration, by using a Doppler sensor.SOLUTION: A data processing apparatus 100 comprises: a data acquisition section 102 for acquiring a Doppler sensor output signal having a frequency of a differential between a frequency of a radiation wave that is an electromagnetic wave radiated to a recognition target 20 and a frequency of a reflection wave obtained by reflecting the radiation wave on the recognition target; a feature amount extraction section 106 for extracting a feature amount indicating features of the Doppler sensor output signal; and an operation determination section 108 for determining operation of the recognition target on the basis of the feature amount.

Description

  The present invention relates to a data processing apparatus and a data processing method, and more particularly, to a data processing apparatus and a data processing method capable of recognizing an action of an object using a Doppler sensor.

  2. Description of the Related Art Conventionally, there is a motion recognition system that can recognize a motion of a person or an object using a sensor. For example, the motion recognition system disclosed in Non-Patent Document 1 uses a three-axis acceleration sensor, and uses machine learning for operations such as walking, standing still, jumping, and running. I recognize.

Zhen-YuHe; Lian-Wen Jin, ”Activity recognition from acceleration data using AR modelrepresentation and SVM,” Proc. Of International Conference on Machine Learning and Cybernetics, pp.2245-2250, July 2008.

  However, since the motion recognition system using the acceleration sensor extracts and uses the motion acceleration, it is difficult to detect a slight motion of the body or a motion that moves at a constant speed, that is, a motion whose acceleration is close to zero. There was a problem that there was. In addition, since it is necessary to attach a sensor to the recognition object itself, the conditions for executing recognition are considerably limited.

  Accordingly, the present invention has been made in view of the above problems, and an object of the present invention is a new and improved technique capable of recognizing the motion of an object without depending on the magnitude of acceleration. Another object of the present invention is to provide a data processing apparatus, an operation recognition system, an operation determination method, and a program.

  In order to solve the above-described problems, according to an aspect of the present invention, the frequency of a radiated wave that is an electromagnetic wave radiated to a recognition object, and the frequency of a reflected wave that is reflected by the recognition object. A data acquisition unit that acquires a Doppler sensor output signal having a frequency of the difference, a feature amount extraction unit that extracts a feature amount indicating a feature of the Doppler sensor output signal, and based on the feature amount, the recognition target object There is provided a data processing device including an operation determination unit that determines an operation.

  With this configuration, the data processing apparatus acquires the Doppler sensor output signal and determines the operation of the recognition target object based on the feature amount indicating the feature of the Doppler sensor output signal. The data indicated by the Doppler sensor output signal changes according to the change in the distance to the recognition object. For this reason, it has been difficult for the conventionally used acceleration sensor to recognize a motion whose acceleration is close to 0, for example, a motion that moves at a constant speed and a slow motion, but it is based on the Doppler sensor output signal. According to this configuration, it is possible to recognize a motion whose acceleration is close to zero. Further, in order to obtain the Doppler sensor output signal, it is only necessary to radiate a radiated wave to a recognition object and receive the reflected wave. Therefore, it is not necessary to attach a sensor to the recognition object itself like an acceleration sensor, and the applicable range Is effective.

  In addition, the operation determination unit determines the operation of the recognition object by pattern matching based on the sample data of the feature amount, and the sample data is output from the Doppler sensor corresponding to a series of operations of the recognition object. It may be the feature amount of the signal.

  In addition, the feature amount extraction unit may extract the feature amount indicating features in the time domain and the frequency domain of the Doppler sensor output signal.

  Further, the feature amount extraction unit, the average value of the voltage in the time domain of the Doppler sensor output signal, the variance value, the maximum value, the minimum value, and the median value, and the average power in the frequency domain of the Doppler sensor output signal, At least one of the average frequency and the average power frequency may be extracted as the feature amount.

  The data acquisition unit acquires two or more Doppler sensor output signals, and the feature amount extraction unit determines the magnitude of the voltage value of the two or more Doppler sensor output signals as the Doppler sensor output signal. May be used as a variable depending on the distance between the Doppler sensor that outputs and the recognition object.

  The motion determination unit determines the motion of the recognition object by pattern matching based on the sample data of the feature amount, and the sample data corresponds to the movement of the recognition object in a plurality of predetermined directions. The feature amount of the Doppler sensor output signal may be used.

  The feature amount extraction unit may extract, as the feature amount, increase / decrease information of the maximum amplitude within a unit time obtained by equally dividing the movement from the start point to the end point by a predetermined natural number N.

  The unit time is preferably at least one cycle of the Doppler sensor output signal.

  The feature amount extraction unit may extract the largest increase / decrease information value as the feature amount from the increase / decrease information corresponding to each of a plurality of consecutive unit times.

  The feature amount extraction unit may extract the feature amount from the normalized voltage signal data of the Doppler sensor output signal.

  Further, the data acquisition unit acquires two Doppler sensor output signals with different phases output from one Doppler sensor, and the feature amount extraction unit is a magnitude relationship between the voltage values of the two Doppler sensor output signals. And the increase / decrease information of the voltage value at the crossing point when the voltage values of the two Doppler sensor output signals are equal to each other is extracted as a feature amount, and the motion determination unit is configured to recognize the recognition target based on the magnitude relationship and the increase / decrease information. Whether the object is approaching or separating from the Doppler sensor that outputs the Doppler sensor output signal may be determined.

  In addition, the operation determination unit may perform the determination at the intersection point in the intersection point where the absolute value of the difference from the voltage value at the previous intersection point is equal to or greater than a predetermined threshold.

  The feature amount extraction unit may compare the magnitude relationship using an average value of the voltage values of the two Doppler sensor output signals to suppress the influence of noise of the Doppler sensor output signal.

  In order to solve the above-described problem, according to another aspect of the present invention, a radiated wave is radiated to a recognition object, the reflected wave is reflected by the recognition object, and the reflected wave is received. A Doppler sensor node having a Doppler sensor that outputs a Doppler sensor output signal having a frequency difference between the frequency of the radiated wave and the reflected wave, a data acquisition unit that acquires the Doppler sensor output signal, and the Doppler sensor output A data processing device comprising: a feature amount extraction unit that extracts a feature amount indicating a feature of a signal; and a motion determination unit that determines a motion of the recognition target based on the feature amount. A motion recognition system is provided.

  In order to solve the above problems, according to another aspect of the present invention, a Doppler sensor node having a Doppler sensor and a Doppler sensor output signal acquired from the Doppler sensor node are processed, and a data acquisition unit, a feature amount In the motion recognition system having the data processing device having the extraction unit and the motion discrimination unit, the Doppler sensor emits a radiated wave to the recognition target, and the Doppler sensor causes the radiated wave to be generated by the recognition target. Receiving a reflected wave reflected; outputting a Doppler sensor output signal having a frequency difference between the frequency of the radiated wave and the frequency of the reflected wave; and the data acquisition unit including the Doppler sensor. A step of acquiring an output signal, and the feature amount extraction unit includes the Doppler An operation determination method comprising: extracting a feature amount indicating a feature of a sensor output signal; and determining the operation of the recognition target object based on the feature amount. Provided.

  In order to solve the above problems, according to another aspect of the present invention, a computer is provided with a frequency of a radiated wave that is an electromagnetic wave radiated to a recognition object, and the radiated wave is reflected by the recognition object. A data acquisition unit that acquires a Doppler sensor output signal having a frequency that is a difference from the frequency of the reflected wave, a feature amount extraction unit that extracts a feature amount indicating a feature of the Doppler sensor output signal, and the feature amount There is provided a program for causing a data processing apparatus to function as a data processing device comprising: an operation determining unit that determines the operation of the recognition object.

  As described above, according to the present invention, a data processing device, a motion recognition system, a motion determination method, and a program capable of recognizing the motion of an object without depending on the magnitude of acceleration using a Doppler sensor. Can be provided.

It is a functional block diagram of a data processor concerning a 1st embodiment of the present invention. It is explanatory drawing which shows the example of the waveform in the time domain of the Doppler sensor output signal when the recognition target object is typing. It is explanatory drawing which shows the power spectrum in the frequency domain of the Doppler sensor output signal when the recognition target object is typing. It is explanatory drawing which shows the example of the waveform in the time domain of the Doppler sensor output signal when the recognition target object is turning the arm. It is explanatory drawing which shows the power spectrum in the frequency domain of a Doppler sensor output signal when the recognition target object is turning the arm. It is explanatory drawing which shows the example of a feature-value. It is explanatory drawing which shows the power spectrum at the time of FFT of the data of sampling frequency fs and N points | pieces for the acquired Doppler sensor output signal. It is explanatory drawing which shows the relationship between the order i at the time of rearranging the power spectrum of FIG. 7 in order with a high electric power value, and the electric power value p (i). It is explanatory drawing which shows the example of an output screen. It is explanatory drawing which shows the example of an output screen. It is explanatory drawing for demonstrating arrangement | positioning of the Doppler sensor node of the action recognition system which concerns on the 2nd Embodiment of this invention, and the definition of recognition object operation | movement. It is explanatory drawing which shows the Doppler sensor output signal corresponding to the operation | movement of a 1st direction, and its increase / decrease information. It is explanatory drawing which shows the Doppler sensor output signal corresponding to the operation | movement of a 2nd direction, and its increase / decrease information. It is explanatory drawing which shows the Doppler sensor output signal corresponding to the operation | movement of a 3rd direction, and its increase / decrease information. It is explanatory drawing which shows the Doppler sensor output signal corresponding to the operation | movement of a 4th direction, and its increase / decrease information. It is description which shows the Doppler sensor output signal corresponding to the operation | movement of a 5th direction, and its increase / decrease information. It is explanatory drawing which shows the Doppler sensor output signal corresponding to the operation | movement of a 6th direction, and its increase / decrease information. It is explanatory drawing which shows the Doppler sensor output signal corresponding to the operation | movement of a 7th direction, and its increase / decrease information. It is explanatory drawing which shows the Doppler sensor output signal corresponding to the operation | movement of an 8th direction, and its increase / decrease information. It is explanatory drawing for demonstrating the Doppler sensor output signal and the feature-value at the time of "approaching" in the action recognition system which concerns on 3rd Embodiment. It is explanatory drawing for demonstrating the Doppler sensor output signal and its feature-value at the time of "separation" in the action recognition system concerning the embodiment. It is explanatory drawing which shows an example of the noise and waveform distortion of a Doppler sensor output signal in the embodiment. It is explanatory drawing which shows the relationship between the Doppler sensor output signal in the same embodiment, and the threshold value for suppressing a misjudgment. It is a flowchart for demonstrating the feature-value extraction and operation | movement discrimination | determination process in the embodiment. It is explanatory drawing for demonstrating the principle of operation of a Doppler sensor. It is explanatory drawing for demonstrating the relationship between the example of a Doppler sensor output signal, and the operation | movement of a recognition target object.

  Exemplary embodiments of the present invention will be described below in detail with reference to the accompanying drawings. In addition, in this specification and drawing, about the component which has the substantially same function structure, duplication description is abbreviate | omitted by attaching | subjecting the same code | symbol.

<Doppler sensor principle>
First, the motion recognition system according to each embodiment of the present invention uses a Doppler sensor. Therefore, in order to facilitate understanding of the motion recognition system according to each embodiment of the present invention, first, the principle of operation of a general Doppler sensor will be described with reference to FIG. 25 and FIG. FIG. 25 is an explanatory diagram for explaining the operating principle of the Doppler sensor. Moreover, FIG. 26 is explanatory drawing for demonstrating the relationship between the example of a Doppler sensor output signal, and operation | movement of a recognition target object.

  Referring to FIG. 25, the Doppler sensor 902 emits an electromagnetic wave (for example, a millimeter wave or a microwave) having a certain frequency with respect to the recognition target object 904, and a radiation wave 906 that is an ultrasonic wave. The Doppler sensor 902 detects a reflected wave 908 that is an electromagnetic wave reflected by the recognition object 904 from the radiated wave 906. At this time, the frequency of the reflected wave 908 changes according to the moving speed of the recognition object 904. The Doppler sensor 902 outputs a Doppler sensor output wave having a frequency difference (hereinafter referred to as “Doppler frequency”) between the radiation wave 906 and the reflected wave 908 as a voltage signal (hereinafter referred to as “Doppler sensor output signal”).

  Here, the Doppler frequency is proportional to the moving speed of the recognition object. Further, the amplitude of the Doppler sensor output signal depends on the reception intensity of the reflected wave.

  Next, the relationship between the waveform of the Doppler sensor output signal and the operation state of the recognition object 904 will be described with reference to the waveform example of the Doppler sensor output signal shown in FIG.

  In a state where the recognition object 904 is stationary as from t0 to t1 and after t3, the voltage value of the Doppler sensor output signal 910 becomes a constant value. This constant voltage is hereinafter referred to as a reference voltage V0. As described above, the amplitude of the Doppler sensor output signal 910 depends on the reception intensity of the reflected wave. However, if the reception intensity of the reflected wave 908 is large when the recognition object 904 is moving, the amplitude from the reference voltage V0 is increased. The amplitude increases. On the other hand, if the reception intensity of the reflected wave 908 is small, the amplitude from the reference voltage V0 is small. That is, the amplitude increases as the distance between the Doppler sensor 902 and the recognition object 904 becomes shorter. The amplitude decreases as the distance between the Doppler sensor 902 and the recognition object 904 increases.

  Thus, by detecting the magnitude of the voltage amplitude of the Doppler sensor output signal 910, the approximate distance between the Doppler sensor 902 and the recognition object 904 can be known. That is, according to the example of FIG. 26, the recognition object 904 approaches the Doppler sensor 902 during the period from t1 to t2 when the amplitude of the Doppler sensor output signal 910 tends to increase, and the amplitude of the Doppler sensor output signal 910 tends to decrease. It can be known from the Doppler sensor output signal 910 that the recognition object 904 is moving away from the Doppler sensor 902 during the period from t2 to t3.

Furthermore, the Doppler frequency changes according to the moving speed of the recognition object 904. That is, it can be seen that the recognition object 904 is moving at a high speed during the period from t5 to t6 when the frequency of the Doppler sensor output signal is high. Here, the Doppler sensor output signal of between t5 t6 the voltage value thereof is varied in value between the minimum voltage V _min and maximum voltage V _max. This is a phenomenon that occurs when, for example, a signal from a Doppler sensor is amplified, and the voltage value of the signal exceeds that due to the restrictions on the maximum output and minimum output of the amplifier used for amplification and the circuits related thereto. It does not rise or fall and is saturated.

  The motion recognition system according to the first to third embodiments of the present invention described below acquires a Doppler sensor output signal using a Doppler sensor, and analyzes the acquired Doppler sensor output signal to recognize the object to be recognized. The function of recognizing the operation of Each of the embodiments has a feature in a feature point extraction method from a Doppler sensor output signal and an operation determination method using the feature point extraction method. Details of each embodiment will be described below.

<First Embodiment>
[System configuration]
First, the configuration of the motion recognition system 10 according to the first embodiment of the present invention will be described with reference to FIG. FIG. 1 is a functional block diagram showing the configuration of the motion recognition system 10 according to the first embodiment of the present invention.

  The motion recognition system 10 has a function of recognizing the motion of the recognition target object 20. The motion recognition system 10 mainly includes at least one Doppler sensor node 200, an A / D converter 300, and a data processing device 100.

  The motion recognition system 10 acquires a Doppler sensor output signal used for recognizing the motion of the recognition target object 20 using the Doppler sensor node 200. The analog signal Doppler signal is converted into a digital signal by the A / D converter 300, and the digital signal is input to the data processing apparatus 100. Thereafter, the motion recognition system 10 analyzes the input Doppler sensor output signal in the data processing device 100 to determine the motion of the recognition target object 20.

(Doppler sensor node 200)
The Doppler sensor node 200 mainly includes a Doppler sensor 202, an amplifier 204, a low-pass filter (LPF) 206, and an output unit 208 in order to recognize an operation from the recognition target object 20.

  The Doppler sensor 202 radiates an electromagnetic wave to the recognition target object 20 (hereinafter, this electromagnetic wave is referred to as a radiated wave), and the reflected wave that is the electromagnetic wave reflected by the recognition target object 20 with the frequency of the radiated wave. This is a module that detects a Doppler frequency that is a difference from the frequency of the output and outputs a Doppler sensor output signal that is a signal having the Doppler frequency. The Doppler sensor 202 may be configured using the general Doppler sensor module described above. Then, the Doppler sensor 202 inputs the acquired Doppler sensor output signal to the amplifier 204.

  The amplifier 204 has a function of amplifying the Doppler sensor output signal input from the Doppler sensor 202. The amplifier 204 may be configured using a general amplifier circuit. The amplifier 204 inputs the Doppler sensor output signal after the amplification process to the low-pass filter 206.

  The low-pass filter 206 has a function of removing high-frequency noise from the Doppler sensor output signal input from the amplifier 204. The low-pass filter 206 may be configured using a general filter circuit. The low-pass filter 206 inputs the Doppler sensor output signal after the filter processing to the output unit 208.

  The output unit 208 has a function of an output interface for the Doppler sensor node 200 to output a Doppler sensor output signal. The output unit 208 inputs the Doppler sensor output signal input from the low pass filter 206 to the A / D converter 300.

  Further, the Doppler sensor node 200 is disposed at a position where the radiation wave from the Doppler sensor 202 can be emitted to the recognition target object 20. Then, the Doppler sensor output signal is acquired at a sampling frequency sufficient for the operation speed of the recognition object 20.

(A / D converter 300)
The A / D (Analog / Digital) converter 300 has a function of converting a Doppler sensor output signal, which is an analog signal received from a Doppler sensor node, into a digital signal.

(Data processing apparatus 100)
The data processing device 100 is a device that performs data processing on the digital signal of the Doppler sensor output signal acquired by the Doppler sensor node 200 and outputs a determination result. The data processing apparatus 100 mainly includes a data acquisition unit 102, a processing data storage unit 104, a feature amount extraction unit 106, an operation determination unit 108, and a determination result output unit 110.

  The data acquisition unit 102 has a function of acquiring data to be processed. In the present embodiment, the data acquisition unit 102 has an interface function of receiving a digitally converted Doppler sensor output signal input from the A / D converter 300.

  The processing data storage unit 104 is a data storage device that stores and holds data to be processed acquired by the data acquisition unit 102. The processing data storage unit 104 can include a storage medium, a recording device that records data on the storage medium, a reading device that reads data from the storage medium, a deletion device that deletes data recorded on the storage medium, and the like. Here, examples of the storage medium include magnetic recording media such as HDD (Hard Disk Drive), EEPROM (Electronically Erasable and Programmable Read Only Memory), flash memory, MRAM (Magnetoresistive Random Access Memory), FeRAM (Ferroelectric Random Access). Non-volatile memories such as Memory) and PRAM (Phase change Random Access Memory) are mentioned, but not limited to the above.

  The feature amount extraction unit 106 has a function of acquiring a Doppler sensor output signal that is data to be processed from the processing data storage unit 104 and extracting a feature amount indicating the feature of the acquired Doppler sensor output signal. In the present embodiment, the feature amount extraction unit 106 extracts feature amounts indicating features in the time domain and the frequency domain of the Doppler sensor output signal. A specific example of the feature amount extracted in the present embodiment will be described later.

  The operation determination unit 108 has a function of determining the operation of the recognition target object 20 based on the feature amount extracted by the feature amount extraction unit 106. In the present embodiment, the motion determination unit 108 determines the motion of the recognition target object 20 using pattern matching based on machine learning. The motion determination unit 108 executes a learning phase in which a machine learning model is constructed from the feature amount of the sample data before the motion is determined. As a machine learning algorithm used here, for example, Support Vector Machine (SVM) can be used. The action determination unit 108 learns what kind of tendency the feature amount of the Doppler sensor output signal shows by the machine learning based on the sample data, and constructs a model used for the determination.

  The determination result output unit 110 has a function of outputting the result determined by the operation determination unit 108. The discrimination result output unit 110 may be configured by a display device such as a CRT (Cathode Ray Tube) display device, a liquid crystal display (LCD) device, an OLED (Organic Light Emitting Display) device, and a lamp. You may comprise with audio | voice output apparatuses, such as a speaker and headphones. In another embodiment, the discrimination result output unit 110 may be a simple output interface, and may output data for displaying the discrimination result on a device having a display device or a sound output device. An example of the screen to be output will be described later.

  It should be noted that the functions of each unit of the data processing apparatus 100 described above are actually ROM and RAM (both of which store a control program describing a processing procedure for a control device such as a CPU (not shown) to realize these functions. This is achieved by reading a control program from a storage medium such as (not shown) and interpreting and executing the program.

[Recognition target]
Next, examples of recognition targets recognized by the operation determination unit 108 of the data processing apparatus 100 according to the present embodiment will be described with reference to FIGS. FIG. 2 is an explanatory diagram showing an example of a waveform in the time domain of the Doppler sensor output signal when the recognition target is typing. FIG. 3 is an explanatory diagram showing a power spectrum in the frequency domain of the Doppler sensor output signal when the recognition target is typing. FIG. 4 is an explanatory diagram illustrating an example of a waveform in the time domain of the Doppler sensor output signal when the recognition target is turning the arm. FIG. 5 is an explanatory diagram showing a power spectrum in the frequency domain of the Doppler sensor output signal when the recognition target is turning the arm.

As operations that the data processing apparatus 100 according to the present embodiment determines, for example, operations listed below are assumed.
・ Typing ・ Rotating arms ・ Wandering around (walking with your hands hanging around)
・ Slowly move your hand close to or move away from the sensor.・ Wave your hand vertically ・ Wave your hand horizontally ・ Wave your hand diagonally

  Among these, as an example, in the frequency domain obtained by fast Fourier transform (FFT) of the waveform in the time domain of the Doppler sensor output signal corresponding to “typing” and “turning the arm” and the data in the time domain The power spectrum is shown in FIGS.

  In the present embodiment, the feature amount of the recognition target data and the feature amount as sample data are extracted from the Doppler sensor output signal corresponding to a series of operations of the recognition target object 20 as exemplified above.

  As shown in FIG. 2 to FIG. 5, the waveform in the time domain and the power spectrum in the frequency domain of the Doppler sensor output signal corresponding to each operation show different features, and the feature amounts should naturally be different. The operation determination unit 108 of the data processing apparatus 100 according to the present embodiment determines the operation of the recognition target object 20 based on this feature amount.

[Feature extraction]
Here, a specific example of the feature amount extracted from the Doppler sensor output signal will be described with reference to FIG. FIG. 6 is an explanatory diagram illustrating an example of the feature amount.

  The feature amount extracted by the feature amount extraction unit 106 of the data processing apparatus 100 according to the present embodiment is, for example, an average value, variance value, maximum value, minimum value, and The median is mainly cited. Examples of frequency domain feature quantities include, for example, an average value of power and data to be processed are divided into a plurality of frequency domains, and the average value of power, average frequency, and power in each of the regions are higher. The average of the frequency of (a%) (however, you may remove | exclude the electric power in a frequency 0Hz below), an average power, and an average power frequency are mainly mentioned.

  Here, the calculation method of the average, average power, and average power frequency of frequencies with higher power (a%) will be described in detail with reference to FIGS. FIG. 7 is an explanatory diagram showing a power spectrum when the acquired Doppler sensor output signal is subjected to FFT on sampling frequency fs and data at N points. FIG. 8 is an explanatory diagram showing the relationship between the order i and the power value p (i) when the power spectrum of FIG. 7 is rearranged in descending order of the power value.

Here, 1 ≦ i ≦ N / 2, p (1) is the highest power value, and p (N / 2) is the lowest power value. The frequency corresponding to the i-th frequency from the higher power p (i) is defined as f (i). Using the data of the upper a% from the one with the higher power in the rearranged power spectrum, the average value f _ave of the frequency is the following formula (1), and the average power p _ave is the following formula ( 2) The average power frequency fp _ave can be calculated using the following formula (3).

[Output example]
Next, an example of an output screen output from the discrimination result output unit 110 of the data processing apparatus 100 according to the first embodiment of the present invention will be described with reference to FIGS. 9 and 10. 9 and 10 are explanatory diagrams illustrating examples of output screens.

  For example, the discrimination result output unit 110 may output the output screen 1100 shown in FIG. The output screen 1100 mainly has a character display area 1102 and an image display area 1104. At this time, the discrimination result output unit 110 displays a character corresponding to the result determined by the motion determination unit 108 in the character display area 1102 and further corresponds to the result determined by the motion determination unit 108 in the image display area 1106. An image to be displayed (a still image or a moving image) may be displayed. In the example of FIG. 9, the discrimination result output unit 110 displays character information “I am typing” and an image of a person who is typing.

  For example, the discrimination result output unit 110 may output the output screen 1110 shown in FIG. The output screen 1110 is a Doppler sensor for each of a = rotate arm, b = slow, c = flapping, d = poor swaying, e = butterfly, f = typing, g = stationary state, and h = seating state. It shows collectively how many times the output signal has been discriminated out of 50 times. For example, it is indicated that the arm rotation is determined to be 49 “arm rotation” and is determined to be “around” once.

[Example of effects]
Here, an example of the effect of the motion recognition system 10 according to the present embodiment will be described. First, unlike the motion recognition system using the acceleration sensor, the motion recognition system 10 according to the present embodiment does not need to be attached to the recognition target 20 itself. For this reason, it is possible to recognize the operation of the recognition object 20 to which it is difficult to attach the sensor. In addition, there is an effect that it is possible to recognize a slow movement that is difficult to recognize with a motion recognition system using an acceleration sensor, and to determine a movement that moves at a constant speed (acceleration is close to 0). . Furthermore, it is possible to determine the operation with high accuracy by using various feature amounts using not only time domain data but also frequency domain data.

<Second Embodiment>
Next, a second embodiment of the present invention will be described. The data processing apparatus 100 according to the present embodiment has a function of determining the moving direction of the recognition target object 20 that has moved on the two-dimensional plane with high accuracy using as few Doppler sensors as possible. In general, the greater the number of sensors used, the higher the discrimination accuracy because more features can be extracted. However, when the number of sensors is increased, the installation cost of the sensors increases, and the processing for determination increases. Therefore, the data processing apparatus 100 according to the present embodiment uses not only a sensor that radiates radio waves in the direct movement direction but also a Doppler sensor output signal of a sensor that does not emit radio waves in the direct movement direction. It is possible to detect movements in various directions with fewer sensors.

  The functional configuration of the data processing apparatus 100 according to the present embodiment is the same as that of the data processing apparatus 100 according to the first embodiment shown in FIG. When the data processing apparatus 100 according to the present embodiment and the data processing apparatus 100 according to the first embodiment are compared, the format of the sample data used in the operation determination unit 108 is different, and the feature amount extraction unit 106 extracts it accordingly. Features are different. In the following, differences from the first embodiment will be mainly described.

[Recognized operation]
First, the arrangement of the Doppler sensor node 200 according to the present embodiment and an outline of sample data will be described with reference to FIG. First, the operation of the recognition target as sample data is defined as eight directions on the two-dimensional plane shown in FIG. For example, eight directions are defined as shown below using the reference numerals in FIG.

-First direction (510 to 550)
Second direction (550 to 510)
Third direction (570 to 530)
-Fourth direction (530 to 570)
・ Fifth direction (580 to 540)
6th direction (540 to 580)
7th direction (520 to 560)
・ Eighth direction (560 to 520)

  The two Doppler sensor nodes 200a and 200b are arranged so that their radiation directions are orthogonal to each other. In the present embodiment, the Doppler sensor node 200a is arranged so that the radiation direction is the first direction, and the Doppler sensor node 200b is arranged so that the radiation direction is the third direction.

  Here, as for the arrangement of the plurality of Doppler sensor nodes 200, it is preferable that the radiation directions do not overlap. Furthermore, it is desirable that the radiation directions of the respective Doppler sensor nodes 200 are arranged so as to be dispersed as much as possible. For this reason, when two Doppler sensor nodes 200 are used as in the present embodiment, they are arranged so that their radial directions are orthogonal to each other. When N Doppler sensor nodes having a number of 3 or more are used, it is preferable that the angle formed by the radial direction is (360 ÷ N) degrees. For example, when three Doppler sensor nodes 200 are used, it is preferable that the three Doppler sensor nodes 200 are arranged such that the radiation directions form an angle of 120 degrees.

  At this time, the Doppler sensor output signals acquired at the Doppler sensor node 200a and the Doppler sensor node 200b have different amplitude increase / decrease patterns and combinations depending on directions, for example, as shown in FIGS. A feature amount extraction method for discriminating an action based on this will be described below.

[Feature extraction]
First, the feature amount extraction unit 106 normalizes the input Doppler sensor output signal in order to eliminate the influence of the difference between the maximum value and the minimum value of the voltage due to the difference in performance of the devices. For example, normalized Doppler sensor output signal data can be obtained using the following equation.

Normalized voltage value = (V−V _min ) ÷ (V _max −V _min )
Here, V is the voltage value of the Doppler sensor output signal before normalization, V _min is the minimum voltage value determined by the device, and V _max is the maximum voltage value determined by the device. As a result of the above calculation, the normalized voltage value takes a value from 0 to 1. In the following description, it is assumed that the reference voltage value is a voltage value when no operation is observed. Hereinafter, the voltage value of the Doppler sensor output signal refers to the normalized voltage value.

  Next, the feature amount extraction unit 106 detects the start and end points of the operation from the input Doppler sensor output signal, and determines the start and end times of the Doppler sensor output signal corresponding to one operation. There are various methods for detecting the start point and the end point. Here, in either the Doppler sensor node 200a or the Doppler sensor node 200b, the voltage value from the reference voltage is a certain period T1th or more during a certain period. An operation start time point ts is a time point when the state of being below a certain threshold value A1th continues and the magnitude of the voltage value becomes equal to or greater than the threshold value A1th. Then, a Doppler sensor output signal corresponding to one operation is detected with a time point when the voltage value becomes equal to or less than a certain threshold A2th for a certain period T2th or more as an operation end time te.

  Next, the feature quantity extraction unit 106 equally divides the time from the start point ts to the end point te into N (N is a natural number), and sets one section divided into N as unit time. The unit time is set to be at least one wave period. This is to ensure that the maximum amplitude of the waveform is included within the unit time. The magnitude of the maximum amplitude within the unit time is set as the section maximum amplitude. The unit time varies depending on the speed of operation per one time. For example, when the operation speed is fast, the time from the start point ts to the end point te is shorter than when the operation speed is slow, and thus the unit time is inevitably shortened.

  Then, the feature amount extraction unit 106 detects the section maximum amplitude within the unit time, and extracts N section maximum amplitude values from each of the Doppler sensor node 200a and the Doppler sensor node 200b. At this time, the magnitude of the amplitude of the Doppler sensor output signal is a variable that depends on the distance between the recognition object 20 and the Doppler sensor 202. Specifically, when the distance between the Doppler sensor 202 and the recognition object 20 is long, the distance is smaller than when the distance is short. Further, the feature amount extraction unit 106 compares the section maximum amplitude value of the i-th (0 <i ≦ N) unit time with the section maximum amplitude value of the immediately preceding (i−1) -th unit time, and calculates the value. Increase / decrease information is extracted as a feature amount, which is “increased” if the increase is, and “decrease” if it is decreasing. At this time, for example, as described with reference to FIG. 26, the same value may be continued in a saturated state. In this case, the increase / decrease information is set to “no change”.

  As described above, the feature amount extraction unit 106 extracts (N−1) pieces of increase / decrease information from the Doppler sensor node 200a and the Doppler sensor node 200b, respectively, and 2 (N−1) pieces of increase / decrease information as feature amounts. .

  For example, FIGS. 12 to 19 are explanatory diagrams showing waveforms of the Doppler sensor node 200a and the Doppler sensor node 200b and increase / decrease information in each unit time when N = 7. As shown here, the increase / decrease information shows a different pattern for each operation. Using this increase / decrease information, the action determination unit 108 executes a learning phase in which a machine learning model is constructed from the feature amounts of these sample data by machine learning.

  For example, the waveform and feature amount of the Doppler sensor output signal corresponding to the operation in the first direction shown in FIG. 12 are shown. At this time, the recognition object 20 has moved from 510 to 550 shown in FIG. The distance from the Doppler sensor node 200a to the recognition object 20 continues to increase while moving from 510 to 550. For this reason, as shown in FIG. 12A, the amplitude of the Doppler sensor output signal acquired at the Doppler sensor node 200a continues to decrease.

  On the other hand, the distance from the Doppler sensor node 200b to the recognition target object 20 decreases until the recognition target object 20 reaches the center point 590 from 510, and increases until it reaches the center point 550 from 550. For this reason, as shown in FIG. 12B, the amplitude of the Doppler sensor output signal acquired at the Doppler sensor node 200b increases until the recognition target 20 reaches the center point 590, and then decreases.

  Then, the feature amount extraction unit 106 also performs the normalization, the detection of the start point and the end point, and the extraction of the increase / decrease information for the output signal of the Doppler sensor to be discriminated in the same manner as the processing for the sample data. Then, the motion determination unit 108 determines a motion by pattern matching according to a machine learning model.

  When extracting the increase / decrease information as a feature amount, the feature amount extraction unit 106 further divides a plurality of continuous increase / decrease information into M groups and reduces the increase / decrease for each group. One representative value of information may be determined. For example, if four groups of increase / decrease information are grouped, and the increase / decrease information changes to “increase”, “increase”, “decrease”, “increase”, the representative value of this group is set to “increase” Also good. When the value of N is large and the number of increase / decrease information increases, the effect of speeding up the process and the influence of slight fluctuation of the increase / decrease information are removed, and data reflecting the overall trend can be acquired. There is an effect.

[Example of effects]
An example of the effect of the motion recognition system 10 according to the second exemplary embodiment of the present invention will be described. The motion recognition system 10 according to the present embodiment is an effect obtained by using a Doppler sensor instead of an acceleration sensor among the examples of the effects in the first embodiment. In addition to “recognizing the movement of the object 20”, “recognizing slow movements and determining movements moving at a constant speed (acceleration close to 0)”, the following effects are obtained: I can hope.

  By using a combination of features of Doppler sensor output signals acquired by a plurality of sensors, it is possible to determine movement in a sparse direction even when there is no sensor emitting in the direction of movement of the recognition object. There is an effect. Further, since the operation can be determined only by the time domain feature without extracting the frequency domain feature using FFT, the operation can be determined by a simple process. Also, since the start point and end point of the operation are detected and the feature amount is extracted for each region obtained by dividing the time from the start point to the end point by a predetermined natural number, a fixed number of features are obtained regardless of the operation speed. The amount can be extracted.

<Third Embodiment>
[Recognition target]
Next, an operation recognition system according to the third embodiment of the present invention will be described. The motion recognition system 10 according to the present embodiment uses two Doppler sensor output signals output from one Doppler sensor 202 for recognition of the recognition object 20. The two Doppler sensor output signals have different phases. FIG. 20 is an explanatory diagram showing a Doppler sensor output signal when the recognition object 20 is “approaching” the Doppler sensor 202 in the motion recognition system 10 according to the present embodiment. FIG. 21 is an explanatory diagram showing a Doppler sensor output signal when the recognition object 20 is “separated” from the Doppler sensor 202 in the motion recognition system 10 according to the present embodiment.

  As shown in FIGS. 20 and 21, the Doppler sensor output signal used in the motion recognition system 10 according to the present embodiment is indicated by a solid line in FIGS. 20 and 21, and the voltage value is indicated by v1 (t). If the Doppler sensor output signal to be output is the first Doppler sensor output signal, the Doppler sensor output signal indicated by the dotted line and the voltage value indicated by v2 (t) is the second Doppler sensor output signal, FIG. As shown in FIG. 5, the phase of the second Doppler sensor output signal is advanced by 90 degrees with respect to the first Doppler sensor output signal when “approaching”. As shown in FIG. 21, the phase of the second Doppler sensor output signal is delayed by 90 degrees with respect to the first Doppler sensor output signal at the time of “separation”. As described above, the motion recognition system 10 according to the present embodiment uses the Doppler sensor adjusted so that the first Doppler sensor output signal and the second Doppler sensor output signal have a phase difference. Therefore, by detecting the phase difference between the two Doppler sensor output signals output from one Doppler sensor 202, it is determined whether the recognition target 20 is approaching or separating from the Doppler sensor 202. It can be discriminated.

  For example, the phase difference can be determined by using hardware such as a general phase comparator. However, in general, a device in which hardware processing and software processing are mixed has a problem that the device size increases. Therefore, a data processing apparatus 100 is proposed below that can determine whether the recognition target 20 is approaching or moving away from the Doppler sensor 202 only by software processing. Furthermore, a data processing apparatus 100 for accurately determining operation even in an environment where noise is mixed in the Doppler sensor output signal will be described.

[Feature extraction]
Next, feature amounts extracted by the feature amount extraction unit 106 of the data processing apparatus 100 according to the present embodiment will be described with reference to FIGS. In the present embodiment, the data acquisition unit 102 of the data processing apparatus 100 acquires two Doppler sensor output signals having different phases output from one Doppler sensor. Then, the feature amount extraction unit extracts a feature amount described below from the two Doppler sensor output signals.

The voltage value of the first Doppler sensor output signal at time t is v1 (t), and the voltage value of the second Doppler sensor output signal is v2 (t). Then, a difference value v1 (t) -v2 (t) between two Doppler sensor output signals at each observation time is obtained, and from the time when the value of v1 (t) -v2 (t) becomes zero, time switched to a negative value, or a time when switched from a negative to a positive value and when the waveform crosses each intersection point from ascending order of time and t _k (k is a natural number.). Further, the voltage value at each crossing point _{t k} and _{V cross (t k).}

As shown in FIG. 20, when the recognition target 20 is approaching the Doppler sensor 202, between two consecutive intersection times t _k−1 and t _k , that is, t _k−1 <t. When <t _k and v1 (t) −v2 (t) is a negative value, the value of V _cross (t _k ) −V _cross (t _k−1 ) is a positive value. When v1 (t) −v2 (t) is a positive value, the value of V _cross (t _k ) −V _cross (t _k−1 ) is a negative value.

On the other hand, as shown in FIG. 21, when the recognition target 20 is separated from the Doppler sensor 202, between two consecutive intersection times t _k−1 and t _k , that is, t _{k When −1} <t <t _k and the value of v1 (t) −v2 (t) is a negative value, the value of V _cross (t _k ) −V _cross (t _k−1 ) is also negative. When the value of v1 (t) −v2 (t) is a positive value, the value of V _cross (t _k ) −V _cross (t _k−1 ) is also a positive value. Yes.

The operation determination unit 108 according to the first embodiment and the second embodiment of the present invention determines the operation using pattern matching according to machine learning, but the operation determination unit 108 according to the present embodiment Thus, the positive / negative information of the value of v1 (t) −v2 (t) and the positive / negative of the value of V _cross (t _k ) −V _cross (t _k−1 ) when t _k−1 <t <t _k The motion determination unit 108 can determine whether the recognition target object 20 is approaching or moving away from the Doppler sensor 202 according to the pattern of the relationship with the information.

  As described above, in order to determine whether the recognition target 20 is approaching or moving away from the Doppler sensor 202 only by software processing, two Doppler sensor output signals output from one Doppler sensor 202 are The basic principle of the method of extracting the phase difference characteristics of the Doppler sensor output signal and the phase difference characteristics of the Doppler sensor output wave at the time of "separation" has been explained. The signal includes noise and distortion, and the movement determination unit 108 may not be able to correctly recognize the movement of the recognition target object 20 due to the influence of the noise and distortion.

  Therefore, a pattern that cannot be recognized correctly due to the effects of noise and distortion and the configuration of the data processing apparatus 100 for suppressing the influence will be described with reference to FIGS. 22 and 23. FIG. FIG. 22 is an explanatory diagram illustrating an example of noise and waveform distortion of the Doppler sensor output signal. FIG. 23 is an explanatory diagram showing a relationship between a Doppler sensor output signal and a threshold value for suppressing erroneous determination.

For example, the two Doppler sensor output signals shown in FIG. 22 include noise and waveform distortion. For this reason, a plurality of intersections are observed in a short time in the region α and the region β. The above feature quantity extraction method assumes a Doppler sensor output signal having an ideal waveform as shown in FIGS. 21 and 22, for example. For this reason, if the feature amount extraction unit 106 extracts feature amounts for all the intersections observed here, the operation determination unit 108 cannot obtain a correct determination result. Therefore, the operation determination unit 108 uses the predetermined threshold value Vth to suppress the influence of noise and waveform distortion of the Doppler sensor output signal. Specifically, when the absolute value of _{V cross (t k) -V cross} (t k-1) at the intersection point _{t k} is equal to or less than the predetermined threshold value Vth, the operation determination unit 108, such cross point _{t k} Is not recognized as a crossing point. That is, the operation determination unit 108 performs the operation determination at the crossing point in which the difference from the voltage value at the previous crossing point is equal to or greater than a predetermined threshold among the crossing points.

Furthermore, by simply setting the threshold value Vth, when the Doppler sensor output signal as shown in FIG. 23 includes a noise, _{'V cross} in _{(t k' t k) -V} cross (t k-1) value is not recognized as intersecting below Vth, in _{t k} is when the value of _{V cross (t k) -V cross} (t k-1) is identified as intersecting become more Vth. At this time, the difference between the voltage values of the first Doppler sensor output signal and the second Doppler sensor output signal is switched from positive to negative and should be recognized as “approaching”, but is recognized as “separation”. May be. In order to suppress such erroneous determination, the value of v1 (t) −v2 (t) and the positive / negative information of the value of v1 (t) −v2 (t) May be used. The average value may be calculated for each predetermined unit section Tunit. With this configuration, it is possible to reduce the possibility of erroneous determination due to noise.

  Next, operations of the feature amount extraction unit 106 and the operation determination unit 108 according to the present embodiment will be described with reference to FIG. FIG. 24 is a flowchart for explaining feature amount extraction and operation determination processing according to the present embodiment.

  First, the feature amount extraction unit 106 extracts voltage values v1 (t) and v2 (t) from two Doppler sensor output signal data (S102). Then, the feature amount extraction unit 106 calculates an average value of v1 (t) −v2 (t) for each unit section, for example, for each unit (S104).

Then, the feature amount extraction unit 106 determines whether or not the average value of v1 (t) −v2 (t) has changed from negative to positive (S106). If the time changes from negative to positive, assuming that time is t _k , the value of V _cross (t _k ) −V _cross (t _k−1 ) is calculated and input to the operation determination unit 108. On the other hand, if the average value of v1 (t) -v2 (t) has not changed from negative to positive, then whether or not the average value of v1 (t) -v2 (t) has changed from positive to negative? Is determined (S114). When the average value of v1 (t) −v2 (t) changes from positive to negative, the feature amount extraction unit 106 similarly calculates V _cross (t _k ) −V _cross (t _k−1 ). A value is calculated and input to the operation determination unit 108. Here, when the average value of v1 (t) −v2 (t) does not change from positive to negative, the process returns to the acquisition of the voltage value in step S102 again.

When the average value of v1 (t) −v2 (t) changes from positive to negative in the determination in step S106, the operation determination unit 108 receives the received V _cross (t _k ) −V _cross (t _It is determined whether or not the value of ( _k-1 ) is greater than a predetermined threshold value Vth (S108). As a result of the determination in step S108, when it is determined that the value of V _cross (t _k ) −V _cross (t _k−1 ) is greater than the predetermined threshold value Vth, the operation determination unit 108 determines that the recognition target object 20 is It recognizes that it is “approaching” to the Doppler sensor 202 (S112). On the other hand, as a result of the determination in step S108, if it is not determined that the value of V _cross (t _k ) −V _cross (t _k−1 ) is greater than the predetermined threshold value Vth, then V _cross (t _k ) −V _cross (t _k−1 ) It is determined whether the value is smaller than a predetermined threshold −Vth (S110). In this determination, when it is determined that the value is smaller than the predetermined threshold −Vth, the motion determination unit 108 recognizes that the recognition target 20 is “separated” with respect to the Doppler sensor 202 (S120). On the other hand, if it is not determined in step S110 that the value is smaller than the predetermined threshold −Vth, the process returns to the extraction of the voltage value in step S102 again.

On the other hand, if it is determined in step S114 that the average value of v1 (t) −v2 (t) has changed from positive to negative, then V _cross (t _k ) −V _cross (t _It is determined whether or not the value of ( _k-1 ) is smaller than a predetermined threshold value -Vth (S116). In this determination, when it is determined that the value is smaller than −Vth, the motion determination unit 108 recognizes that the recognition target 20 is “approaching” to the Doppler sensor 202 (S112). On the other hand, when it is not determined that the value is smaller than −Vth, the operation determination unit 108 further determines whether or not the value of V _cross (t _k ) −V _cross (t _k−1 ) is larger than Vth. Is determined (S118). If it is determined as a result of the determination in step S118 that it is greater than Vth, the operation determination unit 108 recognizes that the recognition target 20 is “separated” from the Doppler sensor 202 (S120). On the other hand, if it is not determined in step S118 that it is greater than Vth, the process returns to the extraction of the voltage value in step S102.

  After the above processing is executed and “approach” is recognized in step S112, or after “separation” is recognized in step S120, the motion determination unit 108 determines whether or not there is further data to be recognized. Determination is made (S122). If there is no further data to be recognized, the operation determination process is terminated. If there is more data to be recognized, the process returns to the extraction of the voltage value in step S102 for the data.

  As described above, the data processing apparatus 100 that determines the distinction between approach and separation from two Doppler sensor output signals output from one Doppler sensor without using hardware such as a phase comparator has been described. For example, the determination of the “approach” and “separation” states may be corrected afterwards in order to further reduce the influence of noise on the Doppler sensor output signal. For example, the unit time is divided every unit time, and the one with the larger number of times of discrimination between “approach” and “separation” within the unit time is set as the representative value of the state of the unit time. Moreover, you may add correction | amendment from the context of a state. For example, when it is determined that the separation is performed for only a short time while the approaching state continues, the “separation” determination may be regarded as an influence of noise and may be corrected to “approach”.

[Example of effects]
The operation recognition system 10 according to the third embodiment of the present invention has been described above. An example of the effect of the operation recognition system 10 will be described. The motion recognition system 10 according to the present embodiment is an effect obtained by using a Doppler sensor instead of an acceleration sensor among the examples of the effects in the first embodiment. In addition to “recognizing the movement of the object 20”, “recognizing slow movements and determining movements moving at a constant speed (acceleration close to 0)”, the following effects are obtained: I can hope.

  First, it is possible to determine whether the recognition target object 20 is approaching or moving away only by software processing without using hardware such as a phase comparator. For this reason, it is possible to reduce the size of the device itself as compared with the case where hardware is installed. Further, even when two Doppler sensor output signals output from one Doppler sensor for use in discrimination include noise, the operation can be discriminated with high accuracy.

  The preferred embodiments of the present invention have been described in detail above with reference to the accompanying drawings, but the present invention is not limited to such examples. It is obvious that a person having ordinary knowledge in the technical field to which the present invention pertains can come up with various changes or modifications within the scope of the technical idea described in the claims. Of course, it is understood that these also belong to the technical scope of the present invention.

  For example, in the first embodiment, the configuration includes one Doppler sensor node, but the present invention is not limited to such an example. For example, it is good also as a structure containing two or more.

  Further, for example, in the second embodiment, eight directions are defined and the movement in the defined direction is recognized, but the present invention is not limited to such an example. For example, the number of directions to be defined is not necessarily eight. Further, the movement is not limited to a linear direction, and can be applied as long as the increase / decrease information of the voltage value of the Doppler sensor output signal shows a specific pattern.

  For example, in the second embodiment, the increase / decrease information of a predetermined natural number N is used as sample data, but the present invention is not limited to this example. For example, after extracting the increase / decrease information of a predetermined natural number N, information on the increase / decrease ratio may be used as sample data.

  Moreover, although the said embodiment demonstrated the operation | movement recognition system using only a Doppler sensor, this invention is not limited to this example. For example, it can be used in combination with data obtained from other sensors.

  In addition, the motion recognition system according to each of the embodiments described above not only displays and outputs recognized motion data on an output unit such as a display unit, but also outputs the data to be used as, for example, device control data. You can also.

  In this specification, the steps described in the flowcharts are executed in parallel or individually even if they are not necessarily processed in time series, as well as processes performed in time series in the described order. Including processing to be performed. Further, it goes without saying that the order can be appropriately changed even in the steps processed in time series.

DESCRIPTION OF SYMBOLS 10 Action recognition system 20 Recognition target object 100 Data processing apparatus 102 Data acquisition part 104 Process data memory | storage part 106 Feature-value extraction part 108 Operation | movement discrimination | determination part 110 Discrimination result output part 200 Doppler sensor node 300 A / D converter

Claims (7)

A data acquisition unit that acquires a Doppler sensor output signal having a frequency that is a difference between a frequency of a radiated wave that is an electromagnetic wave radiated to a recognition object and a frequency of a reflected wave that is reflected by the recognition object; ,
A feature amount extraction unit that extracts a feature amount indicating the feature of the Doppler sensor output signal;
An action discriminating unit for discriminating an action of the recognition object based on the feature amount;
With
The data acquisition unit acquires two or more Doppler sensor output signals,
The feature amount extraction unit uses the magnitude of the amplitude of the voltage value of the two or more Doppler sensor output signals as a variable depending on the distance between the Doppler sensor that outputs the Doppler sensor output signal and the recognition object,
The operation determination unit determines the operation of the recognition object by pattern matching based on the sample data of the feature amount,
The sample data is the feature amount of the Doppler sensor output signal corresponding to the movement of the recognition object in a plurality of predetermined directions,
The feature amount extraction unit extracts, as the feature amount, increase / decrease information of a maximum amplitude within a unit time obtained by equally dividing a start point to an end point of the movement by a predetermined natural number N. .   The data processing apparatus according to claim 1, wherein the unit time is at least a length of one cycle or more of the Doppler sensor output signal.   The data according to claim 1, wherein the feature amount extraction unit extracts the largest increase / decrease information value as the feature amount from the increase / decrease information corresponding to each of the plurality of consecutive unit times. Processing equipment.   The data processing apparatus according to claim 1, wherein the feature amount extraction unit extracts the feature amount from voltage signal data normalized from the Doppler sensor output signal. A Doppler sensor output that emits a radiated wave to a recognition object, receives a reflected wave reflected by the recognition object, and has a frequency difference between the frequency of the radiated wave and the frequency of the reflected wave A Doppler sensor node having a Doppler sensor for outputting a signal;
A data acquisition unit for acquiring the Doppler sensor output signal;
A feature amount extraction unit that extracts a feature amount indicating the feature of the Doppler sensor output signal;
An action discriminating unit for discriminating an action of the recognition object based on the feature amount;
Have
The data acquisition unit acquires two or more Doppler sensor output signals,
The feature amount extraction unit uses the magnitude of the amplitude of the voltage value of the two or more Doppler sensor output signals as a variable depending on the distance between the Doppler sensor that outputs the Doppler sensor output signal and the recognition object,
The operation determination unit determines the operation of the recognition object by pattern matching based on the sample data of the feature amount,
The sample data is the feature amount of the Doppler sensor output signal corresponding to the movement of the recognition object in a plurality of predetermined directions,
The feature amount extraction unit extracts, as the feature amount, increase / decrease information of a maximum amplitude within a unit time obtained by equally dividing a start point to an end point of the movement by a predetermined natural number N. When,
A motion recognition system comprising: An operation recognition system comprising a Doppler sensor node having a Doppler sensor and a data processing device that processes the Doppler sensor output signal acquired from the Doppler sensor node and includes a data acquisition unit, a feature amount extraction unit, and an operation determination unit.
The Doppler sensor emitting a radiated wave to a recognition object;
The Doppler sensor receiving a reflected wave of the radiated wave reflected by the recognition object;
The Doppler sensor outputting a Doppler sensor output signal having a difference frequency between the frequency of the radiated wave and the frequency of the reflected wave;
The data acquisition unit acquiring the Doppler sensor output signal;
The feature amount extraction unit extracting a feature amount indicating a feature of the Doppler sensor output signal; and
The operation determining unit determining the operation of the recognition object based on the feature amount,
The data acquisition unit acquires two or more Doppler sensor output signals,
The feature amount extraction unit uses the magnitude of the amplitude of the voltage value of the two or more Doppler sensor output signals as a variable depending on the distance between the Doppler sensor that outputs the Doppler sensor output signal and the recognition object,
The operation determination unit determines the operation of the recognition object by pattern matching based on the sample data of the feature amount,
The sample data is the feature amount of the Doppler sensor output signal corresponding to the movement of the recognition object in a plurality of predetermined directions,
The feature amount extraction unit extracts, as the feature amount, increase / decrease information of a maximum amplitude within a unit time obtained by equally dividing a start point to an end point of the movement by a predetermined natural number N. . Computer
A data acquisition unit that acquires a Doppler sensor output signal having a frequency that is a difference between a frequency of a radiated wave that is an electromagnetic wave radiated to a recognition object and a frequency of a reflected wave that is reflected by the recognition object; ,
A feature amount extraction unit that extracts a feature amount indicating the feature of the Doppler sensor output signal;
An action discriminating unit for discriminating an action of the recognition object based on the feature amount;
With
The data acquisition unit acquires two or more Doppler sensor output signals,
The feature amount extraction unit uses the magnitude of the amplitude of the voltage value of the two or more Doppler sensor output signals as a variable depending on the distance between the Doppler sensor that outputs the Doppler sensor output signal and the recognition object,
The operation determination unit determines the operation of the recognition object by pattern matching based on the sample data of the feature amount,
The sample data is the feature amount of the Doppler sensor output signal corresponding to the movement of the recognition object in a plurality of predetermined directions,
As the data processing apparatus, the feature amount extraction unit extracts, as the feature amount, increase / decrease information of a maximum amplitude within a unit time obtained by equally dividing a start point to an end point of the movement by a predetermined natural number N. A program to make it work.

Images