阅读说明：本技术 情绪估计装置 (Mood estimation device ) 是由 佐藤宁 于 2018-03-08 设计创作，主要内容包括：提供一种情绪估计装置,能够使用非接触式的脉搏检测技术来精确地估计测定对象者的情绪、精神状态。情绪估计装置针对将心跳信号进行数字化所得到的数字生物体数据,按RR间期提取出一个周期的数据之后,强制进行重采样处理,并由DCT变换处理部获得谐波分量的系数。针对这些谐波分量的系数,利用LPF去除AC分量来获得系数数据列。将该系数数据列与作为表示情绪或心理状态的特征量的词典数据组进行比较来计算相似度,由此根据受验者的心跳信号来估计出受验者的情绪或心理状态。(A kind of mood estimation device is provided, is able to use contactless pulse detection technology to accurately estimate mood, the state of mind of measure object person.Mood estimation device, which is directed to, digitize obtained digital biometric volume data for heartbeat signal, after the data that a cycle is extracted by RR interphase, forces to carry out resampling processing, and obtain the coefficient of harmonic component by dct transform processing unit.For the coefficient of these harmonic components, AC component is removed using LPF and is arranged to obtain coefficient data.The coefficient data is arranged and is compared to calculate similarity with the dictionary data group as the characteristic quantity for indicating mood or psychological condition, the mood or psychological condition of subject are estimated thus according to the heartbeat signal of subject.)

1. a kind of mood estimation device, has:

Resampling processing unit, from heartbeat signal is transformed in the obtained digital biometric volume data of numerical data extract with RR The data of a cycle on the basis of interphase, and the data of a cycle are transformed to pre-determined sample number；

Dct transform processing unit, to the standardized digital organism data for securing sample number by the resampling processing unit Implement discrete cosine transform；

Low-pass filter group is arranged for the coefficient data exported by the dct transform processing unit, is reduced by each coefficient data AC compounent；And

Estimate processing unit, the coefficient data obtained from the low-pass filter group is arranged and is directed to every kind of mood or psychology The i.e. dictionary data group of the aggregate for the dictionary data that state is produced is compared, to estimate most like mood or psychological shape State.

2. mood estimation device according to claim 1, which is characterized in that

The resampling processing unit is directed to the data in the one period extracted from the digital biometric volume data, is based on The peak value of R wave carries out automatic gain adjustment.

Technical field

The present invention relates to a kind of mood estimation devices of mood that people is estimated according to pulse signal.

Background technique

Many practitioner, technical staff are developing the fatigue state of the driver of detection driving a kind of, are mentioning Preceding estimation precarious position and the anti-locking system of fatigue driving sounded an alarm to driver.

Patent Document 1 discloses a kind of vehicles of mood that motroist is estimated according to the expression of motroist and pulse Control device is coped with occupant's mood.

Patent Document 2 discloses a kind of biological signal detectors for detecting pulse in a non contact fashion.

Summary of the invention

Problems to be solved by the invention

It is known that the technology of fatigue is detected according to pulse.

The pulse of people is about 1Hz or so, but the fluctuation with 0.1Hz~0.3Hz or so.When the pulse to people carries out When Fourier transform, the component of the fluctuation can be extracted.

It is known that the wave component of 0.3Hz is as caused by the breathing of people.When people sucks oxygen, due to the amount of the oxygen Increase and so that the saltation velocity of heart is slightly become faster.

It is known that the wave component of 0.1Hz is as caused by the blood pressure of people.

It is known that when the size relation of the wave component of wave component and 0.1Hz to 0.3Hz is analyzed, energy Enough know degree of fatigue, the tense situation of people.

But when obtaining the size relation of the wave component, with the proviso that needing accurately to obtain pulse signal.

If it is the electrocardiographic wave for using the electrode for being pasted on skin to obtain, then more accurately waveform can be obtained, But using media such as electric wave, light to obtain pulse, the position (phase) of peak on a timeline is easy to happen partially It moves.Then, the offset at the peak becomes the biggish noise for wave component, to be difficult to obtain accurately fluctuation point Amount.

Therefore, the reliability of the measurement result of the fatigue strength measuring device constituted using such prior art etc. is low. That is it is low using the reliability of the fatigue strength detection of the pulse detection technology progress using electric wave, if only adopted as former state With the prior art, then accurate mood estimation can not be carried out for measure object person, and then cannot achieve fatigue driving and prevent and be System.

The present invention is to complete in view of the above-mentioned problems, is able to use contactless arteries and veins its purpose is to provide a kind of Detection technique fight to accurately estimate the mood estimation device of the mood of measure object person, the state of mind.

The solution to the problem

To solve the above-mentioned problems, mood estimation device of the invention has: resampling processing unit, from by heartbeat signal It is transformed to extract the data of a cycle on the basis of RR interphase in the obtained digital biometric volume data of numerical data, and will The data of a cycle are transformed to pre-determined sample number；Dct transform processing unit is fixed to by resampling processing unit The standardized digital organism data of sample number implements discrete cosine transform；Low-pass filter group, for by dct transform The coefficient data column of reason portion output, reduce AC compounent by each coefficient data；And estimation processing unit, it will be from low-pass filtering The coefficient data column and the aggregate i.e. dictionary number for the dictionary data produced for every kind of mood or psychological condition that device group obtains It is compared according to group, to estimate most like mood or psychological condition.

The effect of invention

In accordance with the invention it is possible to provide it is a kind of be able to use contactless pulse detection technology accurately estimate measurement The mood estimation device of the mood of object, the state of mind.

Problem, structure and effect other than the above are clear by the explanation of following embodiment.

Detailed description of the invention

Fig. 1 is the outside drawing of mood estimation device involved in embodiments of the present invention.

Fig. 2 is the block diagram for showing the hardware configuration of mood estimation device.

Fig. 3 is the block diagram for showing the software function of mood estimation device.

Fig. 4 is the change for showing each harmonic component of beats and heartbeat when certain subject experiences the haunted house of recreation ground The curve graph of change.

Fig. 5 is the beats and the heart shown when another subject experiences freely falling body (the free fall) of recreation ground The curve graph of the variation for each harmonic component jumped.

Specific embodiment

[mood estimation device: summary]

Inventor, which gives up according to the waveform of pulse, obtains this previous method of wave component, and contemplates according to progress The method for standardizing obtained pulse wave to obtain harmonic component.

In the waveform of the pulse of actual people, before and after next to the ventricular contraction for generating R wave, small is micro- Vibrational waveform persistently generates.The titles such as P, Q, S, T are assigned to these vibrational waveforms.

Therefore, it when carrying out frequency analysis including these micro-vibration waveforms, can observe using RR interphase as fundamental wave Harmonic component.

When constantly observing the harmonic component it is found that people is in scaring, when pleasant state, anxiety etc., harmonic component Ratio changes significantly.

For example, being noted the scene of lie in certain game or travelling in roller-coaster, the haunted house etc. for participating in recreation ground In the scene of facility etc., aggravate to the bounce " rub-a-dub rub-a-dub " of heart sometimes.Should " rub-a-dub rub-a-dub " such aggravation beated be presented on heartbeat Harmonic component in.

Therefore, by obtaining the harmonic component of heartbeat, and the ratio of the component is classified and is carried out by mood, state Analysis, mood, the psychological condition of people can be estimated without using brain wave.

[mood estimation device: synoptic diagram and outside drawing]

Figure 1A is mounted in the synoptic diagram of the mood estimation device 101 of automobile.

Biological signal detector 102 irradiates electric wave from antenna 103 to driver 104, and receives the reflection of the electric wave Wave.Moreover, implementing defined signal processing by using radiated wave and back wave, to detect the pulse signal of driver 104.It should The technology of biological signal detector 102 is proposed via the present inventor (referring to patent document 2).

Mood estimation device 101 is carried out when receiving signal of organism from biological signal detector 102 in inside Aftermentioned A/D transformation and data processing, to estimate mood, the psychological condition of driver 104.

The mood estimation result information that mood estimation device 101 exports is sent in the same manner as mood estimation device 101 The anti-locking system of fatigue driving (not shown) being equipped on automobile, and integrated with other information, synthetically to determine to drive The state of member 104.Be judged as it is necessary to when, measure that the anti-locking system of fatigue driving is implemented to sound an alarm driver 104 etc..

Figure 1B is the outside drawing for showing an example of the pulse measuring instrument 111 with wrist-watch form.

Fig. 1 C is the outside drawing for showing the back side of the pulse measuring instrument 111 with wrist-watch form.

In recent years, the pulse measuring instrument 111 of Wristwatch-type is popularized on the market.Its main part 112 in wrist-watch form Back side setting issue green light LED 113 and light receiving element 114, by the light-receiving intensity generated due to blood flow Variation is detected as pulse signal.Operating principle is identical as pulsation BOLD contrast (the pulse oximeter) that is worn on ear-lobe.That is, Pulse measuring instrument 111 is one kind of biological signal detector.Thus, additionally it is possible to it will be involved by embodiments of the present invention Mood estimation device 101 be embedded in the pulse measuring instrument 111 of the Wristwatch-type.

[mood estimation device 101: hardware configuration]

Fig. 2 is the block diagram for showing the hardware configuration of mood estimation device 101.

The mood estimation device formed by computers such as one board microcomputers (one-board microcomputer) 101 have the CPU 202 connecting with bus 201, ROM 203, RAM 204, nonvolatile memory 205.As non-volatile Memory 205, such as use flash memory.The flash memory can also be also used as ROM 203.

In bus 201, biological signal detector 102 is also connected with by A/D converter 206.That is, organism is believed The simulation signal of organism that number detection device 102 exports is transformed to digital biometric volume data by A/D converter 206.But from The directly output of biological signal detector 102 carries out simulation signal of organism to digitize obtained digital biometric volume data In the case where, do not need A/D converter 206.

In addition, according to the form of mood estimation device 101, additionally it is possible to which display unit 207, operation portion 208 are connected to bus 201, the estimated result of mood or psychological condition is presented to user.

Like this, mood estimation device 101 can be realized by general microcomputer.Mood estimation device 101 Entity be the information processing function provided by program.Thus, additionally it is possible to the program function of mood estimation device 101 is tied up It is scheduled in the software for the microcomputer of a part for constituting biological signal detector 102, is detected with signal of organism Device 102 is integrally formed.

[mood estimation device 101: software function]

Fig. 3 is the block diagram for showing the software function of mood estimation device 101.

The simulation signal of organism that biological signal detector 102 exports is transformed to digital biometric by A/D converter 206 Volume data.Digital biometric volume data is temporarily stored in organism data buffer 301.Organism data buffer 301 has There is the memory capacity for saving minimum two digital biometric the volume datas more than period.

Synchronous test section 302 reads in the digital biometric volume data for being stored in organism data buffer 301, and detects conduct The R wave at the peak of digital biometric volume data, i.e. pulse data.It is then detected that indicating the RR interphase of a cycle of heartbeat.

The RR interphase information that synchronous test section 302 detects is input into resampling processing unit 303.Resampling processing unit 303 read in RR interphase information and organism data buffer 301, and adopt again for the implementation of the digital biometric volume data of a cycle Sample processing and standardizations that (Auto Gain Control is " AGC " below) 304 progress are adjusted using automatic gain.

The period of the pulse of people is various according to factors such as health status, active states.Resampling processing unit 303 will There are the digital biometric volume data of a cycle of deviation, forcibly application is the data of fixed sample number, and is written to and adopts again Sample buffer 305.For the ease of the processing of aftermentioned dct transform processing unit 306, the sample number of resampling buffer 305 is preferably 2 Power.For example, 256 samples, 512 samples or 1024 samples.

In addition, the simulation signal of organism that biological signal detector 102 exports is not necessarily its amplitude by standard Change.Especially using contactless biological signal detector 102 of electric wave, signal of organism is simulated Amplitude is possible to the transmitting and receiving environment according to electric wave and changes.Therefore, by having in the inside of resampling processing unit 303 The data for being stored in resampling buffer 305 are standardized by the function of AGC 304 for amplitude, thus in aftermentioned DCT The intensity of correct harmonic component can be obtained in conversion process portion 306.

The standardized digital organism data for being stored in resampling buffer 305 is read into dct transform processing unit 306.

Dct transform processing unit 306 implements known discrete cosine transform, to export the sample with resampling buffer 305 The equal coefficient data column of number.

In addition, harmonic component refers to the frequency component of the integral multiple of fundamental wave.In other words, except the frequency of the integral multiple of fundamental wave point Frequency component not instead of harmonic component other than amount, noise.Therefore, the coefficient data exported for dct transform processing unit 306 Column, by each coefficient setting low-pass filter (being below " LPF ").

The coefficient data of fundamental wave is input into 0 rank LPF 307a, and the coefficient data of 1 subharmonic is input into 1 rank LPF 307b, the coefficient data of 2 subharmonic are input into 2 rank LPF 307c, and the coefficient data of nth harmonic is input into n rank LPF 307n.Hereafter, it unites to 0 rank LPF 307a, 1 rank LPF 307b, 2 rank LPF 307c, n rank LPF 307n In the case where title, referred to as LPF group 307.The cutoff frequency for constituting each LPF (low-pass filter group) of LPF group 307 is 0.1Hz Left and right.

In this way, coefficient data column 308 obtained from each LPF Jing Guo LPF group 307 are estimated as estimation processing unit 309 It is closest with which coefficient data mode for including in dictionary data group 310.

Dictionary data group 310 has multiple coefficient datas related with various moods, the psychological condition of measured object Mode.Such as excited, unhappy, terrified, fatigue state etc..

Coefficient data column are one-dimensional ordered series of numbers.Similarity between the one-dimensional ordered series of numbers that estimation has identical element number When, such as enumerate and calculate residual sum of squares (RSS) etc..

Estimation processing unit 309 is come to calculate phase to coefficient data column 308 by each coefficient data mode of dictionary data group 310 Like degree, using mood related with the coefficient data mode of similarity highest (the distance between ordered series of numbers is short), psychological condition as estimating Count result output.

[mood estimation device 101: movement]

As already described, the acquisition heartbeat of mood estimation device 101 involved in embodiments of the present invention Harmonic component and the technology for capturing its feature are important key technologies.Therefore, how accurately inventor has thought deeply to obtain The harmonic component of heartbeat.

As described above, in contactless heartbeat detection technology, noise component(s) is very more.Heartbeat just has wave originally Dynamic component, due also to organism activity and so that heartbeat is become faster or slack-off.

But the harmonic component of heartbeat refers to the frequency of the integral multiple of the fundametal compoment (RR interphase) of heartbeat.That is, The frequency component for the integral multiple of fundametal compoment can not cast out as noise all.It is also noise, energy due to fluctuating It is enough that the wave component is initially being removed by standardization.

Firstly, carrying out the detection synchronous with RR interphase of the pulse signal in multiple periods by resampling processing unit 303.And And resampling is carried out by each period forcing.No matter pulse be it is fast or slow, all forcibly make the sample value of pulse cycle in a word Unanimously.Then, the number of the sample value of whole pulse cycles is consistent.Sample number is, for example, 256 samples.It is handled by resampling Portion 303 forcibly makes the period of heartbeat consistent, thus wave component of the removal as noise from pulse.Then, more clearly Show the harmonic component for including in pulse sample.

Then, the pulse sample after being removed noise is provided to dct transform processing unit 306, in dct transform processing By carry out discrete cosine transform (DCT) in portion 306.For example, can be obtained when the pulse sample to 256 samples implements DCT The coefficient data column 308 of identical sample number (256).0 rank be fundamental wave, 1 rank is 2 times of harmonic waves, 2 ranks be 3 times of harmonic waves, 255 ranks are that 256 times of harmonic waves are sequentially connected in this way.

It removes since wave component should be resampled processing unit 303, is exported from dct transform processing unit 306 Harmonic component is rendered as DC component.That is, the AC component of non-DC component is noise.Therefore, apply for these components and use In the LPF of removal wave component.In this way, coefficient number corresponding with the harmonic component of pulse can be obtained via LPF group 307 According to column 308.By above processing, the small harmonic component for including in signal of organism can be clearly showed.

Inventor copies the formant of sound and the coefficient data column 308 is known as organism formant.

That is, mood estimation device 101 involved in embodiments of the present invention is conceived to and had not paid close attention to so far The harmonic component of heartbeat, and extract the harmonic component.It is in the prior art as detection during extracting the harmonic component The wave component of the heartbeat of object is almost completely neglected as the noise for interfering detection harmonic component and disregards and be removed.

[experimental result]

Fig. 4 is the change for showing each harmonic component of beats and heartbeat when certain subject experiences the haunted house of recreation ground The curve graph of change.Horizontal axis is the time, and the longitudinal axis is the size of harmonic component.

The time point T401 of monster is encountered in the curve graph of Fig. 4 it is found that in haunted house, the pulse rate of subject is sharply Rise, and produces significant change (waveform P402) at 8 subharmonic of heartbeat.

Fig. 5 is each harmonic wave point for showing beats and heartbeat when another subject experiences the freely falling body of recreation ground The curve graph of the variation of amount.

In the curve graph of Fig. 5, the heartbeat of the subject when the cockpit portion for being loaded with subject starts mobile is clearly known The situation (time point T501, waveform P502) that sharp rises of 8 subharmonic.And know, hereafter with slowly on tower top It rises, 8 subharmonic of the heartbeat of subject are gently reduced.This is considered as initial subject while cockpit portion starts mobile Feel that terrified, terrified later mood is reduced as cockpit portion continues rising.

Then, when cockpit portion is begun to decline, emotional change (time point point T503) sharply is generated in a short time.Though So be difficult to distinguish in Fig. 5, but specially 1 subharmonic, 3 subharmonic and 8 subharmonic sharply decline, 2 subharmonic, 4 subharmonic, 5 subharmonic sharply increase.Then, 8 subharmonic sharply increase (waveform P504).The increase is presumed to subject and feels very emerging It puts forth energy.

As shown in Figure 4 and Figure 5 it is found that the variation of the mood, psychological condition of the harmonic component and people of heartbeat accordingly generates Significant change.Therefore, these data are collected in large quantities, and are classified by every kind of mood to these data.Moreover, preparing more A and related coefficient data column 308 of identical mood, psychological condition carry out multiple coefficient data column 308 for example minimum The calculation process such as square law, to make dictionary data.

Estimate that processing unit 309 is directed to some coefficient data column 308, the multiple words for including in operation and dictionary data group 310 The distance between allusion quotation data column.Mood related with the smallest dictionary data of the value in calculated distance or psychological condition are feelings The estimated result of thread or psychological condition.

The estimation processing unit 309 can apply multiple regression analysis, principal component analysis, Bayesian Estimation, support vector machines Etc. various algorithm for estimating.

Mood estimation device 101 is disclosed in the embodiments of the present invention.

Mood estimation device 101 is directed to and carries out heartbeat signal to digitize obtained digital biometric volume data, between RR After phase extracts the data of a cycle, force to carry out resampling processing, and obtain harmonic wave point by dct transform processing unit 306 The coefficient of amount.Then, for the coefficient of harmonic component, AC component is removed using LPF to obtain coefficient data column 308.It is by this Number data column 308 are compared to calculate similar to the dictionary data group 310 as the characteristic quantity for indicating mood or psychological condition Degree, the mood or psychological condition of subject are estimated thus according to the heartbeat signal of subject.

It, can be real in a non contact fashion by utilizing mood estimation device 101 involved in the embodiments of the present invention The detection etc. of the security risk of driver in existing automobile.

More than, embodiments of the present invention are illustrated, but the present invention is not limited to above embodiment, not It is detached from the range of the gist of the invention documented by claims, including other variations, application examples.

For example, about above-mentioned embodiment, in order to should be readily appreciated that mode of the invention illustrates, in detail and specifically The structure of device and system is illustrated, but is not necessarily limited to the mode for the entire infrastructure for having illustrated.In addition, can be by certain A part of the structure of a embodiment is replaced into the structure of other embodiment, and can also be to the knot of some embodiment The structure of structure addition other embodiment.In addition it is possible to carry out other structures to a part of the structure of each embodiment Addition/deletion/displacement.

In addition, about above-mentioned each structure, function, processing unit etc., it can also be for example, by designing it in integrated circuits Part or all etc. come with hardware realization.In addition, above-mentioned each structure, function etc. can also be by for by processors It explains and executes the software for the program for realizing each function to realize.For realizing the information of the program of each function, table, file etc. It is able to maintain in volatibility or non-volatile storages such as memory, hard disk, SSD (Solid State Drive: solid state hard disk) In the recording mediums such as device or IC card, CD.

In addition, necessary line is thought in the expression of control line, information wire in explanation, all controls are not necessarily shown on product Line processed, information wire.In fact, it is also assumed that almost all of structure is connected with each other.

Description of symbols

101: mood estimation device；102: biological signal detector；103: antenna；104: driver；111: pulse Measurement device；112: main part；113:LED；114: light receiving element；201: bus；202:CPU；203:ROM；204:RAM； 205: nonvolatile memory；206:A/D converter；207: display unit；208: operation portion；301: organism data buffer； 302: synchronous test section；303: resampling processing unit；304:AGC；305: resampling buffer；306:DCT conversion process portion； 307:LPF group；308: coefficient data column；309: estimation processing unit；310: dictionary data group.