CN103230267A - Anti-movement-interference extraction method for pulse rates - Google Patents
Anti-movement-interference extraction method for pulse rates Download PDFInfo
- Publication number
- CN103230267A CN103230267A CN2013101764588A CN201310176458A CN103230267A CN 103230267 A CN103230267 A CN 103230267A CN 2013101764588 A CN2013101764588 A CN 2013101764588A CN 201310176458 A CN201310176458 A CN 201310176458A CN 103230267 A CN103230267 A CN 103230267A
- Authority
- CN
- China
- Prior art keywords
- pulse frequency
- frequency value
- value
- prediction
- similar
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Abstract
The invention discloses an anti-movement-interference extraction method for pulse rates. According to the method, the anti-movement-interference performance of pulse rate obtaining can be improved greatly, and pulse rates can be extracted accurately in a movement state. Frequency pairs with approximate double relations are searched from peak values of pulse frequency spectra, frequency values with relations with pulse prediction values accordant with set approximate conditions are searched from the searched frequency pairs, and optimal pulse rates are determined from the frequency values; and the selected optimal pulse rates are used as pulse rate prediction values used for a next round of pulse rate extraction. Accuracy, stabilities and real-time properties of pulse rates can be guaranteed when the anti-movement-interference extraction method is applied to a clinical and portable monitor, and a good practical value is provided.
Description
Technical field
The present invention relates to the pulse frequency extractive technique in the field of medical technology, be specifically related to a kind of pulse frequency extracting method of anti-motion artifacts.
Background technology
Pulse signal is the human-body biological medical signals, belong to small-signal, it is little and be subject to the defective of interference effects such as random noise, breathing, motion to have signal intensity, although use FIR filtering can filtering not the high and low frequency in human body pulse frequency scope disturb, filtering method can't the interference of filtering in human body pulse frequency scope.So extracting the pulse frequency value in containing the pulse signal of interference effect is to need the problem of solution badly.
The extracting method of pulse frequency mainly contains time domain approach and frequency domain method two big classes at present, and wherein time domain approach can be divided into difference method, extreme point method and other improved time domain approachs again.And the method dominant mechanism of conventional frequency domain is for obtaining the peaked position of frequency spectrum, and formula is passed through in the peaked position of the frequency spectrum that gets access to is converted to corresponding pulse frequency value.
Under normal (peace and quiet) situation, above-mentioned various time domains and frequency domain method can obtain pulse frequency value more accurately.And under the motion artifacts situation, especially under the situation of some interference ratio big (talent with special skills is selected as choice such as pilot, athlete and daily outdoor exercise motion etc.), for time domain approach, motion artifacts can occur and flood the pulse signal waveform of human body substantially on time domain waveform, any in the case time domain approach all is difficult to extract correct pulse frequency value; For frequency domain method, though on frequency domain, also there is the information of human pulse signal, disturb corresponding spectrum amplitude value will be far longer than the spectrum amplitude value of pulse signal, common frequency domain method is also helpless to this.
Generally speaking; extract the problem of anti-motion artifacts aspect at pulse frequency and blood oxygen saturation; under the stronger situation of motion artifacts; above-mentioned time domain approach and frequency domain method all extract less than real-time, accurate, stable pulse frequency and oximetry value; the remedial measure of taking generally also can be pulse frequency and blood oxygen saturation value protection; be current pulse frequency value can't obtain the time, use previous pulse frequency to replace.The greatest problem that this kind measure exists is exactly, if under the stronger situation of long-time interference ratio, if adopt the pulse frequency protection always, though have stability, obviously do not satisfied the accuracy of human body monitoring equipment and the requirement of real-time.
Summary of the invention
In view of this, the invention provides a kind of pulse frequency extracting method of anti-motion artifacts, improved the interference free performance that pulse frequency is obtained greatly.
The pulse frequency extracting method of this anti-motion artifacts comprises the steps:
Step 1, be converted to spectrum signal after gathering pulse signal and obtaining its alternating component;
Wherein, step 3 is specially: the frequency that step 2 search is obtained is to being designated as [A, B], and wherein, B is approximately two times of A, calculates B/2=C, and A is stored into array TA with C formation approximate frequency to [A, C];
In array TA, exist and only exist 1 [A, C] met A all to be similar to C and to predict the pulse frequency value, and the range value of A and C position correspondence all the amplitude than other positions is big, then current is non-kinestate, from A and C the two to choose one be best pulse frequency value; Otherwise current have a motion artifacts, finds to meet [the A that A and C all are similar to prediction pulse frequency value, C], and meet [A, the C] that A or C more are similar to prediction pulse frequency value, from the A that is similar to prediction pulse frequency value that finds and C and the A that more is similar to prediction pulse frequency value that finds or C, choose best pulse frequency value;
The described condition that is similar to prediction pulse frequency value is: data are closed within the scope in one of prediction pulse frequency value;
The described condition that is similar to prediction pulse frequency value more is: data are closed within the scope more for a short time prediction pulse frequency value.
Preferably, in described step 2, deposit the peak value frequency value corresponding in the spectrum signal in peak value array Pm; In described step 3; all [A in array TA; C] in A and the C length that all is not similar to prediction pulse frequency value or TA array be 0 o'clock; then travel through array Pm: be that Y/2 meets the described condition of predicting the pulse frequency value that is similar to more if there be and only exist 1/2nd of a value X or a value Y; choose then that to meet the X or the Y/2 that are similar to prediction pulse frequency value condition more be best pulse frequency value; otherwise take the pulse frequency protective measure, namely best pulse frequency value is continued to use the value of a circulation.
Preferably, in the step 3, further from the A that is similar to prediction pulse frequency value that finds and C and the A that more is similar to prediction pulse frequency value that finds or C, choose alternative pulse frequency value; In the time in this pulse frequency value leaching process, can't finding best pulse frequency value, further adopt the alternative pulse frequency value of determining in the circulation to substitute the prediction pulse frequency value repeating step three of this use; If still can not obtain best pulse frequency value, then take the pulse frequency protective measure, namely best pulse frequency value is continued to use the value of a circulation.
Beneficial effect:
(1) the present invention is based on the frequency multiplication relation that first-harmonic and second harmonic exist, and in conjunction with the characteristics that can not suddenly change usually between actual pulse frequency value consecutive frame, provide a kind of pulse frequency extracting method of anti-motion artifacts, improved the interference free performance that pulse frequency is obtained greatly.
(2) consider [A in the present invention, C] though in have only an A or C to satisfy approximate condition, but it also satisfies more harsh approximate condition more simultaneously, so with the chosen candidate value that this A or C also are further used as best pulse frequency, can improve best pulse frequency value like this and be selected to power.
(3) the present invention divides multiple situation that kinestate is discussed, and makes that the pulse frequency of obtaining under the kinestate is more targeted and accurate.
(4) the present invention also provides a processing mode of choosing best pulse frequency value according to peak value array Pm, as a supplement, can further improve accuracy of the present invention and success rate.
(5) the present invention also chooses an alternative pulse frequency value, carries out secondary and judges, has improved accuracy of the present invention, real-time.
(6) though the present invention will consider the measure of pulse frequency protection in the computational process of pulse frequency; but this measure only just can be used when disturbing very strong complexity, has just ensured accuracy, reliability, stability and the real-time of result of calculation most possibly at this.
(7) and, extracting method of the present invention is brief uncomplicated, and is not high to operand and request memory, so not only improved measuring speed, and can use in portable pulse frequency and blood oxygen saturation monitoring product, has good practical values.
Description of drawings
Fig. 1 is the pulse signal spectrum diagram under a kind of motion conditions.
Fig. 2 (a) and Fig. 2 (b) are time-domain signal oscillogram and the spectrogram of the non-motion pulse of the present invention.
Fig. 3～Fig. 6 is the situation 1 with motion artifacts)～4) the pulse signal spectrogram.
Fig. 7 is basic flow sheet of the present invention;
Fig. 8 (a) and Fig. 8 (b) are pulse signal time domain waveform figure and the spectrograms under the motion artifacts that produces of daily life;
Fig. 9 (a) and Fig. 9 (b) are pulse signal time domain waveform figure and the spectrograms under strenuous exercise disturbs.
The specific embodiment
Generally speaking, human pulse signal is the signal with periodic feature of beating and producing because of the rhythmicity of human heart, namely has the feature of periodic signal, and establishing human pulse signal is f (t), and then its Fourier expansion formula is:
Wherein, A
0Be constant term or DC component, A
nBe the amplitude of n harmonic wave,
Be the phase place of n harmonic wave, w
1Be fundamental frequency.
In actual applications, through FIR(the response of limit for length's unit impulse is arranged at the pulse time-domain signal, Finite Impulse Response) in filtering and the frequency spectrum doing to obtain after the FFT conversion, the frequency content of pulse signal mainly contains first-harmonic (first harmonic), second harmonic and triple-frequency harmonics.Fig. 1 is the pulse signal spectrum diagram under a kind of motion conditions.For convenient the analysis contrasts, the various piece in the sketch map is made the following assumptions:
Suppose that first-harmonic (being first harmonic) and second harmonic composition in the pulse signal frequency spectrum all exist, A among Fig. 1 and B are respectively the spectrum component of first harmonic and second harmonic correspondence, A is that the frequency values of first harmonic (is defined as a frequency multiplication here, refer to that it is a times of normal pulse rate), B is that the frequency values of second harmonic (is defined as two frequencys multiplication here, refer to that it is two times of normal pulse rate), for normal pulse signal, B is approximate two times of A.Be convenient and observe, make that the two corresponding amplitude Characteristics is spectrum signature under the normal condition.A among the figure
1, A
2, B
1And B
2Corresponding spectrum component is the spectrum component by the motion artifacts generation under the current motion conditions, because the randomness of motion artifacts, this has just caused the position distribution of spectrum component and amplitude size all to have randomness.
In the time domain pulse frequency extracting method under existing motion conditions, because pulse signal has been submerged in the motion artifacts signal usually, so be difficult to find the human pulse signal feature, therefore adopt time domain approach to be difficult to extract pulse frequency value accurately.
For the frequency domain extraction method of routine, common way is the search spectrum peak, and determines the pulse frequency value according to the peak-peak frequency value corresponding.For example in motion artifacts situation shown in Figure 1, because spectrum component is more, the peak-peak frequency value corresponding does not also correspond to real pulse frequency value.When therefore motion artifacts is stronger, determine that according to search frequency spectrum peak-peak the method for pulse frequency value is difficult to get access to pulse frequency value accurately.
In addition, other improved frequency domain extraction methods are searched for corresponding spectrum peak according to nominal pulse frequency value usually in frequency spectrum, and determine the pulse frequency value according to corresponding frequency values.This method can improve anti-interference to a certain extent, but under most of motion conditions, owing to be subjected to the influence of motion artifacts, the amplitude of first-harmonic is much smaller than the interfering frequency composition, if this moment, the interfering frequency composition was superimposed upon near the first-harmonic, then just can't get access to pulse frequency value accurately by the method.
The present invention is based on the frequency multiplication relation of first-harmonic and second harmonic existence, and in conjunction with the characteristics that can not suddenly change usually between actual pulse frequency value consecutive frame, has proposed a kind of pulse frequency extracting method of anti-motion artifacts, can realize the accurate extraction of pulse frequency under the motion conditions.Its basic procedure is:
Step 1, be converted to spectrum signal after gathering pulse signal and obtaining its alternating component;
The principle that arranges to a kind of preferable approximate condition in the above-mentioned flow process is described in detail below.
Observe pulse signal and be not difficult to find that the time-domain signal of non-motion pulse is shown in Fig. 2 (a), frequency-region signal is shown in Fig. 2 (b), and the pulse time domain waveform is clear, and spectrum component is also very clean, is mainly spectrum component and each harmonic component thereof of human pulse.In the case, only exist a pairing approximation two frequencys multiplication relation to [A in the pulse spectrum signal, B], just only there is a pair of [A in half C of calculating B, C], and the A and the C that are somebody's turn to do in [A, C] are very close, and all are similar to a prediction pulse frequency value, and the two corresponding range value all the amplitude than other positions is big, so adopt this condition can more accurately judge the pulse signal that does not have motion artifacts.Wherein, prediction pulse frequency value selects for use the pulse frequency value of previous circulation to extract the result, and the initial value of prediction pulse frequency value can select for use measurement to begin the pulse frequency value of obtaining when not moving.For non-motion pulse signal, because A and C ten minutes are close, almost equal, therefore adopt A or C all can as the pulse frequency value.
Yet the pulse signal under motion artifacts, there is a large amount of frequency domain interference compositions in it, just as the dotted line composition shown in Fig. 1, but the just a kind of situation wherein shown in Fig. 1, and is fairly simple a kind of.By a large amount of pulse signal samples is analyzed and can be learnt, according to the distributing position relation of the frequency content of various interference, when the pulse signal under the motion artifacts belongs to following wherein a kind of situation, can adopt the inventive method to determine the pulse frequency value:
1) two couples of [A, C] that meet approximate prediction pulse frequency value condition appear;
2) occur and a pair of [A, the C] that meets approximate prediction pulse frequency value condition only arranged, but A and C do not meet all conditions bigger than the amplitude of other positions of the two corresponding range value;
3) a pair of [A, the C] of approximate prediction pulse frequency value condition and the A in a pair of [A, C] or the C condition that meets more approximate prediction pulse frequency value in addition of meeting appears;
4) condition that A in a pair of [A, C] or C meet more approximate prediction pulse frequency value appears;
Wherein, being similar to the condition of predicting the pulse frequency value refers to: data are closed within the scope in one of prediction pulse frequency value; The condition that is similar to prediction pulse frequency value more refers to: data are closed within the scope more for a short time prediction pulse frequency value.
Below in conjunction with Fig. 3～Fig. 6 above-mentioned four kinds of situations are described.
Fig. 3 is situation 1), [A
1, B
1/ 2] and [A
2, B
2/ 2] all meet approximate condition, wherein [A
1, B
1/ 2] be actual normal pulse rate, [A
2, B
2/ 2] for disturbing, still owing to the frequency value difference between four data is very little, therefore can be from A
1, B
1/ 2, A
2, B
2If any selected best pulse frequency value of conduct in/2 is selected certainly near one of prediction pulse frequency value then be best selecting.
Fig. 4 is situation 2) a kind of, though occur and only have a pair of [A, the C] that meets approximate prediction pulse frequency value condition, A and C not to meet all conditions (here the component of other amplitudes not shown) bigger than the amplitude of other positions of the two corresponding range value, then can be from A
1, B
1If/2 any selected best pulse frequency values of conduct are selected certainly near one of prediction pulse frequency value then be best selecting.
Fig. 5 is situation 3) a kind of, A wherein
2(not being shown among the figure) be not though satisfy approximate condition, B
2/ 2 satisfy more close to the condition of prediction pulse frequency value, also can be used as a chosen candidate value of pulse frequency value, thereby are all finding comparatively accurate pulse frequency as much as possible in each circulation, rather than adopt the pulse frequency protective measure easily.From 3 chosen candidate values, select the best pulse frequency value of conduct to get final product.Perhaps from [A
1, B
1/ 2] select one near prediction pulse frequency value to be designated as M1 in, then from M1 and B
2Select one arbitrarily in/2, though optional reason is B here
2/ 2 from numerical value more approaching prediction pulse frequency value, but because A
2Therefore do not satisfy condition, make B
2/ 2 credibility decreases, and can't determine which is more close to true pulse frequency.
Fig. 6 is situation 4) a kind of owing to only exist B in a pair of [A, B] to be similar to prediction pulse frequency value, so get B
1/ 2 is the pulse frequency value.In this case; though for once harmonic wave or second harmonic meet approximate condition; because it meets more harsh more approximate condition; also can be used as a chosen candidate value of pulse frequency value; thereby in each circulation, all find comparatively accurate pulse frequency as much as possible, rather than adopt the pulse frequency protective measure easily.
Need to prove, consider the randomness that other interference components occur, it is not illustrated among Fig. 3～Fig. 6, but do not influence above-mentioned conclusion.
Based on above-mentioned analysis, can sum up, under the motion artifacts situation, may find and meet [the A that A and C all are similar to prediction pulse frequency value more than a pair of or a pair of, C], and meet [A, the C] that A or C more are similar to prediction pulse frequency value, so just choosing one the A that can predict the pulse frequency value from the A that is similar to prediction pulse frequency value that finds and C and more being similar to of finding or the C is best pulse frequency value.
In practice, above-mentioned four kinds of situations can contain the motion conditions of the overwhelming majority, and other situations are some phenomena, and for example three to [A, C] etc., but adopts above-mentioned strategy still can obtain the pulse frequency more accurate, real-time than prior art.
Further, in order to improve the success rate that pulse frequency is extracted, can from the chosen candidate value that finds, select one again as alternative pulse frequency value.Each is taken turns in the circulation, and in the time can't obtaining best pulse frequency value, the alternative pulse frequency value that can adopt determine a circulation time in current circulation is once judged and selected.If still can not obtain best pulse frequency value, take the pulse frequency protective measure again, namely best pulse frequency value is continued to use the value of a circulation.Certainly, if do not determine alternative pulse frequency value in the last circulation, then do not carry out this operation.Specific implementation process to this better embodiment is described in detail below, has also added the situation of seeking best pulse frequency value according to Pm in this better embodiment, has further improved pulse frequency value success rate of extracting.This flow process comprises the steps:
Step 1, adopt to set wavelength light beam and shine finger tip (finger tip or refer to abdomen), obtain pulse signal.
Peak value in step 3, the search spectrum signal is stored into peak value array Pm with the peak value frequency value corresponding.
In step 4, the search Pm array all to have the frequency of approximate two times of relations right, be designated as [A, B], wherein, B is approximately two times of A, half that calculate B is designated as C, will be similar to frequency [A, C] is stored into array TA.
Here, A and B are similar to two times being judged to be in [A, B]: A closes within the scope B/2's, and this scope of closing on can be identical or different with the Rule of judgment of aforementioned " being similar to prediction pulse frequency value ".
In this step, the approximate frequency among the TA is compared with prediction pulse frequency value one by one to [A, C].In the present embodiment, the initial value of prediction pulse frequency value is chosen the not pulse frequency value under the setting in motion situation of object when measuring beginning, and later prediction pulse frequency value is the best pulse frequency value that a last circulation is obtained.
The branch situation is determined best pulse frequency value:
A) in array TA, exist and exist 1 [A, C] met A all to be similar to C and to predict the pulse frequency value, and the range value of A and C position correspondence all the amplitude than other positions is big, then current state is non-kinestate, from A and C the two to choose one be best pulse frequency value.
Following situation b)～g) be kinestate:
B) when in array TA, having 2 to [A, C] when meeting A and C and all being similar to prediction pulse frequency value, then current state is kinestate, and one that chooses from two A and two C near prediction pulse frequency value is best pulse frequency value, chooses an alternative pulse frequency value that time approaches prediction pulse frequency value.
C) when in array TA, existing and only having 1 to [A, C] meet A and C and all be similar to prediction pulse frequency value, but A and C do not meet all conditions bigger than the amplitude of other positions of the two corresponding range value, then from A and C the two to choose one be best pulse frequency value, another is alternative pulse frequency value.
D) when in array TA, existing and having 1 to [A, C] meet A and C and all be similar to prediction pulse frequency value, when existing 1 the A in [A, C] or C also be similar to prediction pulse frequency value and its and be similar to prediction pulse frequency value more simultaneously, then current state is kinestate, from last [A, C] in choose A, the C one is approximation 1, the back is chosen among one [A, the C] and is met more that A or the C of approximate condition are approximation 2, choosing one then from approximation 1 and approximation 2 is best pulse frequency value, and another is alternative pulse frequency value.E) when in array TA, having and only have 1 to [A, C] in A or C meet the condition that is similar to prediction pulse frequency value more, then current state is kinestate, choose that to meet the A or the C that are similar to prediction pulse frequency value more be best pulse frequency value, and that alternative pulse frequency value is continued to use last circulation time determined alternative pulse frequency value is constant.Do not have alternative pulse frequency value when for the first time carrying out this method, and if last circulation still do not determine alternative pulse frequency value, then alternative pulse frequency value is still vacant here.
F) when in array TA all A and C length of all not being similar to prediction pulse frequency value or TA array in [A, C] be 0 o'clock, then travel through array Pm; If having and only exist 1/2nd of a value X or a value Y is that Y/2 meets the described condition that is similar to prediction pulse frequency value more, choose then that to meet the X or the Y/2 that are similar to prediction pulse frequency value condition more be best pulse frequency value, it is constant that alternative pulse frequency value is continued to use last circulation time determined alternative pulse frequency value, if in like manner alternative pulse frequency value is not still determined in last circulation, then alternative pulse frequency value is still vacant here, otherwise belongs to situation g).
G) a kind of mode is: directly take the pulse frequency protective measure, namely best pulse frequency value is continued to use the value of a circulation.It is constant that alternative pulse frequency value is continued to use last circulation time determined alternative pulse frequency value, if in like manner last the circulation still do not determined alternative pulse frequency value, then alternative pulse frequency value is still vacant here.
Another kind of mode is: earlier current prediction pulse frequency value is replaced with the alternative pulse frequency value of obtaining in the last circulation, repeated execution of steps three is to obtain best pulse frequency value; If still can not obtain best pulse frequency value, then take the pulse frequency protective measure again; If do not have alternative pulse frequency value in the current circulation, then directly take the pulse frequency protective measure yet.
The present invention is applicable to various situations, comprises the situation (shown in Fig. 8 (a) and Fig. 8 (b)) of the motion artifacts that daily life produces and the situation of disturbing in strenuous exercise (shown in Fig. 9 (a) and Fig. 9 (b)).
Fig. 8 (a) is the pulse signal time domain waveform figure under the motion artifacts that daily life produces, and Fig. 8 (b) is corresponding diagram 8(a) the pulse signal spectrogram.By Fig. 8 (a) as can be seen; for conventional or other improved time domain approachs (main thought is by getting access to the pulse frequency value through certain method again behind the crest in the search pulse signal and the trough); can extracting more in the case, the pulse frequency value has compared difficulty; even the possibility that pulse frequency is protected can appear, its consequence is exactly to have run counter to pulse frequency and the real-time of blood oxygen saturation extraction and the requirement of accuracy.And by Fig. 8 (b) as can be seen, for conventional or other improved frequency domain methods (thereby main thought is obtained pulse frequency for the spectrum peak by the pulse frequency correspondence in certain method search frequency spectrum), the spectrum peak of pulse frequency correspondence can be easier obtained in the case, thereby accurate, real-time, reliable pulse frequency value can be obtained.
Yet under the situation that strenuous exercise disturbs, Fig. 9 (a) is the pulse signal time domain waveform figure under strenuous exercise's interference, and Fig. 9 (b) is corresponding diagram 9(a) the pulse signal spectrogram.By Fig. 9 (a) as can be seen, because motion artifacts has been flooded the pulse signal of human body fully on time domain, if think to use in the case routine or other improved time domain approachs obtain accurately, in real time, reliable pulse frequency and blood oxygen saturation almost do not had probability.And by Fig. 9 (b) as can be seen, the spectrum information of pulse frequency correspondence does not in the case have disturbed flooding, for conventional or other improved frequency domain methods, the spectrum information that extracts the pulse frequency correspondence from frequency spectrum is relatively more difficult, interfering frequency under the different condition also is not quite similar in addition, and use is conventional or other improved frequency domain methods extract accurately, real-time, reliable pulse frequency has also just become almost impossible thing.
By relative analysis as can be seen; at above-mentioned two type games disturbed conditions; routine or other improved time domains and frequency domain method all can not satisfy the requirement that extracts accurate, reliable, real-time, stable pulse frequency; especially under the situation that strenuous exercise disturbs; two class methods all are difficult to extract pulse frequency; but in order to guarantee stability requirement, can use the measure of pulse frequency protection usually, this measure can be used when some are forced.Yet, if under the situation that long-time strenuous exercise disturbs, accurate, reliable, the real-time requirement that the medical monitoring instrument has just been run counter in the measure of using pulse frequency to protect for a long time undoubtedly.
Anti-motion artifacts pulse frequency extracting method simple and fast of the present invention, various interference components in the filtering strenuous exercise and obtain pulse frequency have improved accuracy, the reliability, real-time and stable of pulse blood oxygen instrument widely effectively.Meanwhile, the present invention is because algorithm is simple, therefore especially be fit to be applied to volume small and exquisite, conveniently wear, in the high jamproof portable pulse blood oxygen instrument.
Need to prove, the employed approximate condition of above embodiment has all been considered following situation, i.e. [A, C] though in have only an A or C to satisfy approximate condition and its to satisfy more harsh approximate condition more, therefore this A or C also are further used as the chosen candidate value of best pulse frequency, can improve best pulse frequency value like this and be selected to power.In practice, also can not consider this situation, [A, the C] that only will meet approximate condition finds out, and therefrom chooses best pulse frequency value then.Though the result who chooses does not like this have preferred embodiment effective, still have more the ability of anti-motion artifacts than existing time domain and frequency domain extraction scheme.
In sum, more than be preferred embodiment of the present invention only, be not for limiting protection scope of the present invention.Within the spirit and principles in the present invention all, any modification of doing, be equal to replacement, improvement etc., all should be included within protection scope of the present invention.
Claims (5)
1. the pulse frequency extracting method of an anti-motion artifacts is characterized in that, comprises the steps:
Step 1, be converted to spectrum signal after gathering pulse signal and obtaining its alternating component;
Step 2, from the peak value of spectrum signal search to have the frequency of approximate two times of relations right;
Step 3, the frequency centering that obtains from search seek and the pulse frequency predictive value between relation meet the frequency values of setting approximate condition, and therefrom determine best pulse frequency value;
Step 4, the best pulse frequency value that will select are extracted employed pulse frequency predictive value as the next round pulse frequency.
2. the method for claim 1 is characterized in that, described step 3 specifically comprises:
The frequency that step 2 search is obtained is to being designated as [A, B], and wherein, B is approximately two times of A, is that C formation approximate frequency is stored into array TA to [A, C] with half of A and B;
In array TA, exist and only exist 1 [A, C] met A all to be similar to C and to predict the pulse frequency value, and the range value of A and C position correspondence all the amplitude than other positions is big, then current is non-kinestate, from A and C the two to choose one be best pulse frequency value; Otherwise current have a motion artifacts, finds to meet [the A that A and C all are similar to prediction pulse frequency value, C], and meet [A, the C] that A or C more are similar to prediction pulse frequency value, from the A that is similar to prediction pulse frequency value that finds and C and the A that more is similar to prediction pulse frequency value that finds or C, choose best pulse frequency value;
The described condition that is similar to prediction pulse frequency value is: data are closed within the scope in one of prediction pulse frequency value;
The described condition that is similar to prediction pulse frequency value more is: data are closed within the scope more for a short time prediction pulse frequency value.
3. method as claimed in claim 2 is characterized in that,
In described step 2, deposit the peak value frequency value corresponding in the spectrum signal in peak value array Pm;
In described step 3; all [A in array TA; C] in A and the C length that all is not similar to prediction pulse frequency value or TA array be 0 o'clock; then travel through array Pm: be that Y/2 meets the described condition of predicting the pulse frequency value that is similar to more if there be and only exist 1/2nd of a value X or a value Y; choose then that to meet the X or the Y/2 that are similar to prediction pulse frequency value condition more be best pulse frequency value; otherwise take the pulse frequency protective measure, namely best pulse frequency value is continued to use the value of a circulation.
4. as claim 2 or 3 described methods, it is characterized in that, in the step 3, further from the A that is similar to prediction pulse frequency value that finds and C and the A that more is similar to prediction pulse frequency value that finds or C, choose alternative pulse frequency value;
In the time in this pulse frequency value leaching process, can't finding best pulse frequency value, further adopt the alternative pulse frequency value of determining in the circulation to substitute the prediction pulse frequency value repeating step three of this use; If still can not obtain best pulse frequency value, then take the pulse frequency protective measure, namely best pulse frequency value is continued to use the value of a circulation.
5. the method for claim 1 is characterized in that, in described step 2, deposits the peak value frequency value corresponding in the spectrum signal in peak value array Pm; The frequency that search is obtained is to being designated as [A, B], and wherein, B is approximately two times of A, calculates B/2=C, and A is stored into array TA with C formation approximate frequency to [A, C];
Described step 3 comprises:
Approximate frequency among the TA is compared with prediction pulse frequency value one by one to [A, C], and the branch situation is determined best pulse frequency value:
A) in array TA, exist and exist 1 [A, C] met A all to be similar to prediction pulse frequency value with C, and the range value of A and C position correspondence all the amplitude than other positions is big, then judging current is non-kinestate, from A and C the two to choose one be best pulse frequency value; Following situation b)～g) be kinestate;
B) when existing 2 [A, C] met A and all be similar to prediction pulse frequency value with C in array TA, one that chooses from two A and two C near prediction pulse frequency value is best pulse frequency value; One that chooses time approaching prediction pulse frequency value is alternative pulse frequency value;
C) when in array TA, existing and only having 1 to [A, C] meet A and C and all be similar to prediction pulse frequency value, but A and C do not meet all conditions bigger than the amplitude of other positions of the two corresponding range value, then the two one of choosing near prediction pulse frequency value is best pulse frequency value from A and C, chooses another and is alternative pulse frequency value;
D) when in array TA, existing and only having 1 to [A, C] meet A and C and all be similar to prediction pulse frequency value, have 1 simultaneously to [A, C] in A or C when also being similar to prediction pulse frequency value and meeting and being similar to the condition of prediction pulse frequency value more, from last [A, C] in choose A, the C one is approximation 1, from back one [A, C] in choose and meet the A or the C that are similar to prediction pulse frequency value condition more and be defined as approximation 2, choosing one from approximation 1 and approximation 2 is best pulse frequency value, chooses another and is alternative pulse frequency value;
E) to meet the A or the C that are similar to prediction pulse frequency value more be best pulse frequency value when having and only have 1 the A in [A, C] or C met the condition that is similar to prediction pulse frequency value more in array TA, then choosing; Alternative pulse frequency value is continued to use the value of last circulation;
F) all [A in array TA, C] in A and the C length that all is not similar to prediction pulse frequency value or TA array be 0 o'clock, then travel through array Pm: be that Y/2 meets the described condition of predicting the pulse frequency value that is similar to more if there be and only exist 1/2nd of a value X or a value Y, choose then that to meet the X or the Y/2 that are similar to prediction pulse frequency value condition more be best pulse frequency value, alternative pulse frequency value is continued to use the value of last circulation; Otherwise belong to situation g);
G) adopt the alternative pulse frequency value of determining in the last circulation to substitute prediction pulse frequency value repeating step three earlier; If still can not obtain best pulse frequency value, then take the pulse frequency protective measure, namely best pulse frequency value is continued to use the value of a circulation, and alternative pulse frequency value is continued to use the value of last circulation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310176458.8A CN103230267B (en) | 2013-05-14 | 2013-05-14 | Anti-movement-interference extraction method for pulse rates |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310176458.8A CN103230267B (en) | 2013-05-14 | 2013-05-14 | Anti-movement-interference extraction method for pulse rates |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103230267A true CN103230267A (en) | 2013-08-07 |
| CN103230267B CN103230267B (en) | 2015-06-03 |
Family
ID=48878394
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310176458.8A Expired - Fee Related CN103230267B (en) | 2013-05-14 | 2013-05-14 | Anti-movement-interference extraction method for pulse rates |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103230267B (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104739399A (en) * | 2014-02-10 | 2015-07-01 | 北京金日吉通科贸有限公司 | Method for detecting movement offset in pulse wave heart rate calculation |
| CN105796115A (en) * | 2016-04-20 | 2016-07-27 | 广州视源电子科技股份有限公司 | Blood oxygen measurement method and system |
| WO2017107085A1 (en) * | 2015-12-23 | 2017-06-29 | Intel Corporation | Hmm-based adaptive spectrogram track method |
| WO2017214870A1 (en) * | 2016-06-15 | 2017-12-21 | 深圳迈瑞生物医疗电子股份有限公司 | Method for calculating physiological parameters and corresponding medical equipment |
| CN107510462A (en) * | 2017-09-06 | 2017-12-26 | 广州视源电子科技股份有限公司 | Blood oxygen measuring method and device |
| CN108090410A (en) * | 2017-11-15 | 2018-05-29 | 上海师范大学 | A kind of contactless facial rhythm of the heart Spinning and monitoring method |
| CN108882871A (en) * | 2016-04-15 | 2018-11-23 | 欧姆龙株式会社 | Biont information analytical equipment, system, program and Biont information analysis method |
| CN109009159A (en) * | 2018-07-24 | 2018-12-18 | 华中科技大学 | A kind of anti motion interference accelerating circuit suitable for wearable blood oxygen chip |
| CN110123297A (en) * | 2019-04-04 | 2019-08-16 | 平安科技（深圳）有限公司 | Method for measuring heart rate, device, computer equipment and storage medium |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4442845A (en) * | 1981-11-10 | 1984-04-17 | Stephens Frederick R N | Pulse curve analyser |
| CN101991410A (en) * | 2009-08-31 | 2011-03-30 | 深圳市理邦精密仪器股份有限公司 | Pulse rate searching and calculating method |
| CN102988036A (en) * | 2012-12-26 | 2013-03-27 | 中国科学院自动化研究所 | Method for measuring pulse rate |
-
2013
- 2013-05-14 CN CN201310176458.8A patent/CN103230267B/en not_active Expired - Fee Related
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4442845A (en) * | 1981-11-10 | 1984-04-17 | Stephens Frederick R N | Pulse curve analyser |
| CN101991410A (en) * | 2009-08-31 | 2011-03-30 | 深圳市理邦精密仪器股份有限公司 | Pulse rate searching and calculating method |
| CN102988036A (en) * | 2012-12-26 | 2013-03-27 | 中国科学院自动化研究所 | Method for measuring pulse rate |
Cited By (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104739399A (en) * | 2014-02-10 | 2015-07-01 | 北京金日吉通科贸有限公司 | Method for detecting movement offset in pulse wave heart rate calculation |
| CN108430320B (en) * | 2015-12-23 | 2021-06-15 | 英特尔公司 | HMM-based adaptive spectrogram tracking method |
| WO2017107085A1 (en) * | 2015-12-23 | 2017-06-29 | Intel Corporation | Hmm-based adaptive spectrogram track method |
| US11344211B2 (en) | 2015-12-23 | 2022-05-31 | Intel Corporation | HMM-based adaptive spectrogram track method |
| CN108430320A (en) * | 2015-12-23 | 2018-08-21 | 英特尔公司 | Adaptive spectrum figure tracking based on HMM |
| CN108882871A (en) * | 2016-04-15 | 2018-11-23 | 欧姆龙株式会社 | Biont information analytical equipment, system, program and Biont information analysis method |
| US11246501B2 (en) | 2016-04-15 | 2022-02-15 | Omron Corporation | Biological information analysis device, system, and program |
| US11617516B2 (en) | 2016-04-15 | 2023-04-04 | Omron Corporation | Biological information analysis device, biological information analysis system, program, and biological information analysis method |
| CN108882871B (en) * | 2016-04-15 | 2021-04-16 | 欧姆龙株式会社 | Biological information analysis device, system, program, and biological information analysis method |
| CN105796115B (en) * | 2016-04-20 | 2018-11-02 | 广州视源电子科技股份有限公司 | Blood oxygen measuring method and system |
| CN105796115A (en) * | 2016-04-20 | 2016-07-27 | 广州视源电子科技股份有限公司 | Blood oxygen measurement method and system |
| US11872060B2 (en) | 2016-06-15 | 2024-01-16 | Shenzhen Mindray Bio-Medical Electronics Co., Ltd. | Methods and systems for calculating physiological parameters |
| WO2017214870A1 (en) * | 2016-06-15 | 2017-12-21 | 深圳迈瑞生物医疗电子股份有限公司 | Method for calculating physiological parameters and corresponding medical equipment |
| US11154250B2 (en) | 2016-06-15 | 2021-10-26 | Shenzhen Mindray Bio-Medical Electronics Co., Ltd. | Methods and systems for calculating physiological parameters |
| CN107510462A (en) * | 2017-09-06 | 2017-12-26 | 广州视源电子科技股份有限公司 | Blood oxygen measuring method and device |
| CN107510462B (en) * | 2017-09-06 | 2020-05-05 | 广州视源电子科技股份有限公司 | Blood oxygen measuring method and device |
| CN108090410A (en) * | 2017-11-15 | 2018-05-29 | 上海师范大学 | A kind of contactless facial rhythm of the heart Spinning and monitoring method |
| CN109009159B (en) * | 2018-07-24 | 2020-06-02 | 华中科技大学 | Anti-motion interference accelerating circuit suitable for wearable blood oxygen chip |
| CN109009159A (en) * | 2018-07-24 | 2018-12-18 | 华中科技大学 | A kind of anti motion interference accelerating circuit suitable for wearable blood oxygen chip |
| CN110123297A (en) * | 2019-04-04 | 2019-08-16 | 平安科技（深圳）有限公司 | Method for measuring heart rate, device, computer equipment and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103230267B (en) | 2015-06-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103230267B (en) | Anti-movement-interference extraction method for pulse rates | |
| CN103989462B (en) | The extracting method of a kind of pulse wave fisrt feature point and second feature point | |
| CN102499694B (en) | Method for eliminating interference to blood oxygen saturation monitoring | |
| CN105997043B (en) | A kind of pulse frequency extracting method based on wrist wearable device | |
| CN105223576B (en) | A kind of line spectrum signal target automatic testing method based on single vector subsurface buoy | |
| CN104203089B (en) | Pulsation detector and electronic equipment | |
| Ahrabian et al. | A class of multivariate denoising algorithms based on synchrosqueezing | |
| CN101933810B (en) | A kind of method for detecting blood oxygen saturation | |
| Porporato et al. | Clues to the existence of deterministic chaos in river flow | |
| CN101991410B (en) | Pulse rate searching and calculating method | |
| CN109009085A (en) | A kind of R wave of electrocardiosignal detection method and system | |
| Parsa et al. | A comparison of high precision Fo extraction algorithms for sustained vowels | |
| CN105307095B (en) | A kind of high definition audio frequency measurement method based on FFT | |
| CN109350031A (en) | A kind of photoplethysmographic signal feature point detecting method and device | |
| Aboy et al. | Adaptive modeling and spectral estimation of nonstationary biomedical signals based on Kalman filtering | |
| CN106772572B (en) | A kind of pick-up method of micro-seismic monitoring first arrival | |
| CN101897578B (en) | Method for segmenting arterial pressure signal by beats | |
| CN110313903A (en) | A kind of pulse wave frequency domain character parameter extracting method and device | |
| CN106323330A (en) | Non-contact-type step count method based on WiFi motion recognition system | |
| Xu et al. | Digital filter design for peak detection of surface EMG | |
| WO2020061346A1 (en) | Methods and apparatuses for tracking weak signal traces | |
| Gong et al. | Nonlinear vibration feature extraction based on power spectrum envelope adaptive empirical Fourier decomposition | |
| CN108107393B (en) | A kind of spectral peak Credibility judgement method in frequency analysis | |
| CN103099642B (en) | Ultrasonic blood flow signal quality real-time analysis method | |
| Pham et al. | An adaptive computation of contour representations for mode decomposition |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20150603 Termination date: 20160514 |