CN109582713B

CN109582713B - Motion state identification method, motion state identification device and terminal

Info

Publication number: CN109582713B
Application number: CN201811455450.4A
Authority: CN
Inventors: 孟洋; 陈维亮
Original assignee: Goertek Techology Co Ltd
Current assignee: Goertek Techology Co Ltd
Priority date: 2018-11-30
Filing date: 2018-11-30
Publication date: 2023-05-19
Anticipated expiration: 2038-11-30
Also published as: CN109582713A

Abstract

The invention discloses a method for identifying a motion state, which is applied to the technical field of data processing and solves the problem that a walking state and a riding state cannot be accurately distinguished in the prior art, and the method comprises the following steps: acquiring data acquired by a motion sensor, and calculating a correlation coefficient by using the expectation of the data and the preset delay time when the motion state is determined to be a walking or riding fuzzy state according to the data; when the correlation coefficient is located in the first correlation coefficient interval, the motion state is a walking state; when the correlation coefficient is located in the second correlation coefficient interval, the motion state is a riding state; the method can accurately distinguish the walking state from the riding state by utilizing the correlation coefficient, thereby improving the accuracy of identifying the motion state; the invention also discloses a motion state identification device, a terminal and a computer readable storage medium, which have the beneficial effects.

Description

Motion state identification method, motion state identification device and terminal

Technical Field

The present invention relates to the field of data processing technologies, and in particular, to a method and apparatus for identifying a motion state, a terminal, and a computer readable storage medium.

Background

In the prior art, characteristics such as Fourier transform, wavelet transform, mean value, standard deviation and the like are generally adopted for identifying the motion state. However, these are only suitable for theoretical research, and in practical application, the quality of the motion state recognition result is completely dependent on the quality of the acquired data, especially in the fuzzy boundary region of the motion state, the motion state recognition result is inaccurate, so that the accuracy of motion state recognition needs to be improved.

Disclosure of Invention

The invention aims to provide a method, a device, a terminal and a computer readable storage medium for identifying a motion state, which can accurately distinguish a walking state from a riding state and further improve the accuracy of motion state identification.

In order to solve the above technical problems, the present invention provides a method for identifying a motion state, including:

acquiring data acquired by a motion sensor, and calculating a correlation coefficient by using the expectation of the data and the preset delay time when the motion state is determined to be a walking or riding fuzzy state according to the data;

when the correlation coefficient is located in the first correlation coefficient interval, the motion state is a walking state;

When the correlation coefficient is located in the second correlation coefficient section, the motion state is a riding state.

Optionally, before the calculating the correlation coefficient using the expected and preset delay time of the data, the method further includes:

calculating a first sample entropy using the data;

when the first sample entropy is located in the first sample entropy interval, the motion state is a walking state;

when the first sample entropy is located in the second sample entropy interval, the motion state is a riding state;

and when the first sample entropy is in a third sample entropy interval, executing the step of calculating the correlation coefficient by using the expected data and the preset delay time.

Optionally, before said calculating the first sample entropy using said data, further comprising:

calculating a first kurtosis value using the data;

when the first kurtosis value is positioned in the first kurtosis value interval, the motion state is a walking state;

when the first kurtosis value is located in the second kurtosis value interval, the motion state is a riding state;

and when the first kurtosis value is in the third kurtosis value interval, executing the step of calculating the first sample entropy by using the data.

Optionally, before said calculating the first kurtosis value using said data, further comprising:

calculating a stationarity parameter using the data;

executing the step of calculating a first kurtosis value by using the data when the stationarity parameter is located in a first stationarity parameter interval;

when the stability parameter is located in a second stability parameter interval, calculating a second sample entropy by using the data; when the second sample entropy is larger than a sample entropy threshold value, the motion state is a riding state; and when the second sample entropy is smaller than the sample entropy threshold, the motion state is a walking state.

Optionally, the calculating the stability parameter using the data includes:

acquiring the total acceleration of the triaxial acceleration corresponding to a preset window, and determining the minimum total acceleration Min1 of the preset window;

using the formula

Calculating a total acceleration Average value Average1;

determining the minimum combined acceleration Min2q of each sliding small window in the preset window;

using the formula

Calculating a combined acceleration Average value Average2q of each sliding small window;

determining a maximum value Max and a minimum value Min from the total acceleration Average value Average1 and the total acceleration Average value Average2q of each sliding small window;

Taking the ratio Max/Min of Max and Min AS a stability parameter AS;

wherein n is the number of combined acceleration corresponding to the preset window, X (i) is the ith combined acceleration, min2q is the minimum combined acceleration of the q-th sliding small window in the preset window, xq (m) is the m-th combined acceleration of the q-th sliding small window in the preset window, and l is the number of combined acceleration of the sliding small window.

Optionally, when determining that the motion state is running or riding blur state according to the data, the method further comprises:

calculating a second kurtosis value using the data;

when the second kurtosis value is smaller than the first kurtosis threshold value, the exercise state is a running state;

when the second kurtosis value is greater than the first kurtosis threshold, the motion state is a riding state.

Optionally, the determining that the motion state is a walking or riding fuzzy state according to the data includes:

calculating the total acceleration of the obtained triaxial acceleration, and calculating the standard deviation of the total acceleration;

when the standard deviation is in the first standard deviation interval, the motion state is a walking or riding fuzzy state.

Optionally, the calculating the correlation coefficient using the expected and preset delay time of the data includes:

Using the formula a (τ) =e [ (X _t -μ)(X _t+τ -μ)]Respectively calculating a correlation parameter A (0) with a preset delay time of 0 and a correlation parameter A (n) with a preset delay time of n;

taking the ratio A (0)/A (n) of A (0) to A (n) as a correlation coefficient A;

wherein τ is the delay time, X _t For the resultant acceleration, X, obtained within a preset time period t _t+τ For the resultant acceleration obtained by delaying τ for a preset period of time t, μ is X _t E is the desired operation.

The invention also provides a motion state identification device, which comprises:

the correlation coefficient calculation module is used for acquiring data acquired by the motion sensor, and calculating a correlation coefficient by utilizing the expectation of the data and the preset delay time when the motion state is determined to be a walking or riding fuzzy state according to the data;

the first classification module is used for enabling the motion state to be a walking state when the correlation coefficient is located in a first correlation coefficient interval; when the correlation coefficient is located in the second correlation coefficient section, the motion state is a riding state.

The invention also provides a terminal, comprising:

the motion sensor is used for collecting data;

a memory for storing a computer program;

and the processor is used for realizing the steps of the motion state identification method when executing the computer program.

The present invention also provides a computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of the above-described method of identifying a state of motion.

The invention provides a method for identifying a motion state, which comprises the following steps: acquiring data acquired by a motion sensor, and calculating a correlation coefficient by using the expectation of the data and the preset delay time when the motion state is determined to be a walking or riding fuzzy state according to the data; when the correlation coefficient is located in the first correlation coefficient interval, the motion state is a walking state; when the correlation coefficient is located in the second correlation coefficient interval, the motion state is a riding state.

Therefore, the correlation coefficient is utilized to distinguish the walking state and the riding state, and as the correlation coefficient characterizes the correlation degree of the same event between two different periods, the correlation of the same event at different times can be reflected, the movement processes of the walking and the riding are different, namely, the events are different, and the calculated correlation coefficient is not the same, so that the correlation coefficient is calculated by utilizing the expectation of the acquired data and the preset delay time, the correlation degree of the same event between two different periods can be obtained, the walking state and the riding state can be accurately distinguished, and the movement state identification accuracy is improved; the invention also provides a motion state identification device, a terminal and a computer readable storage medium, which have the beneficial effects and are not repeated here.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are required to be used in the embodiments or the description of the prior art will be briefly described below, and it is obvious that the drawings in the following description are only embodiments of the present invention, and that other drawings can be obtained according to the provided drawings without inventive effort for a person skilled in the art.

FIG. 1 is a flowchart of a method for identifying a motion state according to an embodiment of the present invention;

FIG. 2 is a diagram of statistical results of stability coefficients according to an embodiment of the present invention;

FIG. 3 is a schematic diagram of sliding widget partitioning according to an embodiment of the present invention;

fig. 4 is a block diagram of a motion state recognition device according to an embodiment of the present invention.

Detailed Description

For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.

At present, in an application scene requiring motion recognition, motion recognition is mostly performed through features of accelerometer data, such as fourier transform (FFT), wavelet transform, and features of mean value, standard deviation, etc. However, when the theoretical research is applied to practical application, the quality of the motion state recognition result is completely dependent on the quality of the acquired data, and particularly in a fuzzy boundary area of the motion state, the motion state recognition result is inaccurate. In order to solve the above problems, the present embodiment increases the correlation coefficient in the process of identifying the motion state, thereby improving the accuracy of identifying the motion state. Referring to fig. 1 specifically, fig. 1 is a flowchart of a method for identifying a motion state according to an embodiment of the present invention; the method may include:

s110, acquiring data acquired by a motion sensor, and calculating a correlation coefficient by using the expectations of the data and the preset delay time when the motion state is determined to be a walking or riding fuzzy state according to the data.

Specifically, the present embodiment does not limit how to determine, according to the data collected by the motion sensor, that the current state is a walking or riding blur state, and correspondingly does not limit the kind of the data collected by the motion sensor. For example, the data may be acceleration or total acceleration, and the average value, average value or standard deviation of the data may be used correspondingly to compare with a corresponding set threshold interval, so as to determine that the motion state corresponding to the acquired data may be a walking state or a riding state, that is, a walking or riding fuzzy state. It will be appreciated that the user may be able to determine from this data that his corresponding movement state may be a walking state or a riding state. The motion state corresponding to the data can be rapidly and accurately determined to be a walking state or a riding state through the subsequent steps, and the accurate identification of the motion state is finally realized.

Further, to ensure accuracy and computational efficiency of the identification of the walk or ride blur condition. The embodiment determines the classification area corresponding to the walking or riding fuzzy state through the combined acceleration and the standard deviation. In this embodiment, the acquiring the data acquired by the motion sensor, and when determining that the motion state is a walking or riding blur state according to the data may include:

Specifically, the number of the three-axis accelerations is not limited in this embodiment, and the user may determine the three-axis accelerations according to the actual calculation accuracy requirement, the set calculation frequency of the motion state recognition, the frequency of the data collected by the motion sensor, and the like. For example, when the user invokes the algorithm corresponding to the motion state recognition method provided in this embodiment once in 8 seconds, the data collected by the corresponding motion sensor is the data collected in the 8 seconds. The specific amount is also related to the frequency with which the motion sensor collects data. For example, when the frequency of the data collected by the motion sensor is 26Hz, that is, 26 data are collected every second, 208 data are obtained for 8 seconds. After the corresponding data is acquired, the corresponding combined acceleration and the standard deviation of the combined acceleration are calculated. The calculation process is not limited in this embodiment, and the user may refer to a specific manner of calculating the combined acceleration and the standard deviation of the combined acceleration in the related art.

In this embodiment, a specific data interval of the first standard deviation interval is not limited, and a user may determine according to an actual application scenario. When the standard deviation is in the first standard deviation interval, the motion state at the moment is determined to be a walking or riding fuzzy state. Of course, the user may refer to a specific manner of setting the standard deviation interval corresponding to the walk or ride blur state in the related art.

In this embodiment, when it is determined that the standard deviation is located in the first standard deviation interval, that is, when it is determined that the motion state is a walking or riding blur state, it is necessary to calculate a correlation coefficient capable of accurately distinguishing the walking or riding blur state. The specific manner of calculating the correlation coefficient is not limited in this embodiment. The user can select the corresponding calculation method according to the actual calculation precision and the hardware calculation capability. The correlation coefficient in this embodiment is calculated, for example, using the pearson correlation (Pearson correlation) algorithm. Where autocorrelation is defined as the pearson correlation between values at different times in a random process. In particular, a correlation function can be utilized

To calculate a correlation coefficient, wherein the correlation function R (τ) can be expressed as a function of the delay time τ, X _t For the resultant acceleration, X, obtained within a preset time period t _t+τ For the resultant acceleration obtained by delaying τ for a preset period of time t, μ is X _t Sigma is X _t E is the standard deviation of the desired operation.

Of course, the user can modify the existing calculation method of the correlation coefficient to obtain a calculation method of the self-defined correlation parameter which is more suitable for the field of motion state identification. The correlation coefficient characterizes the correlation degree of the same event between two different periods, so that the correlation of different moments can be reflected. Therefore, in this embodiment, two different periods of the same event are obtained through a preset delay time, and the correlation coefficient is calculated in combination with the expectation of the data acquired by the motion sensor. Further, in this embodiment, the complexity of the calculation of the correlation coefficient is reduced by using the self-defined correlation coefficient, so as to improve the calculation efficiency. Preferably, calculating the correlation coefficient using the expected data and the preset delay time in the present embodiment may include:

using the formula a (τ) =e [ (X _t -μ)(X _t+τ -μ)]Respectively calculating a correlation parameter A (0) with a preset delay time of 0 and a correlation parameter A (n) with a preset delay time of n; the correlation coefficient A is the ratio A (0)/A (n) of A (0) to A (n).

Note that a (0) is obtained by substituting τ=0 into a (τ) to obtain a (0) =e [ (X) _t -μ)(X _t -μ)]The same thing a (n) is that τ=n is taken into a (τ) to obtain a (n) =e [ (X) _t -μ)(X _t+n -μ)]. Wherein τ is the delay time, X _t For the resultant acceleration, X, obtained within a preset time period t _t+τ For the resultant acceleration obtained by delaying τ for a preset period of time t, μ is X _t E is the expectation operation, X _t+n The total acceleration obtained by delaying n in a preset time period t is obtained. It can be understood that, in this embodiment, the specific time value of the preset delay time n is not limited, and may be set according to the actual use scenario. For example, n may be 0.5 seconds or 1 second, etc. The value of the preset time period t is not limited in this embodiment either, for example, the time for calling the algorithm corresponding to the motion state recognition method provided in this embodiment once may be, for example, the preset time period t is 8 seconds. Of course, in this embodiment, considering the accuracy of motion state identification and the calculation efficiency, the value of n may preferably be 1 in seconds. The correlation coefficient a corresponding at this time is specifically: a (0)/a (1), wherein a (1) is that τ=1 is taken into a (τ) to obtain a (1) =e [ (X) _t -μ)(X _t+1 -μ)]，X _t+1 For a resultant acceleration delayed by 1 second for a preset period t.

And S120, when the correlation coefficient is positioned in the first correlation coefficient interval, the motion state is a walking state.

And S130, when the correlation coefficient is located in the second correlation coefficient section, the motion state is a riding state.

Specifically, in this embodiment, specific data intervals of the first correlation coefficient interval and the second correlation coefficient interval are not limited, and the user may determine according to an actual application scenario. Since the correlation coefficient corresponding to the walking state is smaller than the correlation coefficient corresponding to the riding state, the upper limit value of the first correlation coefficient section should be the lower limit value of the second correlation coefficient section. For example, when the first correlation coefficient interval is a range of the first correlation coefficient or less, the second correlation coefficient interval is a range of the second correlation coefficient which is larger than the first correlation coefficient. If the first correlation coefficient is A1, the first correlation coefficient range is a range less than or equal to A1, and the second correlation coefficient range is a range greater than A1. Of course, the first correlation coefficient range may be a range smaller than A1, and the second correlation coefficient range may be a range equal to or larger than A1. That is, in this embodiment, it is not limited to which section the value corresponding to the section end point specifically belongs to, and the user may determine according to the actual situation.

In order to overcome the disadvantage of the related art, that is, to improve the accuracy of motion state identification, the present embodiment uses the calculated correlation coefficient to accurately distinguish the walking or riding blur state. The specific process may be that the current motion state is firstly determined to be in a walking or riding fuzzy state (for example, whether the current motion state is in a walking or riding fuzzy state can be quickly determined by calculating a simple standard deviation), then the correlation coefficient is calculated, and the corresponding motion state (namely, the current walking state or riding state) is accurately determined according to the correlation coefficient interval in which the correlation coefficient is located.

Based on the technical scheme, the method for identifying the motion state provided by the embodiment of the invention utilizes the correlation coefficient to distinguish the walking state from the riding state, so that the problem of confusion in identifying the motion state is avoided, the walking state and the riding state can be distinguished accurately, and the accuracy of identifying the motion state is further improved.

Based on the above embodiment, the correlation coefficient may be directly calculated after the motion state is determined to be the walking or riding blur state, and the walking or riding blur state may be accurately distinguished by using the correlation coefficient, or, of course, the motion recognition may be performed by using the features of the data collected by other motion sensors after the motion state is determined to be the walking or riding blur state, and the distinguishing may be performed by using the feature of the correlation coefficient after distinguishing a part of accurate walking states and riding states from the walking or riding blur state and another part of accurate distinguishing may not be performed on the walking or riding blur state by other features. Therefore, the calculated amount of the correlation coefficient can be reduced on the basis of ensuring the accuracy of the motion state identification. Specifically, the method in this embodiment may further include, before calculating the correlation coefficient using the expected data and the preset delay time:

Calculating a first sample entropy using the data;

and when the first sample entropy is in the third sample entropy interval, executing the step of calculating the correlation coefficient by using the expected data and the preset delay time.

Specifically, the sample entropy is a method for detecting the complexity of a time sequence, and the larger the sample entropy is, the more complex the sequence is and the worse the periodicity is. The acceleration signal of human body movement has a certain periodicity, so the human body movement state can be identified by utilizing the sample entropy. The embodiment is not limited to a specific form of calculating the sample entropy by using the data, and the user may calculate the corresponding sample entropy according to the specific type of the data collected by the user.

In this embodiment, specific data intervals of the first sample entropy interval, the second sample entropy interval, and the third sample entropy interval are not limited, and a user may determine according to an actual application scenario. Since the sample entropy corresponding to the walking state is generally smaller than the sample entropy corresponding to the riding state, the upper limit value of the first sample entropy section should be the lower limit value of the third sample entropy section, and the upper limit value of the third sample entropy section should be the lower limit value of the second sample entropy section. For example, when the first sample entropy section is a range smaller than the first set sample entropy, the third sample entropy section is a range greater than or equal to the first set sample entropy and less than or equal to the second set sample entropy, and the second sample entropy section is a range greater than the second set sample entropy, that is, the first set sample entropy is smaller than the second set sample entropy. If the first set sample entropy is S1, the second set sample entropy is S2, the first sample entropy interval is a range smaller than S1, the third sample entropy interval is a range greater than or equal to S1 and less than or equal to S2, and the second sample entropy interval is a range greater than S2. Of course, the first sample entropy section may be a range of S1 or less, the third sample entropy section may be a range of S1 or more and S2 or less, and the second sample entropy section may be a range of S2 or more. That is, in this embodiment, it is not limited to which section the value corresponding to each section end point specifically belongs to, and the user can determine according to the actual situation.

In this embodiment, when the first sample entropy is located in the first sample entropy interval, the motion state can be accurately determined to be a walking state; when the first sample entropy is located in the second sample entropy interval, the motion state can be accurately determined to be the riding state; only when the first sample entropy is located in the third sample entropy section, the current motion state is not accurately distinguished to be the walking state or the riding state by the sample entropy, and then a correlation coefficient needs to be calculated, and the state in the fuzzy area is accurately distinguished by using the correlation coefficient (the specific distinguishing process can refer to the embodiment).

Further, before calculating the first sample entropy using the data in the above embodiment, it may further include:

calculating a first kurtosis value using the data;

when the first kurtosis value is located in the third kurtosis value interval, the step of calculating the first sample entropy by using the data is executed.

Specifically, in this embodiment, when the motion state is determined to be in a walking or riding blur state according to the collected data, the walking or riding blur state is first distinguished by using the feature of kurtosis value, and in the portion where the kurtosis value cannot accurately distinguish the walking or riding blur state, the feature of sample entropy is used to distinguish, and the specific process can refer to the above embodiment. The present embodiment is not limited to a specific form of calculating the first kurtosis value using data, and the user may refer to the related art.

In this embodiment, the reason for distinguishing the walking or riding blur state by using the kurtosis value is that the kurtosis value is simple to calculate and the calculation speed is high. In the embodiment, several characteristics are combined, so that the calculation efficiency is ensured on the basis of ensuring the accuracy of the motion state identification. Firstly, by setting a plurality of kurtosis value intervals, the area which is easy to be confused and the area which can accurately identify the walking state and the riding state are distinguished. And secondly, executing the identification of the walking or riding fuzzy state in the divided area which is easy to be confused by adopting sample entropy with high calculation accuracy, and distinguishing the area which is easy to be confused from the area which can accurately identify the walking state and the riding state by setting a plurality of sample entropy sections. And then, executing the identification of the walking or riding fuzzy state by adopting a correlation coefficient with higher calculation accuracy in the divided area which is easy to be confused, and finally, realizing the accurate identification of the walking or riding fuzzy state through a plurality of correlation coefficient sections. Therefore, the accurate identification of the walking or riding fuzzy state is completed through the combination of the characteristics, and the calculation efficiency is ensured on the basis of ensuring the accuracy of the identification of the motion state.

In this embodiment, specific data intervals of the first kurtosis interval, the second kurtosis interval, and the third kurtosis interval are not limited, and a user can determine according to an actual application scenario. Since the kurtosis value corresponding to the walking state is generally smaller than the kurtosis value corresponding to the riding state, the upper limit of the first kurtosis value interval should be the lower limit of the third kurtosis value interval, and the upper limit of the third kurtosis value interval should be the lower limit of the second kurtosis value interval. For example, when the first kurtosis interval is a range smaller than the first set kurtosis value, the third kurtosis interval is a range greater than or equal to the first set kurtosis value and less than or equal to the second set kurtosis value, and the second kurtosis interval is a range greater than the second set kurtosis value, that is, the first set kurtosis value is smaller than the second set kurtosis value. If the first set kurtosis value is K1, the second set kurtosis value is K2, the first kurtosis value range is a range smaller than K1, the third kurtosis value range is a range greater than or equal to K1 and less than or equal to K2, and the second kurtosis value range is a range greater than K2. Of course, the first kurtosis value range may be equal to or less than K1, the third kurtosis value range may be equal to or greater than K1 and less than K2, and the second kurtosis value range may be equal to or greater than K2. That is, in this embodiment, it is not limited to which section the value corresponding to each section end point specifically belongs to, and the user can determine according to the actual situation.

In this embodiment, when the first kurtosis value is located in the first kurtosis value interval, the motion state can be accurately determined to be a walking state; when the first kurtosis value is positioned in the second kurtosis value interval, the motion state can be accurately determined to be a riding state; only when the first kurtosis value is located in the third kurtosis value interval, the current motion state is not accurately distinguished to be the walking state or the riding state through the kurtosis value, at the moment, sample entropy needs to be calculated, and the state in the fuzzy area is accurately distinguished by the sample entropy; when the first sample entropy is located in the first sample entropy interval, the motion state can be accurately determined to be a walking state; when the first sample entropy is located in the second sample entropy interval, the motion state can be accurately determined to be the riding state; only when the first sample entropy is located in the third sample entropy section, the current motion state is not accurately distinguished to be the walking state or the riding state through the sample entropy, and then a correlation coefficient needs to be calculated, so that the state in the fuzzy area is accurately distinguished by the correlation parameter.

Of course, it can be understood that the user may not select the feature of sample entropy, that is, when the first kurtosis value is located in the first kurtosis value range, the motion state may be accurately determined to be the walking state; when the first kurtosis value is positioned in the second kurtosis value interval, the motion state can be accurately determined to be a riding state; only when the first kurtosis value is located in the third kurtosis value interval, the current motion state is not accurately distinguished to be the walking state or the riding state through the kurtosis value, and the characteristic of the correlation coefficient can be directly adopted to carry out subsequent identification. Namely, the correlation coefficient is directly calculated in the area where the kurtosis value cannot be distinguished, and the state of the fuzzy area where the kurtosis value cannot be distinguished is accurately distinguished by using the correlation coefficient.

Based on the technical scheme, the motion state identification method provided by the embodiment of the invention utilizes the kurtosis value, the sample entropy and the correlation coefficient to distinguish the walking state from the riding state, so that the problem of confusion in motion state identification is avoided, the walking state and the riding state can be distinguished accurately, and the motion state identification accuracy is further improved. And through the combination of a plurality of characteristics, the advantages of each characteristic are exerted, the disadvantages of each characteristic (namely the advantages of small calculation amount of kurtosis value and high calculation speed are exerted, the disadvantages of inaccurate identification results in certain areas are solved by the larger calculation amount, but the sample entropy and the correlation coefficient with high calculation accuracy are solved, and the advantages of high calculation accuracy of the sample entropy and the correlation coefficient are exerted), so that the calculation amount is reduced, and the calculation efficiency is improved.

Based on the above embodiment, in order to further increase the operation speed of motion recognition and greatly reduce the operation amount to save the code power consumption, the present embodiment may further include, before calculating the first kurtosis value using the data:

calculating a stability parameter by using the data;

when the stability parameter is located in the first stability parameter interval, executing the step of calculating a first kurtosis value by using the data;

When the stability parameter is located in the second stability parameter interval, calculating a second sample entropy by using the data; when the second sample entropy is larger than the sample entropy threshold, the motion state is a riding state; when the second sample entropy is smaller than the sample entropy threshold, the motion state is a walking state.

In the embodiment, when determining the walking or riding fuzzy state, in order to improve the efficiency of subsequent calculation, the subsequent recognition of the walking or riding fuzzy state is divided into two parts through the stability parameter, and the two parts select different recognition strategies to accurately distinguish the walking or riding fuzzy state, so that the recognition operation speed is greatly improved, the operation amount is greatly reduced, and the code power consumption is saved on the basis of meeting the recognition accuracy of the walking or riding fuzzy state.

Reference may be made specifically to fig. 2, wherein x is walking, i.e. walking state, + is riding, i.e. riding state. In fig. 2, a large amount of data is tested, the vertical axis is AS, the stability parameter is the AS, the horizontal axis represents the number of tests, and it is determined that the stability parameter is located in the second stability parameter section, that is, the number of data corresponding to the walking state of the area above the black line in the figure is only 7. That is, the stability parameters in the second stability parameter interval are basically data corresponding to the riding state. The stability parameters obtained by calculating the data corresponding to the very few walking states are larger, and the stability parameters obtained by calculating the data corresponding to the basic walking states are smaller. Therefore, in the present embodiment, the data above the black line is only required to be discriminated by the sample entropy. For the data below the black line, the peak value is firstly utilized to distinguish (although the peak value cannot be completely and accurately distinguished, the calculated amount can be greatly reduced), then the sample entropy is utilized to distinguish, and finally the correlation coefficient is utilized to distinguish (the recognition of the walking or riding fuzzy state can be accurately completed through the sample entropy and the correlation coefficient). Of course, the method can also be implemented by directly distinguishing the sample entropy and then distinguishing the correlation coefficient, or by distinguishing the correlation coefficient only, and the methods can realize accurate identification of the fuzzy state of walking or riding, and the difference is only that the calculated amount is different and the calculated parameter quantity is different. Various implementations may refer to the specific processes in the foregoing embodiments.

It can be understood that the acceleration signal in the walking state has certain periodicity and relatively regularity, so that the calculated amount for accurately distinguishing the walking or riding fuzzy state can be greatly reduced by using the stability value, and the operation speed is improved. In addition, the correlation parameters can accurately identify the motion state. Therefore, in the embodiment, the calculation speed of the motion state identification is ensured through the stability coefficient, and the accuracy of the motion state identification is ensured through the kurtosis value, the sample entropy and the correlation coefficient, so that the motion state can be identified more quickly and accurately.

The specific manner of calculating the stability parameter is not limited in this embodiment. The user can select the corresponding calculation method according to the actual calculation precision and the hardware calculation capability. In order to improve the reliability of the calculation of the stability parameters. The process of calculating the stability parameter using the data in this embodiment may include:

acquiring the total acceleration of the triaxial acceleration corresponding to the preset window, and determining the minimum total acceleration Min1 of the preset window;

using the formula

Calculating a total acceleration Average value Average1;

Using the formula

taking the ratio Max/Min of Max and Min AS a stability parameter AS;

Specifically, the size of the preset window is not limited in this embodiment, and the preset window may be set according to a period corresponding to the acquired data acquired by the motion sensor, and may also be understood as a period of invoking an algorithm corresponding to the motion state identification method provided in this embodiment. For example, when the data involved in the movement state recognition calculation is data acquired within 8 seconds, the corresponding preset window is a window corresponding to 8 seconds. At this time, the minimum total acceleration Min1 of the corresponding preset window can be understood as the minimum total acceleration Min1 of all total accelerations in the 8 second window, wherein if the frequency of the data collected by the motion sensor is 26Hz, that is, 26 data are collected per second, and 208 data are obtained in total for 8 seconds. Specifically, the process of obtaining Min1 may be: min1=min (X (1), X (2), X (3), X (208)), where Min () is the operator that takes the minimum value.

After the determination of the minimum combined acceleration Min1, the Average value Average1 of the combined accelerations needs to be calculated. The Average1 is an Average value of corresponding data in a preset window at this time calculated after subtracting the minimum total acceleration from each total acceleration in the preset window, and is referred to as a total acceleration Average value Average1. The above procedure is expressed by the formula:

wherein X (i) is the ith combined acceleration in a preset window, and n is the number of the combined accelerations corresponding to the preset window. Specifically, when the number of the preset window integrated accelerations is 208, the corresponding formula is that

The number of sliding portlets included in the preset window and the generation mode of the sliding portlets are not limited in this embodiment. For example, when the preset window is an 8-second window, the window corresponding to the preset second can be sequentially slid to be used as a sliding small window. The value of the preset second is not limited in this embodiment, and may be 2 seconds, for example. When each sliding widget corresponds to a 2-second window, the preset window of the 8-second window may correspond to 7 sliding widgets, namely, the 1 st sliding widget and the 2 nd sliding widget are respectively the 1 st sliding widget, the 2 nd sliding widget and the 3 rd sliding widget are respectively the 2 nd sliding widget, the 3 rd sliding widget and the 4 th sliding widget … … are respectively the 3 rd sliding widget, the 7 th sliding widget and the 8 th sliding widget are respectively the 7 th sliding widget, and specific reference may be made to fig. 3. At this time, the minimum combined acceleration Min2q of each sliding small window needs to be acquired. I.e., the minimum combined acceleration Min2q of all combined accelerations in each sliding widget. That is, if 7 sliding small windows exist at this time, the corresponding q ranges from 1 to 7, and the minimum combined acceleration of the 7 sliding small windows can be obtained. If the frequency of the data collected by the motion sensor is 26Hz, that is, 26 data are collected every second, 52 data are collected for 2 seconds corresponding to each sliding small window, that is, the minimum total acceleration needs to be determined from 52 total accelerations for each sliding small window.

The specific process of calculating the combined acceleration Average value Average2q of each sliding window in this embodiment may refer to the calculation process of the combined acceleration Average value Average 1. That is, after each combined acceleration in the q-th sliding small window is subtracted from the minimum combined acceleration corresponding to the sliding small window, the Average value of the corresponding data in the q-th sliding small window at this time is calculated, and is called as the combined acceleration Average value Average2q of the q-th sliding small window. Wherein, the value range of q is a value corresponding to the total number of the sliding small windows from 1. The above procedure is expressed by the formula:

wherein Xq (m) is the mth combined acceleration of the qth sliding small window in the preset window, l is the number of combined accelerations of the sliding small windows, and Min2q is the minimum combined acceleration of the qth sliding small window in the preset window. Specifically, when the number of the preset window integrated accelerations is 52, the corresponding formula is +.>

And when the number of the sliding small windows is 7, the value range of q is 1,2,3,4,5,6 and 7.

The maximum value Max and the minimum value Min are determined from the total acceleration Average value Average1 and the total acceleration Average value Average2q of each sliding window. That is, the maximum total acceleration value Max and the minimum total acceleration value Min are determined from the obtained total acceleration average value and the total acceleration average value corresponding to each sliding window. When the number of sliding portlets is 7, then the process finds the maximum Max and minimum Min from the 8 data. Finally, the ratio Max/Min of Max and Min is used AS a stability parameter AS.

Specifically, in this embodiment, specific data intervals of the first stability parameter interval and the second stability parameter interval are not limited, and a user may determine according to an actual application scenario. Since the first stability parameter interval corresponds to an area below the black line in fig. 2, the upper limit value of the first stability parameter interval should be the lower limit value of the second stability parameter interval. For example, when the first stability parameter interval is a range smaller than or equal to the first stability parameter, the second stability parameter interval is a range larger than the first stability parameter. If the first stability parameter is AS1, the first stability parameter interval is equal to or less than AS1, and the second stability parameter interval is greater than AS1. Of course, the first stability parameter interval may be smaller than AS1, and the second stability parameter interval may be equal to or greater than AS1. That is, in this embodiment, it is not limited to which section the value corresponding to the section end point specifically belongs to, and the user may determine according to the actual situation.

Based on the technical scheme, the motion state identification method provided by the embodiment of the invention utilizes the stability coefficient and the correlation coefficient to distinguish the motion states, can effectively and accurately distinguish the situation that the motion states are easy to be confused, reduces the calculated amount, improves the calculation efficiency of motion state identification, and improves the hardware utilization rate.

Based on any of the above embodiments, when determining that the exercise state is a running or riding blur state according to the data in the present embodiment may further include:

calculating a second kurtosis value using the data;

Specifically, the present embodiment does not limit how to determine that the current state is a running or riding fuzzy state according to the data collected by the motion sensor, and correspondingly does not limit the kind of the data collected by the motion sensor. For example, the data may be acceleration or total acceleration, and the average value, average value or standard deviation of the data may be used correspondingly to compare with a corresponding set threshold interval, so as to determine that the motion state corresponding to the acquired data may be running state or riding state, that is, running or riding fuzzy state. It will be appreciated that the user may be able to determine from this data that his corresponding movement state may be a running state or a riding state. The motion state corresponding to the data can be quickly and accurately determined to be the running state or the riding state through the subsequent steps, and the accurate identification of the motion state is finally realized.

Further, to ensure accuracy and computational efficiency of running or riding blur state identification. The embodiment determines the classification area corresponding to the running or riding fuzzy state through the combined acceleration and the standard deviation. In this embodiment, the step of acquiring the data acquired by the motion sensor and determining that the motion state is a running or riding blur state according to the data may include:

when the standard deviation is in the second standard deviation range, the exercise state is running or riding blur state.

Specifically, according to the description of the above embodiment, the number of the acquired triaxial accelerations is not limited in this embodiment, and the user may determine the triaxial acceleration according to the actual calculation accuracy requirement, the set calculation frequency of the motion state recognition, the frequency of the motion sensor acquisition data, and the like. After the corresponding data is acquired, the corresponding combined acceleration and the standard deviation of the combined acceleration are calculated. The calculation process is not limited in this embodiment, and the user may refer to a specific manner of calculating the combined acceleration and the standard deviation of the combined acceleration in the related art.

In this embodiment, the specific data interval of the second standard deviation interval is not limited, and the user may determine according to the actual application scenario. When the standard deviation is in the second standard deviation interval, the motion state at the moment is determined to be running or riding fuzzy. Of course, the user may refer to a specific manner of setting the standard deviation interval corresponding to the running or riding blur state in the related art. Typically, the standard deviation corresponding to a running or riding blur state is greater than a walking or riding blur state. The lower limit value of the second standard deviation interval is therefore greater than or equal to the upper limit value of the first standard deviation interval.

In this embodiment, the running state and the riding state can be distinguished by the single kurtosis threshold, i.e. the first kurtosis threshold, and compared with the mode of distinguishing the running state and the riding state by using sample entropy, the kurtosis value is simpler to calculate, and the running state and the riding state are distinguished by taking the characteristic of kurtosis into consideration of calculation speed and memory saving. Kurtosis and sample entropy are basic statistical features of data, and the principle thereof is not repeated here. Of course, the specific value of the first kurtosis threshold is not limited in the present embodiment, and the corresponding exercise state is specifically a running state or a riding state when the second kurtosis value is equal to the first kurtosis threshold. For being set according to the actual situation.

Based on the technical scheme, the kurtosis value is used for replacing the sample entropy in the embodiment of the invention, so that the data calculation amount is reduced on the basis of not affecting the running state and riding state identification accuracy, and the running state and riding state identification efficiency and the storage space utilization rate are improved.

The motion state recognition device, the terminal and the computer readable storage medium provided in the embodiments of the present invention are described below, and the motion state recognition device, the terminal and the computer readable storage medium described below and the motion state recognition method described above may be referred to correspondingly.

Referring to fig. 4, fig. 4 is a system block diagram of a motion state recognition device according to an embodiment of the present invention, where the device may include:

the correlation coefficient calculation module 100 is configured to obtain data collected by the motion sensor, and calculate a correlation coefficient according to the data expectation and a preset delay time when the motion state is determined to be a walking or riding fuzzy state according to the data;

the first classification module 200 is configured to, when the correlation coefficient is located in the first correlation coefficient interval, determine that the motion state is a walking state; when the correlation coefficient is located in the second correlation coefficient interval, the motion state is a riding state.

Based on the above embodiment, the apparatus may further include:

the sample entropy calculation module is used for calculating a first sample entropy by using the data;

the second classification module is used for enabling the motion state to be a walking state when the first sample entropy is located in the first sample entropy interval; when the first sample entropy is located in the second sample entropy interval, the motion state is a riding state; when the first sample entropy lies in the third sample entropy interval, the correlation coefficient calculation module 100 is triggered.

Based on the above embodiment, the apparatus may further include:

a first kurtosis value calculation module for calculating a first kurtosis value using the data;

The third classification module is used for enabling the motion state to be a walking state when the first kurtosis value is located in the first kurtosis value interval; when the first kurtosis value is located in the second kurtosis value interval, the motion state is a riding state; and when the first kurtosis value is positioned in the third kurtosis value interval, triggering a sample entropy calculation module.

Based on the above embodiment, the apparatus may further include:

the stability parameter calculation module is used for calculating stability parameters by utilizing the data;

the fourth classification module is used for triggering the first kurtosis value calculation module when the stability parameter is located in the first stability parameter interval; when the stability parameter is located in the second stability parameter interval, calculating a second sample entropy by using the data; when the second sample entropy is larger than the sample entropy threshold, the motion state is a riding state; when the second sample entropy is smaller than the sample entropy threshold, the motion state is a walking state.

Based on the above embodiment, the stability parameter calculation module may include:

the first calculation unit is used for acquiring the total acceleration of the triaxial acceleration corresponding to the preset window and determining the minimum total acceleration Min1 of the preset window;

a second calculation unit for using the formula

Calculating a total acceleration Average value Average1;

The third calculation unit is used for determining the minimum total acceleration Min2q of each sliding small window in the preset window;

a fourth calculation unit for using the formula

a fifth calculation unit, configured to determine a maximum value Max and a minimum value Min from the total acceleration Average value Average1 and the total acceleration Average value Average2q of each sliding small window, and take a ratio Max/Min of Max to Min AS a stability parameter AS;

Based on any of the above embodiments, the apparatus may further include:

a second kurtosis value calculation module for calculating a second kurtosis value using the data;

a fifth classification module, configured to, when the second kurtosis value is smaller than the first kurtosis threshold, determine that the exercise state is a running state; when the second kurtosis value is greater than the first kurtosis threshold, the motion state is a riding state.

Based on any of the above embodiments, the correlation coefficient calculation module 100 may include:

The walking or riding fuzzy state determining unit is used for calculating the combined acceleration of the acquired triaxial acceleration and calculating the standard deviation of the combined acceleration; when the standard deviation is in the first standard deviation interval, the motion state is a walking or riding fuzzy state;

a correlation coefficient calculation unit for calculating a correlation coefficient using the formula a (τ) =e [ (X) _t -μ)(X _t+τ -μ)]Respectively calculating a correlation parameter A (0) with a preset delay time of 0 and a correlation parameter A (n) with a preset delay time of n; taking the ratio A (0)/A (n) of A (0) to A (n) as a correlation coefficient A;

It should be noted that, based on any embodiment, the apparatus may be implemented based on a programmable logic device, where the programmable logic device includes an FPGA, a CPLD, a single-chip microcomputer, a processor, and the like.

The embodiment also provides a terminal, including: the motion sensor is used for collecting data; a memory for storing a computer program; a processor for implementing the steps of the method for identifying a motion state as described in any of the embodiments above when executing a computer program. When the processor executes the computer program, the data acquired by the motion sensor are acquired, and when the motion state is determined to be a walking or riding fuzzy state according to the data, the correlation coefficient is calculated by utilizing the expectations of the data and the preset delay time; when the correlation coefficient is located in the first correlation coefficient interval, the motion state is a walking state; when the correlation coefficient is located in the second correlation coefficient interval, the motion state is a riding state.

The motion sensor is not limited in this embodiment, and may be specifically determined according to the selection of the user calculation parameters. Optionally, the motion sensor is specifically a triaxial acceleration sensor.

The terminal is not limited in this embodiment, and the terminal may be an intelligent wearable device such as an intelligent watch, an intelligent bracelet, a tracker, or a mobile terminal such as a mobile phone. When the terminal is used for positioning, for example, the tracker is used for positioning old people, children or valuables, the motion state classification can be used as the positioning basis of the terminal, and the accurate motion state identification result provides basis for the positioning strategy of the tracker. For example, when the device is in a stationary state continuously, frequent positioning is not needed, the device is positioned once in 1 hour, is positioned once in 4 minutes in a walking state, is positioned once in 2 minutes in a running state, and other states have corresponding positioning strategies, which are not described herein. Therefore, the power consumption of the terminal can be greatly reduced through accurate motion state identification, the standby time and the use time of the terminal are improved, and the user experience is improved. Of course, the motion state can be classified into other positioning devices connected with the terminal to provide positioning basis.

The present embodiment also provides a computer-readable storage medium, on which a computer program is stored, which when executed by a processor implements the steps of the method for identifying a motion state according to any of the embodiments described above. When the computer program is executed by the processor, the data acquired by the motion sensor are acquired, and when the motion state is determined to be a walking or riding fuzzy state according to the data, the correlation coefficient is calculated by utilizing the expectations of the data and the preset delay time; when the correlation coefficient is located in the first correlation coefficient interval, the motion state is a walking state; when the correlation coefficient is located in the second correlation coefficient interval, the motion state is a riding state.

The computer readable storage medium may include: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (Random Access Memory, RAM), a magnetic disk, or an optical disk, or other various media capable of storing program codes.

In the description, each embodiment is described in a progressive manner, and each embodiment is mainly described by the differences from other embodiments, so that the same similar parts among the embodiments are mutually referred. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant points refer to the description of the method section.

Those of skill would further appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware, computer software, or combinations of both, and that the various illustrative elements and steps are described above generally in terms of functionality in order to clearly illustrate the interchangeability of hardware and software. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the solution. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.

The steps of a method or algorithm described in connection with the embodiments disclosed herein may be embodied directly in hardware, in a software module executed by a processor, or in a combination of the two. The software modules may be disposed in Random Access Memory (RAM), memory, read Only Memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

The method, the device, the terminal and the computer readable storage medium for identifying the motion state provided by the invention are described in detail. The principles and embodiments of the present invention have been described herein with reference to specific examples, the description of which is intended only to facilitate an understanding of the method of the present invention and its core ideas. It should be noted that it will be apparent to those skilled in the art that various modifications and adaptations of the invention can be made without departing from the principles of the invention and these modifications and adaptations are intended to be within the scope of the invention as defined in the following claims.

Claims

1. A method for identifying a motion state, comprising:

when the correlation coefficient is located in the second correlation coefficient interval, the motion state is a riding state;

before the correlation coefficient is calculated by using the expected and preset delay time of the data, the method further comprises the following steps:

calculating a first sample entropy using the data;

when the first sample entropy is located in a third sample entropy interval, executing the step of calculating a correlation coefficient by using the expected data and the preset delay time;

before said calculating the first sample entropy using said data, further comprising:

calculating a first kurtosis value using the data;

executing the step of calculating a first sample entropy using the data when the first kurtosis value is located in a third kurtosis value interval;

wherein before said calculating the first kurtosis value using said data, further comprises:

calculating a stationarity parameter using the data;

2. The method of claim 1, wherein calculating a stationarity parameter using the data comprises:

Using the formula

Calculating a total acceleration Average value Average1;

using the formula

taking the ratio Max/Min of Max and Min AS a stability parameter AS;

3. The method of claim 1, further comprising, when determining from the data that the motion state is a running or riding blur state:

calculating a second kurtosis value using the data;

4. The method of claim 1, wherein determining that the motion state is a walking or riding blur state based on the data comprises:

5. The method for recognizing a motion state according to any one of claims 1 to 4, wherein the calculating a correlation coefficient using the expectation of the data and a preset delay time includes:

taking the ratio A (0)/A (n) of A (0) to A (n) as a correlation coefficient A;

6. A motion state recognition apparatus, comprising:

The first classification module is used for enabling the motion state to be a walking state when the correlation coefficient is located in a first correlation coefficient interval; when the correlation coefficient is located in the second correlation coefficient interval, the motion state is a riding state;

the motion state identification device further comprises:

a sample entropy calculation module for calculating a first sample entropy using the data;

the second classification module is used for enabling the motion state to be a walking state when the first sample entropy is located in the first sample entropy interval; when the first sample entropy is located in the second sample entropy interval, the motion state is a riding state; when the first sample entropy is located in a third sample entropy interval, triggering the correlation coefficient calculation module;

the motion state identification device further comprises:

the third classification module is used for enabling the motion state to be a walking state when the first kurtosis value is located in the first kurtosis value interval; when the first kurtosis value is located in the second kurtosis value interval, the motion state is a riding state; when the first kurtosis value is located in a third kurtosis value interval, triggering the sample entropy calculation module;

The motion state identification device further comprises:

the stability parameter calculation module is used for calculating stability parameters by using the data;

the fourth classification module is used for triggering the first kurtosis value calculation module when the stability parameter is located in a first stability parameter interval; when the stability parameter is located in a second stability parameter interval, calculating a second sample entropy by using the data; when the second sample entropy is larger than a sample entropy threshold value, the motion state is a riding state; and when the second sample entropy is smaller than a sample entropy threshold value, the motion state is a walking state.

7. A terminal, comprising:

the motion sensor is used for collecting data;

a memory for storing a computer program;

processor for implementing the steps of the method for identifying a state of motion according to any one of claims 1 to 5 when executing said computer program.

8. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, implements the steps of the method of identifying a state of motion according to any one of claims 1 to 5.