WO2019033586A1

WO2019033586A1 - Swimming exercise analysis method based on smartwatch and smartwatch

Info

Publication number: WO2019033586A1
Application number: PCT/CN2017/109857
Authority: WO
Inventors: 戎海峰; 牛浩田; 杨钒沁; 王芳德; 夏啸夫
Original assignee: 东莞市远峰科技有限公司; 广东远峰电子科技股份有限公司
Priority date: 2017-08-16
Filing date: 2017-11-08
Publication date: 2019-02-21
Also published as: CN107376247B; CN107376247A

Abstract

A swimming exercise analysis method based on a smartwatch, comprising: acquiring multiple peak values and trough values in motion signals captured on each coordinate axis of a sensor (41 and 520) of a smartwatch; filtering the peak values and the trough values to produce valid peak values and valid trough values; matching a periodic motion on the basis of the valid peak values and the valid trough values; and determining a swimming state of a wearer on the basis of the match result. Also disclosed is the smartwatch and a computer-readable storage medium.

Description

Swimming watch analysis method based on smart watch and smart watch

Technical field

The present disclosure relates to the field of smart wear technology, for example, to a smart watch based swimming motion analysis method and a smart watch.

Background technique

Today, as people pay more and more attention to health and sports, swimming has become a popular form of aerobic exercise. With the advancement of waterproof technology for smart watches, the monitoring of the use of smart watches for swimming is also gradually being realized. As the technical basis for solving motion monitoring and motion state reminders, motion state recognition is the core of smart watch algorithm and one of the difficulties.

In the field of smart watches, the commonly used swimming monitoring method is that the wearer actively activates the swimming mode, collects the wearer's motion signal through the sensor, and matches with the preset action templates of different swimming postures, thereby judging the swimming posture adopted by the wearer. The amount of calculation is huge. Estimating the calorie consumption of the wearer by timing and stroke characteristics, the data is rough, and most people do not have enough knowledge of calories, and the calories are not intuitive enough compared to the actual mileage.

Summary of the invention

The present disclosure proposes a swimming watch analysis method based on a smart watch and a smart watch, which can automatically recognize the swim state of the wearer through the action signal collected by the smart watch.

The present disclosure provides a swimming watch analysis method based on a smart watch, including:

Obtaining a plurality of peaks and a plurality of valley values from motion signals collected by the sensing elements on each coordinate axis of the multi-axis sensor of the smart watch;

Filtering the plurality of peaks and the plurality of valley values to derive effective peaks and effective valleys therefrom;

Performing a matching of periodic actions according to the effective peak value and the effective valley value;

The wearer's swimming state is determined based on the matching result.

Optionally, the step of filtering the plurality of peaks and the plurality of valley values to obtain an effective peak and an effective valley from the method, comprising:

And selecting, according to a time sequence generated by the motion signal, a peak value and a valley value adjacent to the peak value, and calculating a signal amplitude difference and a time difference between the peak value and the valley value;

When the signal amplitude difference is greater than a preset amplitude threshold, and the time difference is greater than a preset time threshold, Then the peak is an effective peak;

When the peak is an invalid peak, discarding the valley value, and selecting one of two peaks adjacent to the valley value according to a preset filtering rule to discard;

Repeat the above steps until all valid peaks are obtained.

Optionally, the step of filtering the plurality of peaks and the plurality of valleys to obtain an effective peak and an effective valley from the method further includes:

And selecting a valley value and a peak value adjacent to the valley value according to a time sequence generated by the motion signal, and calculating a signal amplitude difference and a time difference;

And when the signal amplitude difference between the peak value and the bottom value is greater than the preset amplitude threshold, and the time difference is greater than the preset time threshold, the valley value is a valid valley value;

When the valley value is an invalid valley value, discarding the peak value, and selecting one of two valley values adjacent to the peak value according to a preset filtering rule to discard;

Repeat the above steps until all effective troughs are obtained.

Optionally, the step of performing matching of the periodic actions according to the effective peak and the effective valley includes:

Setting, in a chronological order of the action signals, each of the effective peaks and the effective valleys, setting each of the effective peaks and the two effective valleys adjacent to the effective peak as an action group, Calculating the signal amplitude difference and time difference between the effective peak and the effective valley in an action group, respectively; or

Setting, in a chronological order of the action signals, each of the effective peaks and the effective valleys, setting each effective valley value and two effective peaks adjacent to the effective valley value as one action group , respectively calculating the signal amplitude difference and time difference between the effective peak and the effective valley in an action group;

Calculating a similarity of two action groups adjacent in time according to the signal amplitude difference and the time difference;

When the similarity meets a preset condition, each action group is recorded as a periodic action;

Record each cycle action as the number of strokes.

Optionally, the step of determining a swim state of the wearer according to the matching result includes:

When the total number of strokes reaches a preset number of times, it is determined that the wearer is in a swimming state.

Optionally, after the step of determining that the wearer is in a swimming state, the method further includes:

The decision tree algorithm is used to determine the wearer's stroke based on the action signal of a periodic action.

Optionally, after the step of determining that the user is in a swimming state, the method further includes:

According to the wearer's stroke, the action signal, the magnetometer threshold, the stroke number threshold, and the time of the circle The threshold is counted.

Optionally, after the step of performing matching of the periodic actions according to the effective peak value and the effective bottom value, the method further includes:

The number of strokes matched by each coordinate axis of the multi-axis sensor is separately counted;

Evaluating the motion signal quality of each coordinate axis according to a preset rule and the number of strokes;

Select the coordinate axis with the best motion signal quality to continue the matching of the periodic motion.

The present disclosure also provides a smart watch including a nine-axis sensor and a processor.

The nine-axis sensor is configured to collect an action signal of the wearer;

The processor includes a signal screening module, a filtering module, an action matching module, and a state detecting module;

The signal screening module is configured to acquire a plurality of peaks and a plurality of valleys in the motion signal;

The filtering module is configured to filter the plurality of peaks and the plurality of valley values to obtain an effective peak value and an effective valley value therefrom;

The action matching module is configured to perform matching of periodic actions according to the effective peak value and the effective bottom value;

The state detection module is configured to determine a swim state of the wearer based on the matching result.

Optionally, the filtering module is configured to:

When the signal amplitude difference is greater than a preset amplitude threshold, and the time difference is greater than a preset time threshold, the peak is an effective peak;

Repeat the above steps until all valid peaks are obtained;

When the signal amplitude difference between the peak value and the valley value is greater than a preset amplitude threshold, and the time difference is greater than a preset time threshold, the valley value is a valid valley value;

Repeat the above steps until all effective troughs are obtained.

Optionally, the action matching module is set to:

Each cycle action is recorded as the number of strokes.

Optionally, the status detection module is configured to:

Optionally, the state detecting module is further configured to: after determining that the wearer is in a swimming state,

Determining the wearer's stroke using a decision tree algorithm based on the action signal of a periodic action;

The circle is counted according to the wearer's stroke, the motion signal, the magnetometer threshold, the stroke number threshold, and the lap time threshold.

Optionally, the action matching module is further configured to: after performing matching of the periodic actions according to the effective peak value and the effective bottom value,

Separating the number of strokes matched by each coordinate axis of the nine-axis sensor;

The coordinate axis optimal for the motion signal is selected to continue the matching of the periodic motion.

The present disclosure also provides a computer readable storage medium storing computer executable instructions for performing any of the methods described above.

The present disclosure also provides a smart watch that includes one or more processors, a memory, and one or more programs, the one or more programs being stored in a memory when executed by one or more processors , perform the above method.

The present disclosure also provides a computer program product comprising a computer program stored on a non-transitory computer readable storage medium, the computer program comprising program instructions, when the program instructions are executed by a computer, Having the computer perform any of the methods described above.

The present disclosure provides a swimming watch analysis method based on a smart watch and a smart watch, which collects an action signal through a sensor of a smart watch, performs filtering to obtain an effective signal, and then according to characteristics of the effective signal The action cycle is matched to determine the swim status of the wearer. In the embodiment of the present application, the action signals collected by the sensor are filtered, the non-standard signal is removed, the basic data amount is reduced, and the wearer is automatically determined according to the periodicity of the signal itself, and the wearer does not need to manually open the swimming mode. Compared with related technologies, a large amount of template matching calculation is not required, the computing load of the smart watch is reduced, and the recognition rate is improved.

DRAWINGS

1 is a flow chart of a swimming motion analysis method provided in Embodiment 1;

2 is a flow chart of a swimming motion analysis method provided in Embodiment 2;

3 is a flow chart of a swimming motion analysis method provided in Embodiment 3;

4 is a schematic structural view of a smart watch provided in Embodiment 4.

FIG. 5 is a schematic diagram of a hardware structure of a smart watch according to an embodiment.

Detailed ways

Embodiment 1

The embodiment provides a swimming sport analysis method based on a smart watch, which is suitable for a scene in which a wearer of a smart watch performs motion monitoring, and can identify whether the wearer is in a swimming state.

1 is a flow chart of a swimming motion analysis method provided in Embodiment 1. As shown in FIG. 1, the swimming motion analysis method includes the following steps:

Step 11. Acquire a plurality of peaks and a plurality of valleys from the motion signals collected by the sensing elements on each coordinate axis of the multi-axis sensor of the smart watch.

The sensor of the smart watch can adopt a nine-axis sensor, including a three-axis accelerometer, a three-axis gyroscope and a three-axis magnetometer. During the process of collecting data by the sensor, each coordinate axis has a corresponding motion signal generated according to the acquired motion signal. The signal amplitude can be plotted as a wavy curve, and the peaks and valleys of the wavy curve are screened according to the signal amplitude. In this embodiment, the motion signals of the accelerometer and the gyroscope can be selected to perform the operation of step 12.

In step 12, the plurality of peaks and the plurality of valley values are filtered to obtain an effective peak value and an effective valley value.

According to the time sequence generated by the action signal, a peak value and a valley value adjacent to the peak value are selected, and a signal amplitude difference and a time difference between the peak value and the valley value are calculated; when the signal amplitude difference is greater than a preset amplitude threshold, and the time difference is greater than a preset time threshold, the peak is a valid peak; when the peak is an invalid peak, the valley is discarded, and one of the two peaks adjacent to the valley is discarded according to a preset filtering rule, and is repeated. The above steps until all valid peaks are obtained.

According to the time sequence generated by the action signal, a valley value and an adjacent peak value after the valley value are selected, and a signal amplitude difference and a time difference between the valley value and the peak value are calculated; when the signal amplitude difference is greater than a preset amplitude threshold, and the time difference is If the threshold is greater than the preset time threshold, the valley value is a valid valley value; when the valley value is an invalid valley value, the peak value is discarded, and one of two valley values adjacent to the peak value is selected according to a preset filtering rule. Discard, repeat the above steps until all effective troughs are obtained.

In the chronological order in which the motion signals are generated, the peak and valley values appear alternately, so the filtering of the peak and valley values is also alternated.

In order to facilitate calculation and judgment, the above signal amplitude difference and the above time difference take absolute values. A reasonable initial value is set for the preset amplitude threshold and the preset time threshold, and the average value of the signal amplitude difference and the average value of the time difference are calculated while filtering, and the amplitude threshold and the time threshold are dynamically adjusted according to the two average values, respectively.

For example, the peaks are represented by p0, p1, p2, ..., and the valleys are represented by v0, v1, v2, ..., and the peaks and valleys p0, v0, p1, v1 are sequentially obtained according to the time sequence generated by the action signals. P2, v2, ..., the corresponding time points of peak and valley occurrence are t0, t1, t2, t3, t4, t5, ...; calculation and judgment: when the signal amplitude difference |p0-v0|> preset amplitude threshold And when the time difference |t1-t0|> preset time threshold, p0 is reserved as the effective peak, and other peak and valley values are calculated in order to obtain the effective peak and the effective valley.

Assuming that |p1-v1|<preset amplitude threshold, or |t3-t2|<preset time threshold, valley value v1 is not a valid valley value, should be discarded, discard flag position 1; in order to ensure that the peak and valley values are sequentially interleaved A peak should be discarded accordingly; for the same reason, if a peak has been discarded, a valley should be discarded accordingly. Assuming that the valley point (t3, v1) is discarded, the flag position is discarded. According to the preset filtering rule, one of the peak point (t2, p1) and the peak point (t4, p2) is selected for discarding.

In this embodiment, the preset filtering rule is: first determining whether p1 and p2 are valid peaks according to the filtering method, and if one of the peaks is not a valid peak, then selecting to discard the value; if both p1 and p2 are Yes or no effective peak, discard one of the smaller signal amplitude differences, or discard the one with a smaller time difference (can be designed according to the actual situation), and discard the flag position 0.

In step 13, matching of the periodic actions is performed according to the effective peak value and the effective valley value.

In the plurality of effective peaks and effective valleys, selecting each effective peak and two effective valleys adjacent to the effective peak as an action group, respectively calculating a signal amplitude difference and a time difference; or, in the above plurality of valid In the peak value and the effective valley value, each effective valley value and two effective peaks adjacent to the effective valley value are selected as an action group, respectively calculating a signal amplitude difference and a time difference; and calculating a temporal phase according to the signal amplitude difference and the time difference The similarity between the two action groups of the neighbor; if the similarity of the two action groups meets the preset condition, each action group is recorded as a periodic action; each cycle action is recorded as a stroke number.

To ensure the strictness of the matching, the amplitude threshold and the time threshold are relatively high when the first two periodic actions are matched. In the subsequent matching, in order to ensure the continuity of the action, the amplitude threshold and the time threshold may be appropriately reduced. .

It is assumed that after step 12, the effective peak and the effective valley are obtained in chronological order: p0, v0, p1, v1, p2, v2, p3, ..., then, divided into (p0, v0, p1), (p1, Multiple action groups such as v1, p2, ), (p2, v2, p3), or multiple action groups such as (v0, p1, v1), (v1, p2, v2), respectively, are calculated in each action group. The signal amplitude difference and time difference between the effective peak value and the effective valley value, and the similarity between the two action groups is judged by this, for example, (p0, v0, p1) matches (p1, v1, p2,), and should correspond Match (p0, v0) with (p1, v1), (v0, p1) and (v1, p2), and match both phases to determine that two action groups match, and there are more than three consecutive similar action groups. Then, each action group is recorded as one cycle action, and each cycle action is recorded as one stroke number, that is, three strokes.

In step 14, the wearer's swimming state is determined according to the matching result.

When the number of strokes reaches a preset number of times, it is determined that the wearer is in a swimming state.

In this embodiment, the motion signal collected by the sensor is filtered, the non-standard signal is removed, the basic data amount is reduced, and the wearer is automatically determined according to the periodicity of the signal itself, and the wearer does not need to manually open the swimming mode. Compared with related technologies, a large number of template matching calculations are not required, which reduces the computational load of the smart watch and improves the recognition rate. This method has a good inhibitory effect on the problem of using false peaks and pseudo-valleys after conventional filtering.

Embodiment 2

This embodiment is improved on the basis of the above embodiment, and can measure the exercise mileage of the wearer, and the data is intuitive.

2 is a flow chart of a swimming motion analysis method provided in the second embodiment. As shown in FIG. 2, the swimming motion analysis method includes the following steps:

Step 21: Acquire a plurality of peaks and a plurality of valleys from the motion signals collected by the sensing elements on each coordinate axis of the multi-axis sensor of the smart watch.

Step 22, filtering the plurality of peaks and the plurality of valley values to obtain effective peaks and effective valleys therefrom.

Step 23: Perform matching of periodic actions according to the effective peak value and the effective bottom value.

In step 24, it is determined that the wearer is in a swimming state.

In step 25, a decision tree algorithm is used to determine the wearer's stroke based on the action signal of one cycle action.

Firstly, the collected large amount of data is sent to the decision tree algorithm, and the mean, variance, maximum value, minimum value, etc. are calculated to obtain a decision tree for judging the stroke; in the actual detection, after the wearer's motion state is stable, a single is selected. The six-axis data of the cycle (3-axis accelerometer and 3-axis gyroscope) recognizes the stroke based on the decision tree algorithm.

In step 26, the circle is calculated according to the wearer's stroke, the action signal, the magnetometer threshold, the stroke number threshold, and the lap time threshold.

When the wearer's swimming posture is unchanged, the magnetometer threshold corresponding to the swimming posture can be selected, and compared with the action signal of the magnetometer in the nine-axis sensor, if the wearer has obvious reentry action, it can be judged by the magnetometer data. Come out, remember to fold back; then get the stroke number threshold and the circle time threshold according to the historical data, and verify the above-mentioned foldback. If the time or the number of strokes is larger than the actual situation, it is obviously unable to count the circle. Then the above foldback is not counted.

If the wearer changes his stroke during the return, especially if he changes from backstroke to other strokes or from other strokes to backstroke, you need to reduce the magnetometer threshold, based on the time of the last fold and the number of strokes. Determine the number of strokes and the time threshold to determine whether to fold back.

Optionally, the length of the pool can be estimated by the number of strokes and time of each fold, the stroke, the height of the wearer, and the like, and the swimmer's swimming mileage can be calculated. Optionally, the swimmer's swimming speed is calculated by the length of the pool and the return time.

In the embodiment, on the basis of identifying the swimming state, the number of reentry times is counted, so that the foldback caused by the change of the swimmer's swimming posture is not obvious, and the circle is not counted, and the number of the laps and the swimming mileage are displayed to the wearer, which is more intuitive. Reflects the actual amount of exercise of the wearer.

Embodiment 3

In combination with the method of the above embodiment, the present embodiment can select a coordinate axis with better sensor data quality to perform matching of periodic actions when matching periodic actions, thereby effectively avoiding data confusion and less data processing.

FIG. 3 is a flowchart of a swimming motion analysis method according to Embodiment 3 of the present application. As shown in FIG. 3, the analysis method of the swimming movement includes the following steps:

Step 31: Acquire a plurality of peaks and a plurality of valleys from the motion signals collected by the sensing elements on each coordinate axis of the multi-axis sensor of the smart watch.

Step 32: Filter the plurality of peaks and the plurality of valley values to obtain an effective peak value and an effective valley value.

Step 33: Perform matching of periodic actions according to the effective peak value and the effective bottom value.

In step 34, the number of strokes matched by each coordinate axis of the sensor is separately counted.

The sensor can adopt a nine-axis sensor, including a three-axis accelerometer, a three-axis gyroscope and a three-axis magnetometer. The three-axis accelerometer, the three-axis gyroscope and the three-axis magnetometer all include three coordinate axes of X, Y and Z, respectively. Count the number of strokes that each axis matches.

Step 35: The motion signal quality of each coordinate axis is evaluated according to a preset rule and the number of strokes.

The preset rule is set to: set an evaluation value for each coordinate axis. Before the start of the matching of the periodic motion, the historical evaluation value is used to determine which coordinate axis is locked for periodic motion matching, or each coordinate axis performs periodic motion. Matching, dynamically update the evaluation value of each coordinate axis according to the matching situation, and then decide which axis to lock for periodic motion matching. The more matches the number of consecutive strokes, the higher the evaluation value, the greater the chance of locking the axis.

In step 36, the coordinate axis optimal for the motion signal is selected to continue the matching of the periodic motion.

Select the coordinate axis with higher evaluation value to perform the next motion matching. When the number of strokes of one coordinate axis is significantly larger than the other two axes, it is possible that the data of the coordinate axis is unreasonable. Choose the better one of the two axes.

In step 37, the swim state of the wearer is determined according to the matching result.

According to the number of strokes matched by the locked coordinate axis to determine whether the wearer is in a swimming state, the length of the swimming pool can be estimated by the number of strokes and time of each reentry, the stroke position, the height of the wearer and the like, and then the length of the pool can be estimated. The swimmer's swimming mileage.

In this embodiment, only one coordinate axis is selected for planning the number of waters at the same time. For some strokes, a certain coordinate axis data may be confusing. In this embodiment, a coordinate axis with high signal quality may be selected. , ignore the axis of signal confusion, reduce the amount of data processing.

Embodiment 4

The embodiment provides a smart watch for performing the swimming motion analysis method described in the above embodiment, and setting a corresponding function module or hardware structure according to the method, and solving the same technical problem with the above analysis method, achieving the same technology. effect.

4 is a schematic structural diagram of a smart watch provided in Embodiment 4 of the present application. As shown in FIG. 4, the smart watch includes a nine-axis sensor 41 and a processor 42, and the nine-axis sensor 41 includes a three-axis accelerometer, a three-axis gyroscope, and a three-axis magnetometer.

The nine-axis sensor 41 is arranged to collect an action signal of the wearer.

The processor 42 includes a signal screening module 421, a filtering module 422, an action matching module 423, and a state detecting module 424.

The signal screening module 421 is configured to acquire a plurality of peaks and a plurality of valleys in the motion signal.

The filtering module 422 is configured to filter the plurality of peaks and the plurality of valleys to obtain an effective peak and an effective valley.

The action matching module 423 is configured to perform matching of periodic actions according to the effective peak value and the effective bottom value.

The status detection module 424 is configured to determine the swim status of the wearer based on the matching result.

The filtering module 422 is configured to:

And selecting, according to a time sequence generated by the motion signal, a peak value and a valley value adjacent to the peak value, and calculating a signal amplitude difference and a time difference between the peak value and the valley value; when the signal amplitude difference is greater than a pre- When the amplitude threshold is set, and the time difference is greater than the preset time threshold, the peak is a valid peak; when the peak is an invalid peak, the valley is discarded, and two adjacent to the valley are selected One of the peaks is discarded; the above steps are repeated until all valid peaks are obtained.

And, according to the time sequence generated by the action signal, selecting a valley value and an adjacent peak value after the valley value, calculating a signal amplitude difference and a time difference; when the signal amplitude difference between the peak value and the valley value is greater than a preset An amplitude threshold, and when the time difference is greater than a preset time threshold, the valley is a valid valley; when the valley is an invalid valley, the peak is discarded, and two adjacent to the peak are selected One of the valleys is discarded; repeat the above steps until all effective valleys are obtained.

The action matching module 423 is configured to:

Setting, in a chronological order of the action signals, each of the effective peaks and the effective valleys, setting each of the effective peaks and the two effective valleys adjacent to the effective peak as an action group, Calculating a signal amplitude difference and a time difference respectively; or, in a plurality of the effective peaks and the effective valleys, selecting each effective valley value and two effective peaks adjacent to the effective valley value as an action group, The signal amplitude difference and time difference between the effective peak and the effective valley in an action group are calculated separately.

Calculating a similarity of two action groups adjacent in time according to the signal amplitude difference and the time difference; when the similarity meets a preset condition, each action group is recorded as one cycle action; The cycle action is recorded as the number of strokes.

The status detection module 424 is configured to:

Optionally, the state detecting module 424 is further configured to:

After determining that the wearer is in the swimming state, the decision tree algorithm is used to determine the swimmer's stroke according to the action signal of one action cycle; according to the wearer's stroke, the action signal, the magnetometer threshold, the stroke number threshold, and the The circle time threshold is counted.

Optionally, the action matching module 423 is further configured to:

After performing the matching of the periodic actions according to the effective peak value and the effective valley value, respectively The number of strokes matched to each coordinate axis of the sensor is determined; the motion signal quality of each coordinate axis is evaluated according to a preset rule and the number of strokes; and the coordinate axis with the optimal motion signal is selected to continue the matching of the periodic motion.

In this embodiment, the motion signals collected by the sensor are filtered, the non-standard signals are removed, and the amount of basic data is reduced, which has a good inhibitory effect on the problem of using pseudo-peaks and valleys after conventional filtering; The periodicity automatically determines whether the wearer is in a swimming state, and does not require the wearer to manually open the swimming mode. Compared with the related technology, a large amount of template matching calculation is not required, the computing load of the smart watch is reduced, and the recognition rate is improved. The number of foldbacks is counted in combination with the change of the swimmer's stroke, so that the foldback caused by the change of the swimmer's stroke is not obvious, and the circle is not counted, and the number of strokes and the swim mileage are displayed to the wearer, and the wearer is more intuitively reflected. The actual amount of motion; it is also possible to further select the high-signal axis of the signal quality, ignore the coordinate axis of the signal confusion, and reduce the amount of data processing.

An embodiment further provides a computer readable storage medium storing computer executable instructions for performing any of the methods described above.

FIG. 5 is a schematic diagram of a hardware structure of a smart watch according to an embodiment. As shown in FIG. 5, the smart watch includes: one or more processors 510 and a memory 520. One processor 510 is taken as an example in FIG.

The smart watch may further include an input device 530 and an output device 540.

The processor 510, the memory 520, the input device 530, and the output device 540 in the smart watch may be connected by a bus or other means, and the bus connection is taken as an example in FIG.

The input device 530 can receive input numeric or character information, and the output device 540 can include a display device such as a display screen.

The memory 520 is a computer readable storage medium that can be used to store software programs, computer executable programs, and modules. The processor 510 executes various functional applications and data processing by executing software programs, instructions, and modules stored in the memory 520 to implement any of the above embodiments.

The memory 520 may include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application required for at least one function; the storage data area may store data created according to the use of the smart watch, and the like. In addition, the memory may include a volatile memory such as a random access memory (RAM), and may also include a non-volatile memory, such as at least one magnetic Disk storage devices, flash memory devices, or other non-transitory solid state storage devices.

Memory 520 can be a non-transitory computer storage medium or a transitory computer storage medium. The non-transitory computer storage medium, such as at least one magnetic disk storage device, flash memory device, or other non-volatile solid state storage device. In some embodiments, memory 520 can optionally include a memory remotely located relative to processor 510 that can be connected to the smart watch over a network. Examples of the above networks may include the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

Input device 530 can be used to receive input numeric or character information and to generate key signal inputs related to user settings and function controls of the smart watch. The output device 540 can include a display device such as a display screen.

The smart watch of the present embodiment may also include a communication device 550 for transmitting and/or receiving information over a communication network.

A person skilled in the art can understand that all or part of the process of implementing the above embodiment method can be completed by executing related hardware by a computer program, and the program can be stored in a non-transitory computer readable storage medium. The program, when executed, may include the flow of an embodiment of the method as described above, wherein the non-transitory computer readable storage medium may be a magnetic disk, an optical disk, a read only memory (ROM), or a random access memory (RAM). Wait.

Industrial applicability

The present disclosure provides a swimming watch analysis method based on a smart watch and a smart watch, which can automatically recognize the wearer's swimming state through an action signal collected by the smart watch.

Claims

A swimming watch analysis method based on a smart watch, comprising:

Obtaining a plurality of peaks and a plurality of valley values from motion signals collected by the sensing elements on each coordinate axis of the multi-axis sensor of the smart watch;

Filtering the plurality of peaks and the plurality of valley values to derive effective peaks and effective valleys therefrom;

Performing a matching of periodic actions according to the effective peak value and the effective valley value;

The wearer's swimming state is determined based on the matching result.
The analysis method according to claim 1, wherein the step of filtering said plurality of peaks and said plurality of valley values to obtain effective peaks and effective valleys therefrom comprises:

And selecting, according to a time sequence generated by the motion signal, a peak value and a valley value adjacent to the peak value, and calculating a signal amplitude difference and a time difference between the peak value and the valley value;

When the signal amplitude difference is greater than a preset amplitude threshold, and the time difference is greater than a preset time threshold, the peak is an effective peak;

When the peak is an invalid peak, discarding the valley value, and selecting one of two peaks adjacent to the valley value according to a preset filtering rule to discard;

Repeat the above steps until all valid peaks are obtained.
The analysis method according to claim 2, wherein the step of filtering the plurality of peaks and the plurality of valley values to obtain an effective peak value and an effective valley value therefrom further comprises:

And selecting a valley value and a peak value adjacent to the valley value according to a time sequence generated by the motion signal, and calculating a signal amplitude difference and a time difference;

And when the signal amplitude difference between the peak value and the bottom value is greater than the preset amplitude threshold, and the time difference is greater than the preset time threshold, the valley value is a valid valley value;

When the valley value is an invalid valley value, discarding the peak value, and selecting one of two valley values adjacent to the peak value according to a preset filtering rule to discard;

Repeat the above steps until all effective troughs are obtained.
The analysis method according to claim 1, wherein the step of matching the periodic actions according to the effective peak value and the effective bottom value comprises:

Setting, in a chronological order of the action signals, each of the effective peaks and the effective valleys, setting each of the effective peaks and the two effective valleys adjacent to the effective peak as an action group, Calculating the signal amplitude difference and time difference between the effective peak and the effective valley in an action group, respectively; or

Setting, in a chronological order of the action signals, each of the effective peaks and the effective valleys, setting each effective valley value and two effective peaks adjacent to the effective valley value as one action group , respectively Calculating the signal amplitude difference and time difference between the effective peak and the effective valley in an action group;

Calculating a similarity of two action groups adjacent in time according to the signal amplitude difference and the time difference;

When the similarity meets a preset condition, each action group is recorded as a periodic action;

Record each cycle action as the number of strokes.
The analysis method according to claim 1 or 4, wherein the step of determining the swim state of the wearer based on the matching result comprises:

When the total number of strokes reaches a preset number of times, it is determined that the wearer is in a swimming state.
The analysis method according to claim 5, wherein after the step of determining that the wearer is in a swimming state, the method further comprises:

The decision tree algorithm is used to determine the wearer's stroke based on the action signal of a periodic action.
The analysis method according to claim 6, wherein after the step of determining that the user is in a swimming state, the method further comprises:

The circle is counted according to the wearer's stroke, the motion signal, the magnetometer threshold, the stroke number threshold, and the lap time threshold.
The analysis method according to claim 4, wherein after the step of matching the periodic actions according to the effective peak value and the effective bottom value, the method further comprises:

The number of strokes matched by each coordinate axis of the multi-axis sensor is separately counted;

Evaluating the motion signal quality of each coordinate axis according to a preset rule and the number of strokes;

Select the coordinate axis with the best motion signal quality to continue the matching of the periodic motion.
A smart watch comprising: a nine-axis sensor and a processor,

The nine-axis sensor is configured to collect an action signal of the wearer;

The processor includes a signal screening module, a filtering module, an action matching module, and a state detecting module;

The signal screening module is configured to acquire a plurality of peaks and a plurality of valleys in the motion signal;

The filtering module is configured to filter the plurality of peaks and the plurality of valley values to obtain an effective peak value and an effective valley value therefrom;

The action matching module is configured to perform matching of periodic actions according to the effective peak value and the effective bottom value;

The state detection module is configured to determine a swim state of the wearer based on the matching result.
The smart watch of claim 9 wherein said filtering module is configured to:

Selecting a peak and an adjacent valley after the peak according to the time sequence generated by the motion signal a value, a signal amplitude difference and a time difference between the peak value and the valley value are calculated;

When the signal amplitude difference is greater than a preset amplitude threshold, and the time difference is greater than a preset time threshold, the peak is an effective peak;

When the peak is an invalid peak, discarding the valley value, and selecting one of two peaks adjacent to the valley value according to a preset filtering rule to discard;

Repeat the above steps until all valid peaks are obtained;

And selecting a valley value and a peak value adjacent to the valley value according to a time sequence generated by the motion signal, and calculating a signal amplitude difference and a time difference;

When the signal amplitude difference between the peak value and the valley value is greater than a preset amplitude threshold, and the time difference is greater than a preset time threshold, the valley value is a valid valley value;

When the valley value is an invalid valley value, discarding the peak value, and selecting one of two valley values adjacent to the peak value according to a preset filtering rule to discard;

Repeat the above steps until all effective troughs are obtained.
The smart watch of claim 9, wherein the action matching module is configured to:

Setting, in a chronological order of the action signals, each of the effective peaks and the effective valleys, setting each of the effective peaks and the two effective valleys adjacent to the effective peak as an action group, Calculating the signal amplitude difference and time difference between the effective peak and the effective valley in an action group, respectively; or

Setting, in a chronological order of the action signals, each of the effective peaks and the effective valleys, setting each effective valley value and two effective peaks adjacent to the effective valley value as one action group , respectively calculating the signal amplitude difference and time difference between the effective peak and the effective valley in an action group;

Calculating a similarity of two action groups adjacent in time according to the signal amplitude difference and the time difference;

When the similarity meets a preset condition, each action group is recorded as a periodic action;

Each cycle action is recorded as the number of strokes.
The smart watch according to claim 9 or 11, wherein the state detecting module is set to:

When the total number of strokes reaches a preset number of times, it is determined that the wearer is in a swimming state.
The smart watch according to claim 12, wherein the state detecting module is further configured to: after determining that the wearer is in a swimming state,

Determining the wearer's stroke using a decision tree algorithm based on the action signal of a periodic action;

According to the wearer's stroke, the action signal, the magnetometer threshold, the stroke number threshold, and the time of the circle The threshold is counted.
The smart watch according to claim 11, wherein the action matching module is further configured to: after performing matching of the periodic actions according to the effective peak value and the effective bottom value,

Separating the number of strokes matched by each coordinate axis of the nine-axis sensor;

Evaluating the motion signal quality of each coordinate axis according to a preset rule and the number of strokes;

The coordinate axis optimal for the motion signal is selected to continue the matching of the periodic motion.
A computer readable storage medium storing computer executable instructions for performing the method of any of claims 1-8.