CN110547805A - Real-time gait analysis method based on plantar pressure - Google Patents

Abstract

The invention relates to a real-time gait analysis method based on plantar pressure. The specific steps are as follows: using a wearable insole to obtain plantar pressure data under the state of human activity; Through comprehensive comparison, the extraction and identification of each independent activity state are realized, the kinematic parameters of gait are calculated at the same time, and the activity state is evaluated and analyzed; finally, the continuous activity state is statistically sorted. The invention can realize point-by-point real-time extraction and analysis of human body long-range activity information based on wearable devices, and can be applied to the fields of gait recognition, daily health detection, and evaluation of related diseases.

Description

A real-time plantar pressure-based gait analysis method

technical field

The invention relates to a real-time gait analysis method based on plantar pressure, which is a gait analysis and evaluation method based on a wearable device.

Background technique

There is a high correlation between the dynamic information of the sole of the foot and the physiological structure of the human body and the walking mode. The acquisition and analysis of gait pressure information is helpful to discover the detailed nature of human motion, and it is not only of great significance to the evaluation of related diseases, such as small broken steps in patients with Parkinson's disease, freezing gait and hemiplegic gait in patients with stroke, etc.; At the same time, it can also be applied to fields such as gait recognition (identity recognition).

At present, the auxiliary means for gait analysis mainly include multi-camera image sequence acquisition, nuclear magnetic resonance, 3D force platform, pressure trail, and wearable devices such as smart insoles. With the help of cameras, force measuring platforms, nuclear magnetic resonance and other methods, there are disadvantages such as high cost, complex testing process or high operational requirements, and it is impossible to conduct more popular test evaluations. In contrast, monitoring the daily activity status of patients based on wearable devices and providing quantitative analysis results can provide a more objective basis for doctors' diagnosis. At the same time, it can also provide scientific and effective guidance for patients' rehabilitation in real time.

The smart insole with pressure sensor is more suitable for gait analysis in the human body's daily activity state. However, the existing pressure sensor-based gait analysis has problems such as insufficient use of pressure information; the state classification process is cumbersome; it cannot explain and analyze other detailed information of the active state, so it is difficult to achieve an effective evaluation of the active state and It is difficult to meet the real-time requirements.

SUMMARY OF THE INVENTION

In view of the above problems, the purpose of the present invention is to propose a real-time plantar pressure-based gait analysis method. The method of the invention can realize the classification and evaluation of different activity states of the human body through the real-time analysis of the dynamic changes of the plantar pressure.

A real-time gait analysis method based on plantar pressure signal proposed by the present invention, the specific steps are as follows:

(1) Parameter initialization:

Two thresholds of the plantar pressure signal of each subject are set based on the individual differences of the subjects: the amplitude threshold th1 at the sole of the foot and the amplitude threshold th2 at the heel; the judgment time threshold of the sitting state is set to t_sit, and the corresponding The counting parameter count_sit is 0; the judgment time threshold of the station status is set to t_std, and the corresponding count parameter count_std is 0; the judgment time threshold of the standing and sitting status is set to t_st when the status is sorted, and the judgment time threshold of the walking status is t_wk;

(2) Read the pressure signal of the human body's bipedal or monopodial under various active states point by point:

The pressure signal is the pressure data collected at a certain sampling frequency f by a pressure sensor placed on the sole of the foot in the insole; the pressure sensor on the sole of the foot is generally located at the sole and the heel, and multiple pressures can be set separately at the sole and heel. sensors to obtain multi-channel pressure data; generally, the number of pressure sensors set at the sole of the foot is not less than 2, and the number of pressure sensors at the heel is not less than 1;

(3) Identify and evaluate the independent activity state of the pressure data obtained in step (2):

Detect whether each pressure data has at least one rising edge, that is, determine whether the pressure amplitude at the sole of the foot is greater than the amplitude threshold th1 at the sole of the foot or whether the pressure amplitude at the heel is greater than the amplitude threshold th2 at the heel; If it is detected that there is a rising edge in at least one way, the window length value wlen of the sliding window will be adaptively selected, and the plantar power state (standing, walking, running) will be executed adaptively. and up (down) stairs) and extract the corresponding feature parameters;

The window length wlen of the sliding window needs to be adjusted according to the pressure at the heel: when the peak pressure at the heel is greater than the amplitude threshold th2 at the heel, the window length of the observation window is set to wlen1; otherwise, it is considered that the corresponding high frequency is at this time. If a motion state, such as running, requires a smaller analysis window length, select the observation window window length as wlen2, where: wlen2<wlen1;

Among them: the determination of the state of plantar weakness (sitting) includes the following steps:

Count the data points before the rising edge detected by each channel of pressure data, that is, if a data point appears, then count_sit is incremented by 1 until the rising edge is detected;

② Judgment steps Whether the count_sit after counting is greater than f * t_sit; if it is greater, that is, the accumulated period exceeds the set time threshold t_sit, it is marked as sitting state and the start and end points of the state are recorded, and the count parameter count_sit is cleared at the same time;

The determination of plantar force status includes the following:

① Set the initial length of the data segment to be analyzed through the adaptive sliding window window length wlen, that is, when the peak pressure at the heel is greater than the amplitude threshold th2 at the heel, set the initial length of the data segment to be analyzed as wlen1; otherwise wlen2;

② Extract characteristic parameters including stride frequency, peak value, landing delay, effective pressure duration and corresponding start and end time from each pressure data;

③ Determine whether the falling edge of the pressure data curve is detected in the current observation window, that is, the pressure value at the sole is less than the amplitude threshold th1 at the sole or the pressure value at the heel is less than the amplitude threshold th2 at the heel; if so, perform each dynamic motion state ( Identification and evaluation of walking, running, and going up (down) stairs; if not, judgment and marking of standing status;

(4) Determine whether the state classification and evaluation process based on pressure data continues; if so, return to step (2); if not, perform statistical sorting on each independent activity state, and mark the corresponding transition state.

In the present invention, in the determination of the power state of the sole of the foot, the step frequency in step (2) is the reciprocal of the time difference between two adjacent landings, wherein the landing time is defined as the minimum value of the peak time of each channel of pressure data; the peak value is the maximum value of the relative amplitude of pressure in each pressure data curve under the current sliding window; the landing delay is the time difference between the pressure peak moment at the heel and the pressure peak moment at the sole; the effective pressure duration and The start and end times are the time periods and start and end times when each pressure curve is within the range of the respectively set amplitude thresholds under the current observation window.

In the present invention, in the determination of the power state of the sole of the foot, the identification and evaluation methods of each dynamic motion state (walking, running, and going up (down) stairs) described in step ③ are as follows: if the landing time delay is positive and relatively If it is small, when the stride frequency is high, it is marked as running (RUN); when the stride frequency is low, it is marked as going up (down) stairs (Stair); if the landing delay is positive and large, it is marked as normal walking (WALK). ); if the landing delay is negative, it is marked as abnormal walking (Abnorm) where the toe touches the ground before the heel; if the pressure at the sole or heel is low under the given amplitude thresholds (th1 and th2), the If the absolute value of the difference between the two pressure peaks on the sole of the foot is greater than a certain set value, the balance state of the force applied by the sole of the foot is marked as unbalanced (balance: No), otherwise it is balanced ( balance : Yes);

The method for judging and marking the station status is as follows: assign the value f *wlen to the counting parameter count_std of the station status, and calculate the duration that the pressure is greater than the threshold value, that is, every time the pressure sensor data is read, if the falling edge is not detected, then Increase count_std by 1; when count_std is greater than f * t_std, mark the station status and clear the counter count_std; otherwise, once a falling edge is detected, clear the counter count_std, update to step (2), and enter the analysis of subsequent data.

In the present invention, the selection criterion of the pressure peak moment at the sole of the foot described in step (3) is: since the number of sensors set at the sole of the foot is not less than 2, when the pressure peaks of each pressure sensor at the sole of the foot exist, then take The minimum value of the corresponding time of each channel; when the pressure peak of only one pressure sensor exists, the corresponding time is selected as the pressure peak time at the sole of the foot; if the peak value of each pressure sensor does not exist, the end point of the current observation window is taken The moment is the moment of peak pressure on the sole of the foot.

In the present invention, the abnormal state of the sole force applied in step (3) includes two situations of insufficient force applied to the sole (toe-less) and insufficient force applied to the heel (heel-less). When the maximum value of each signal pressure peak value at the sole of the foot is less than 2 times th1, it is marked as toe-less; when the maximum value of each signal pressure peak value at the heel is less than 1.5 times th2, it is marked as Heel-less.

In the present invention, the statistical sorting of the states in step (4) refers to marking each independent activity state in a statistical sense. For sitting (SIT) and standing (STD); the activities of walking, running and going up (down) stairs with a state duration greater than or equal to t_wk are uniformly marked as walking (WLK); and follow: WLK-TURN-WLK, WLK- The definitions of WLKSIT-SIT, WLK-WALKSTD-STD, SIT-SITWLK-WLK, SIT-SITSTD-STD, STD-STDWLK-WLK and STD-STDSIT-SIT mark transitions between corresponding states; among them, WLK-TURN -WLK indicates that the two consecutive states before and after are marked as WLK (walking), and the transition state during the period is defined as TURN (turning); WLK-WLKSIT-SIT indicates that the two preceding and following states are WLK (walking) and SIT (sitting) respectively. , the transition state during the period is defined as WLKSIT (Walking to Sitting); and so on; in addition, the unrecognized state is marked as UnKnown.

The present invention has the following beneficial effects:

1. The processing algorithm involved in the present invention is simple, can realize real-time data analysis, and can timely evaluate the activity state of the subject;

2. The present invention can realize the extraction and analysis of long-range activity information of the human body based on wearable devices, and is expected to be applied to the monitoring of the daily activity state of the human body;

3. The characteristic parameters extracted by the method of the present invention can be effectively used for the assessment of the activity state, suitable for fields such as gait analysis and identity recognition, and can also assist in the diagnosis, treatment and rehabilitation of related diseases.

Description of drawings

Figure 1 is a general flow diagram of the present invention.

FIG. 2 is a schematic diagram of a specific flow of “classification and evaluation of activity states” in FIG. 1 .

FIG. 3 is a schematic diagram of extracting characteristic parameters of a single activity by the method of the present invention. The abscissa is time, and the ordinate is the relative amplitude of plantar pressure. The three curves in the window are the pressure curves corresponding to two points on the forefoot and one point on the heel. (t, V) represents the coordinates of the peak point of each curve, t is the time, and V is the pressure amplitude. th1 and th2 are the foot pressure amplitude threshold and the heel pressure amplitude threshold, respectively. Hw1, Hw2 and Hw3 represent the effective pressure duration corresponding to the two-way ball and one-way heel pressure data, respectively.

FIG. 4 is a schematic diagram of the evaluation result of the independent walking state based on the adaptive sliding window of the present invention. The abscissa is time, and the ordinate is the relative amplitude of plantar pressure. Each dark rectangular area in the figure corresponds to an observation window. The three curves in the window are the dynamic pressure change curves of two soles and one heel respectively. The curve with a higher peak reflects the dynamic pressure change at the heel, and the other two The curves correspond to the two pressure changes at the sole of the foot, respectively. In the figure, WALK is the record of the independent walking state; balance is used to indicate the balance of the current walking state; delay is the landing delay between the sole and the heel; Abnormal state of bottom force; freqz refers to cadence.

FIG. 5 is a schematic diagram of the state sorting result of the present invention for the movement process of walking and sitting. The abscissa is time, and the ordinate is the relative amplitude of plantar pressure. The three curves in each cycle are the pressure dynamic change curves of two channels of soles and one channel of heels. Among them, the light-colored curve with a relatively high peak value represents the pressure change at the heel, and the other two curves are the pressure changes of the two channels of signals at the sole of the foot. Pressure changes; stand and Sit represent independent standing and sitting states, respectively, and the meanings of other symbols are the same as in Figure 4. The color segments of different grayscales in Figure 5 correspond to different activity state sorting results, and the corresponding relationship between grayscales and states is shown in Figure 6.

FIG. 6 is a schematic diagram of grayscale marks corresponding to the statistical arrangement of active states in Embodiment 1 of the present invention, wherein 0-10 are numbers corresponding to each state.

Detailed ways

The method of the present invention and its application will be further described in detail below with reference to the accompanying drawings and embodiments. The examples are only used to illustrate the present invention and not to limit the conditions. On the basis of the technical solutions of the present invention, any modification or modification of the above-mentioned embodiments according to the principles of the present invention shall not be excluded from the protection scope of the present invention.

Example 1: The movement state of the subject from walking to sitting was analyzed and evaluated by the plantar pressure-based gait analysis method of the present invention. The pressure data in this embodiment is the plantar pressure data of the subject's single foot collected by using the wearable insole, in which 2 channels are collected at the sole of the foot and 1 channel is collected at the heel, totaling 3 channels of signals; the sampling frequency f is 50Hz. The workflow is as follows:

(1) Parameter initialization:

In this embodiment, the amplitude threshold th1 at the sole of the foot is set to be 0.2 times the peak pressure of the sole of the subject under the general activity state (see the dotted line corresponding to the amplitude threshold th1 at the sole of the foot in Fig. 3 ); the amplitude threshold th2 at the heel It is set to be 0.3 times the peak heel pressure of the subject in the general activity state (see the dotted line corresponding to the amplitude threshold th2 at the heel in Figure 3); the judgment time thresholds for the independent sitting state and the independent standing state are set to 1s, that is t_sit = t_std = 1s, the corresponding counting parameters count_sit and count_std are both initialized to 0; when setting the state sorting time threshold for standing and sitting states t_st = 3.5s, for walking state determination time threshold t_wk = 2s; the setting is optional The sliding window window length values are wlen1 = 0.7s and wlen2 = 0.5s respectively;

(2) 3 channels of pressure data are read point by point;

(3) Identify and evaluate the independent activity state of the read pressure data:

The 3-way pressure data read in step (2) is compared with the amplitude threshold th1 at the sole of the foot and the amplitude threshold th2 at the heel, respectively, to detect whether there is a rising edge and the force state of the sole of the foot. When the heel pressure peak value is greater than the amplitude threshold th2 at the heel, a rising edge occurs and indicates the power of the sole of the foot. Therefore, it is transferred to the judgment of the strength of the sole of the foot, and the window length of the sliding window is selected as wlen1=0.7s. As shown in Figure 3, information such as pressure peak and corresponding time are extracted from each pressure curve, and the coordinates of the peak point of each curve are represented by (t, V) with subscripts (t is the time, V is the pressure amplitude). The minimum value of the peak time in the road pressure data is recorded as the landing time; then mark the start and end time of each pressure curve within the range of the respective set amplitude thresholds, and calculate the effective pressure duration (Hw1, Hw2 and Hw3 respectively). Indicates the effective pressure duration corresponding to the two-way sole and one-way heel pressure data); the stride frequency (freqz) is calculated according to the reciprocal of the time difference between two adjacent landings; the sole is obtained according to the difference between the peak pressure moment at the heel and the peak pressure moment at the sole Characteristic parameters such as the landing delay (delay) between the heel and the heel.

Under the current observation window, if the pressure amplitude is smaller than the amplitude threshold th1 at the sole of the foot or the amplitude of the heel pressure is smaller than the amplitude threshold th2 at the heel in the two-way sole pressure data, that is, the falling edge of the pressure curve is detected, and the Identification and evaluation of dynamic motion state; and marking and evaluation of specific independent states according to the characteristic parameters calculated above. As shown in Figure 4, in this embodiment, the landing delay is greater than 0.2s, so it is marked as a walking state (WALK); the absolute value of the difference between the peak pressures of the two-way soles is greater than 1.5 times th1, so the balance status is marked as balance: No ; The maximum value of the two-way pressure peak is less than 2 times th1, so it is marked as a toe-less abnormal state of plantar force.

If the falling edge of any channel of pressure data is not detected in an observation window, and the peak pressure at the heel is greater than th2, the value f * 0.7s is assigned to the count parameter count_std of the station state, and the duration of the pressure data above the threshold is calculated. In this embodiment, since the duration is longer than t_std=1s, the corresponding independent active state is marked as a station (corresponding to the middle color segment among the three gray-scale color segments in FIG. 5 ).

In addition, when the three-way pressure data values are all about 0, that is, there is no rising edge, the judgment of the plantar weakness state is turned to. Before the rising edge is detected, the count parameter count_sit of the sitting state is accumulated. In this embodiment, since count_sit is greater than f * t_sit, that is, the duration of the plantar weakness state before the rising edge is greater than t_sit=1s, the independent active state is marked as sitting (corresponding to the three gray-scale color segments in Fig. 5 ). the right color segment).

(4) Determine whether the analysis of the pressure data is completed. If the analysis is completed, the active state is statistically sorted, and the corresponding transition state is marked.

As shown in Figure 5, since the total duration of each independent walking state is greater than t_wk, the continuous independent walking state is marked as WLK (shown in the light-colored segment on the left in Figure 5, using the grayscale corresponding to the WLK state in Figure 6) Number 1); similarly, organize and mark the continuous independent sitting state as SIT (shown in the dark section on the right in Figure 5, using the grayscale number 2 corresponding to the SIT state in Figure 6); and the process of walking to sitting , the previously marked independent station state (shown in the middle color segment in Figure 5, using the grayscale number 3 corresponding to the STD state in Figure 6) is analyzed as a transition segment because its duration is less than t_st, and it is analyzed by the adjacent front and rear The state is marked as WLKSIT, and its corresponding gray segment is finally represented by the corresponding gray number 5 in FIG. 6 .

Claims (7)

1. a real-time gait analysis method based on plantar pressure signal, its characteristic concrete steps are as follows: (1) Parameter initialization: Two thresholds of the plantar pressure signal of each subject are set based on the individual differences of the subjects: the amplitude threshold th1 at the sole of the foot and the amplitude threshold th2 at the heel; the judgment time threshold of the sitting state is set to t_sit, and the corresponding The counting parameter count_sit is 0; the judgment time threshold of the station status is set to t_std, and the corresponding count parameter count_std is 0; the judgment time threshold of the standing and sitting status is set to t_st when the status is sorted, and the judgment time threshold of the walking status is t_wk; (2) Read the pressure signal of the human body's bipedal or monopodial under various active states point by point: The pressure signal is the pressure data collected at a certain sampling frequency f by a pressure sensor placed on the sole of the foot in the insole; the pressure sensor on the sole of the foot is generally located at the sole and the heel, and multiple pressures can be set separately at the sole and heel. sensors to obtain multi-channel pressure data; generally, the number of pressure sensors set at the sole of the foot is not less than 2, and the number of pressure sensors at the heel is not less than 1; (3) Identify and evaluate the independent activity state of the pressure data obtained in step (2): Detect whether each pressure data has at least one rising edge, that is, determine whether the pressure amplitude at the sole of the foot is greater than the amplitude threshold th1 at the sole of the foot or whether the pressure amplitude at the heel is greater than the amplitude threshold th2 at the heel; If it is detected that there is a rising edge in at least one way, the window length value wlen of the sliding window will be adaptively selected, and the plantar power state (standing, walking, running) will be executed adaptively. and up (down) stairs) and extract the corresponding feature parameters; The window length wlen of the sliding window needs to be adjusted according to the pressure at the heel: when the peak pressure at the heel is greater than the amplitude threshold th2 at the heel, the window length of the observation window is set to wlen1; otherwise, it is considered that the corresponding high frequency is at this time. If a motion state, such as running, requires a smaller analysis window length, select the observation window window length as wlen2, where: wlen2<wlen1; Among them: the determination of the state of plantar weakness (sitting) includes the following steps: Count the data points before the rising edge detected by each channel of pressure data, that is, if a data point appears, then count_sit is incremented by 1 until the rising edge is detected; ② Judgment steps Whether the count_sit after counting is greater than f * t_sit; if it is greater, that is, the accumulated period exceeds the set time threshold t_sit, it is marked as sitting state and the start and end points of the state are recorded, and the count parameter count_sit is cleared at the same time; The determination of plantar force status includes the following: ① Set the initial length of the data segment to be analyzed through the adaptive sliding window window length wlen, that is, when the peak pressure at the heel is greater than the amplitude threshold th2 at the heel, set the initial length of the data segment to be analyzed as wlen1; otherwise wlen2; ② Extract characteristic parameters including stride frequency, peak value, landing delay, effective pressure duration and corresponding start and end time from each pressure data; ③ Determine whether the falling edge of the pressure data curve is detected in the current observation window, that is, the pressure value at the sole is less than the amplitude threshold th1 at the sole or the pressure value at the heel is less than the amplitude threshold th2 at the heel; if so, perform each dynamic motion state ( Identification and evaluation of walking, running, and going up (down) stairs; if not, judgment and marking of standing status; (4) Determine whether the state classification and evaluation process based on pressure data continues; if so, return to step (2); if not, perform statistical sorting on each independent activity state, and mark the corresponding transition state. 2. The method according to claim 1, wherein: in the determination of the power state of the sole of the foot, the step frequency described in step 2 is the reciprocal of the time difference between two adjacent landings, wherein the landing time is defined as the pressure of each road. The minimum value of the peak time of the data; the peak value is the maximum value of the relative amplitude of the pressure in each pressure data curve under the current sliding window; the landing delay is the time between the peak pressure moment at the heel and the pressure peak moment at the sole of the foot. Time difference; the effective pressure duration and start and end times are the time periods and start and end times of each pressure curve within the respectively set amplitude threshold range under the current observation window. 3. The method according to claim 2, characterized in that: in the determination of the power state of the sole of the foot, the identification and evaluation method of each dynamic motion state (walking, running and going up (down) stairs) described in step 3. They are: if the landing delay is positive and small, when the stride frequency is high, it is marked as running (RUN); when the stride frequency is low, it is marked as going up (down) stairs (Stair); if the landing delay is If it is positive and larger, it is marked as normal walking (WALK); if the landing delay is negative, it is marked as abnormal walking (Abnorm) where the toe touches the ground before the heel; if it is limited by the given amplitude thresholds (th1 and th2) If the pressure value at the sole or heel is low, it is marked as an abnormal state of plantar force; if the absolute value of the difference between two pressure peaks on the sole of the foot is greater than a certain set value, the balance of the force applied by the sole of the foot is marked as not. Balance (balance : No), otherwise balance (balance : Yes); The method for judging and marking the station status is as follows: assign the value f *wlen to the counting parameter count_std of the station status, and calculate the duration that the pressure is greater than the threshold value, that is, every time the pressure sensor data is read, if the falling edge is not detected, then Increase count_std by 1; when count_std is greater than f * t_std, mark the station status and clear the counter count_std; otherwise, once a falling edge is detected, clear the counter count_std, update to step (2), and enter the analysis of subsequent data. 4. The method according to claim 1, characterized in that: in step (3), the selection criteria for the pressure peak moment at the sole of the foot is: since the number of sensors set at the sole of the foot is not less than 2, when the If the pressure peaks of all the pressure sensors exist, the minimum value of the corresponding time of each channel is taken; when there is only one pressure peak of the pressure sensor, the corresponding time is selected as the pressure peak time at the sole of the foot; if the peak value of each pressure sensor If neither exists, the end time of the current observation window is taken as the pressure peak time at the sole of the foot. 5 . The method according to claim 1 , wherein the abnormal state of the sole force application in step (3) includes two types of insufficient force application on the sole of the foot (toe-less) and insufficient force application on the heel (heel-less). a situation. 6. When the maximum value of each signal pressure peak value at the sole of the foot is less than 2 times th1, it is marked as toe-less; when the maximum value of each signal pressure peak value at the heel is less than 1.5 times th2, Marked as heel-less. 7. The method according to claim 1, characterized in that: the statistical sorting of the states in step (4) refers to marking each independent activity state in a statistical sense. (Sit) and stand (stand) states are marked as sitting (SIT) and standing (STD) respectively; the activities of walking, running and going up (down) stairs with a state duration greater than or equal to t_wk are uniformly marked as walking (WLK); And according to the definitions of: WLK-TURN-WLK, WLK-WLKSIT-SIT, WLK-WALKSTD-STD, SIT-SITWLK-WLK, SIT-SITSTD-STD, STD-STDWLK-WLK and STD-STDSIT-SIT WLK-TURN-WLK means that when the two consecutive states before and after are marked as WLK (walking), the transition state during the period is defined as TURN (turning); WLK-WLKSIT-SIT means that the two states before and after are respectively When it is WLK (walking) and SIT (sit), the transition state during the period is defined as WLKSIT (walking to sitting); and so on; in addition, the unrecognized state is marked as UnKnown.