Gesture monitoring method and system based on smart ring

Abstract

The present invention relates to the field of data processing, and provides a gesture monitoring method and system based on a smart ring, including: using a sensor to collect sensor data corresponding to a gesture in real time at a first sampling frequency to obtain current data; determining an optimization factor of the current data based on the change between the current data and the reference data before the current data; the reference data before the current data is a preset number of sensor data that is continuous in time with the current data; calculating the similarity between the current data and a preset gesture instruction sequence based on the optimization factor; determining the sampling frequency optimization factor of the sensor based on the similarity, and optimizing the first sampling frequency of the sensor using the sampling frequency optimization factor, so that the sensor collects sensor data corresponding to the gesture in real time at a second sampling frequency. The method can optimize the sampling frequency of sensor data, optimize the accuracy of gesture monitoring of the smart ring, and improve the accuracy of gesture recognition.

Description

Technical Field

[0001] The present application relates to the field of data processing, and in particular to a gesture monitoring method and system based on a smart ring. Background Art

[0002] During the use of smart rings, it is necessary to monitor the data of various sensors in the smart rings, upload the real-time collected sensor data to the cloud computing server, and perform gesture recognition in the cloud computing server. The various sensors of smart rings include accelerometers, gyroscopes, magnetometers, etc. In the process of gesture recognition, it is necessary to recognize real-time data, so gesture recognition needs to be completed by time series data streaming matching.

[0003] The existing technology monitors the sensor data of the smart ring at every moment, and activates gesture recognition by comparing the similarity between the time series window and the known gesture command data. When the gesture command is detected, it is necessary to continue matching with the subsequent newly added sensor data until the gesture command corresponding to a period of time series data is determined. However, in the process of gesture command recognition, because there will be similar gestures in the sensor data corresponding to the set gesture command and because each user has different movement habits, even if it is the same gesture, there will be different sensor data fluctuations.

[0004] In the process of data collection by multiple sensors in existing smart rings, sensor data is collected through a fixed sampling frequency, which may cause inaccurate recognition when recognizing gestures for different usage habits. Summary of the invention

[0005] The present invention provides a gesture monitoring method and system based on a smart ring, which can improve the accuracy of gesture recognition.

[0006] In a first aspect, the present application provides a gesture monitoring method based on a smart ring, comprising:

[0007] Using the sensor to collect sensor data corresponding to the gesture in real time at a first sampling frequency to obtain current data;

[0008] Determine an optimization factor of the current data based on a change between the current data and reference data before the current data; the reference data before the current data is a preset number of sensor data that are continuous in time with the current data;

[0009] Calculating the similarity between the current data and a preset gesture instruction sequence based on the optimization factor;

[0010] A sampling frequency optimization factor of the sensor is determined based on the similarity, and the first sampling frequency of the sensor is optimized using the sampling frequency optimization factor, so that the sensor collects sensor data corresponding to the gesture in real time at a second sampling frequency.

[0011] Optionally, determining the optimization factor of the current data based on the change between the current data and the reference data before the current data includes:

[0012] Calculate the outlier degree of current data based on reference data;

[0013] Calculate the first average similarity between the data sequence with the jth data in the reference data as the starting data and the known gesture instruction sequence before obtaining the current data;

[0014] After obtaining the current data, calculate the second average similarity between the data sequence with the jth data in the reference data as the starting data and the known gesture instruction sequence;

[0015] An optimization factor for current data is determined based on the outlier degree and a difference between the first average similarity and the second average similarity.

[0016] Optionally, the optimization factor of the current data is calculated using the following formula:

[0017]

[0018] Among them, COF K＝10 (p i ) represents the current data p i The outlier factor obtained through reference data, the number of reference data is K, the value is 10, Norm(COF K＝10 (p i )) represents the outlier factor COF K＝10 (p i ) is the degree of outlier obtained after linear normalization, △ε ji represents the difference between the first average similarity and the second average similarity.

[0019] Optionally, calculating the similarity between the current data and a preset gesture instruction sequence based on the optimization factor includes:

[0020] Calculate the initial similarity between the data sequence starting with the current data and the preset gesture command sequence;

[0021] The initial similarity is processed based on the optimization factor to obtain the similarity between the current data and the preset gesture instruction sequence.

[0022] Optionally, the similarity between the current data and the preset gesture instruction sequence is calculated using the following formula:

[0023]

[0024] in, is the optimization factor for the current data, Indicates calculating the data sequence starting with the current data With the preset gesture command sequence T c The distance between them indicates the similarity between them, Norm() indicates linear normalization, ε i,c Indicates the similarity between the current data and the preset gesture command sequence.

[0025] Optionally, determining a sampling frequency optimization factor of the sensor based on the similarity includes:

[0026] Calculate a first similarity between a data sequence starting with the current data and the most similar gesture instruction data in a preset gesture instruction sequence, and calculate a second similarity between a data sequence starting with the current data and the second most similar gesture instruction data in a preset gesture instruction sequence;

[0027] A sampling frequency optimization factor of the sensor is determined based on the difference between the first degree of similarity and the second degree of similarity.

[0028] In a second aspect, the present application provides a gesture monitoring system based on a smart ring, comprising:

[0029] A collection module, used to collect sensor data corresponding to the gesture in real time using a sensor at a first sampling frequency to obtain current data;

[0030] An optimization factor calculation module, used to determine the optimization factor of the current data based on the change between the current data and the reference data before the current data; the reference data before the current data is a preset number of sensor data that are continuous in time with the current data;

[0031] A similarity calculation module, used for calculating the similarity between the current data and a preset gesture instruction sequence based on the optimization factor;

[0032] The optimization module is used to determine a sampling frequency optimization factor of the sensor based on the similarity, and optimize the first sampling frequency of the sensor using the sampling frequency optimization factor, so that the sensor collects sensor data corresponding to the gesture in real time at a second sampling frequency.

[0033] Optionally, the optimization factor calculation module is further used for:

[0034] Calculate the outlier degree of current data based on reference data;

[0035] Calculate the first average similarity between the data sequence with the jth data in the reference data as the starting data and the known gesture instruction sequence before obtaining the current data;

[0036] After obtaining the current data, calculate the second average similarity between the data sequence with the jth data in the reference data as the starting data and the known gesture instruction sequence;

[0037] An optimization factor for current data is determined based on the outlier degree and a difference between the first average similarity and the second average similarity.

[0038] Optionally, the similarity calculation module is further used for:

[0039] Calculate the initial similarity between the data sequence starting with the current data and the preset gesture command sequence;

[0040] The initial similarity is processed based on the optimization factor to obtain the similarity between the current data and the preset gesture instruction sequence.

[0041] Optionally, the optimization module is further used for:

[0042] Calculate a first similarity between a data sequence starting with the current data and the most similar gesture instruction data in a preset gesture instruction sequence, and calculate a second similarity between a data sequence starting with the current data and the second most similar gesture instruction data in a preset gesture instruction sequence;

[0043] A sampling frequency optimization factor of the sensor is determined based on the difference between the first degree of similarity and the second degree of similarity.

[0044] The beneficial effects of the present application are different from the prior art. The gesture monitoring method and system based on a smart ring of the present application include: using a sensor to collect sensor data corresponding to a gesture in real time at a first sampling frequency to obtain current data; determining the optimization factor of the current data based on the change between the current data and the reference data before the current data; the reference data before the current data is a preset number of sensor data that is continuous in time with the current data; calculating the similarity between the current data and a preset gesture instruction sequence based on the optimization factor; determining the sampling frequency optimization factor of the sensor based on the similarity, and optimizing the first sampling frequency of the sensor using the sampling frequency optimization factor, so that the sensor collects sensor data corresponding to the gesture in real time at a second sampling frequency. This method can optimize the sampling frequency of sensor data, optimize the accuracy of gesture monitoring of the smart ring, and improve the accuracy of gesture recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1It is a flowchart of an embodiment of a gesture monitoring method based on a smart ring of the present invention;

[0046] Figure 2 It is a structural schematic diagram of an embodiment of a gesture monitoring system based on a smart ring of the present invention. DETAILED DESCRIPTION

[0047] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0048] The method of the present application, when performing gesture command recognition represented by sensor data collected in real time by the smart ring in the cloud computing server, adjusts the subsequent smart ring sampling frequency by matching similarity of gesture commands of real-time data points, optimizes the accuracy of gesture monitoring of the smart ring, and improves the gesture recognition accuracy. The present application is described in detail below with reference to the accompanying drawings and embodiments.

[0049] See also Figure 1 , Figure 1 The figure is a flow chart of an embodiment of a gesture monitoring method based on a smart ring of the present invention, which specifically includes:

[0050] Step S11: using a sensor to collect sensor data corresponding to the gesture in real time at a first sampling frequency to obtain current data.

[0051] The sensor data template of the smart ring gesture command includes the sensor data time series feature values ​​of the smart ring gesture command and the change information of the time series data. The corresponding gesture command is determined by calculating the similarity between the smart ring sensor data collected in real time and the sensor data of each gesture command.

[0052] Specifically, the sensor data corresponding to the gesture is collected in real time using a sensor at a first sampling frequency to obtain current data.

[0053] Step S12: determining an optimization factor of the current data based on a change between the current data and reference data before the current data; the reference data before the current data is a preset number of sensor data that are temporally continuous with the current data.

[0054] After obtaining the real-time sensor data of the smart ring, the similarity calculation with each gesture instruction can be started through the real-time data. However, in the process of similarity calculation, each real-time sensor data point may be the starting data point of a new gesture instruction, or it may be the same gesture connected to the previous sensor data. In addition, when a user uses a gesture instruction, there may be a situation where the user wants to interrupt the gesture and start another one. Therefore, in the process of similarity judgment of the real-time sensor data of the smart ring, it is necessary to judge whether the data point monitored in real time is the starting data point of other gesture instructions. The judgment basis is the optimization factor of the gesture matching similarity obtained by the similarity measurement between the nearest neighbor subsequences.

[0055] After obtaining the optimization factor, the similarity between the real-time data point and the known gesture command can be obtained according to the optimization factor, and the sampling frequency optimization method of the subsequent sensor can be obtained through the difference between the similarities.

[0056] The real-time collected data points of the smart ring sensor are compared with the sensor feature data corresponding to all known gesture commands to obtain the numerical similarity of the real-time newly added data points for each gesture command (for the newly added single data point, only the numerical value can be compared). For the newly added data points, the next round of similarity measurement is performed between the next newly added data point and the previous data point.

[0057] In the process of measuring the similarity of gestures by taking each collected data point as the gesture starting data point, if the sequence formed by the real-time data point and the previous data point has a higher data continuity with the similarity of the existing gesture command, then this data point does not need to be used as a new starting data point for gesture matching. This can ensure that the amount of calculation in the gesture command monitoring process of the smart ring sensor data is lower and the accuracy of gesture recognition is guaranteed. For each new data point p i (i.e., current data), the path change between the data point and the previous data point is used as the optimization factor for the newly added data point, and the data points are screened by the optimization factor, so that the data points without path deviation are not measured as the starting data points.

[0058] In one embodiment, the outlier degree of the current data is calculated based on the reference data. Specifically, the current data p is calculated i COF outlier factor K＝10 (p i ). Among them, the reference data is the current data p i The K data that are consecutive in time with the current data, where K is taken as 10.

[0059] The first average similarity between a data sequence in which the j-th data in the reference data is used as the starting data and a known gesture instruction sequence is calculated before the current data is obtained; the second average similarity between a data sequence in which the j-th data in the reference data is used as the starting data and a known gesture instruction sequence is calculated after the current data is obtained; and the optimization factor of the current data is determined based on the outlier degree and the difference between the first average similarity and the second average similarity.

[0060] In one embodiment, the optimization factor of the current data is calculated using the following formula:

[0061]

[0062] Among them, COF K＝10 (p i ) represents the current data p i The outlier factor obtained through reference data, the number of reference data is K, the value is 10, Norm(COF K＝10 (p i )) represents the outlier factor COF K＝10 (p i ) is the degree of outlier obtained after linear normalization, △ε ji represents the difference between the first average similarity and the second average similarity.

[0063] In one embodiment, the data sequence with the jth data in the reference data as the starting data is recorded as T j , the known gesture command sequence is recorded as T i , then the similarity calculation formula between the data sequence with the jth data in the reference data as the starting data and the known gesture command sequence is:

[0064] ε ji =Norm(DTW(T j ,T i ));

[0065] Among them, DTW(T j ,T i ) represents the data sequence T starting with the jth data j With the known gesture command sequence T i The DTW distance between the two is measured by using the data sequence T j The real-time length is determined by the known gesture command sequence T i The DTW distance is measured for sequences of the same length starting from the beginning.

[0066] The construction of the optimization factor needs to be judged by the deviation caused by the newly added data points and the previous data points. A data point deviates from the previous time series. The more deviations there are, the more likely it is an anomaly. At the same time, when it causes a decrease in the degree of similarity, it can also be said that the data point is more likely to be the starting data point of another gesture that is interrupted by the gesture. Therefore, the optimization factor can be judged by the COF outlier factor and similarity difference between the data point and the previous data point.

[0067] Based on the above formula, the optimization factor of whether the current data is used as a starting data point can be calculated. Compared with taking each new data point as a starting data point for gesture command matching, a large amount of cloud computing server occupancy can be saved while ensuring the accuracy of gesture command recognition.

[0068] Step S13: calculating the similarity between the current data and a preset gesture instruction sequence based on the optimization factor.

[0069] After obtaining the optimization factor of the current data, you can optimize the matching data point in the process of measuring the similarity between the matching data point and the known gesture command data. In the process of calculating the similarity, Screen data points and optimize factors The higher the value, the more likely the data point should be used as a starting data point to start a new gesture command match as an interrupt gesture.

[0070] In one embodiment, the initial similarity between the data sequence starting with the current data and the preset gesture instruction sequence is calculated. Specifically, the initial similarity ε i,c The calculation method is:

[0071]

[0072] in, Indicates a data sequence starting with the current data With the preset gesture command sequence T c , and Norm() represents linear normalization.

[0073] The initial similarity is processed based on the optimization factor to obtain the similarity between the current data and the preset gesture instruction sequence. Specifically:

[0074]

[0075] in, is the optimization factor for the current data, Indicates calculating the data sequence starting with the current data With the preset gesture command sequence T c The distance between them indicates the similarity between them, Norm() indicates linear normalization, ε i,c Represents the similarity between the current data and the preset gesture command sequence, ε′ i,c Indicates the similarity between the current data and the preset gesture command sequence.

[0076] When measuring the similarity of the real-time sensor data of the smart ring, the DTW distance measurement is used to measure the similarity between the real-time sensor data and the feature sequence of the gesture command, that is, The similarity of the first 100 data points of the smart ring is measured in this way. In the subsequent data points, the information formed by the previous data points can be used to determine whether the new data points are used as the starting data points. That is, In the subsequent similarity measurement process, This can achieve the goal of screening new data points, thereby optimizing the amount of calculation in the smart ring gesture matching process.

[0077] Step S14: determining a sampling frequency optimization factor of the sensor based on the similarity, and optimizing the first sampling frequency of the sensor using the sampling frequency optimization factor, so that the sensor collects sensor data corresponding to the gesture in real time at a second sampling frequency.

[0078] After obtaining the similarity between the newly added data points and the characteristic data of each gesture command sensor, the subsequent smart ring sensor sampling frequency can be optimized through the difference information of the similarity between each known gesture command sensor data. In the process of gesture recognition, when the existing known sensor data has similarities with multiple known gesture commands, it means that the existing data points are not good for the recognition of gesture commands. In order to ensure more accurate gesture recognition when the gesture is completed, it is necessary to increase the subsequent sampling frequency of the smart ring to obtain more data points for gesture recognition. When the sensor data points increase, the sensor data can contain more user's finger movement information, thereby improving the accuracy of gesture recognition.

[0079] In a specific embodiment, a first similarity degree between a data sequence starting with the current data and the most similar gesture instruction data in a preset gesture instruction sequence is calculated. Calculate the second similarity between the data sequence starting with the current data and the second similar gesture instruction data in the preset gesture instruction sequence A sampling frequency optimization factor of the sensor is determined based on the difference between the first degree of similarity and the second degree of similarity.

[0080] In a specific embodiment, the similarity between the current data and the preset gesture instruction sequence ε ic , the sampling frequency optimization factor η is calculated based on the difference i The acquisition is as follows:

[0081]

[0082] It should be noted that the preset gesture command sequences are all known gesture command sequences in the database.

[0083] For the sampling frequency optimization factor, when the newly added data point, i.e. the current data, is similar to the feature sequences of each gesture command, it is necessary to increase the sampling frequency of the smart ring in the subsequent smart ring sampling process. As described above, the difference between the most similar feature command and the second most similar feature command is measured to obtain the optimization factor of the sampling frequency.

[0084] Based on the above formula, the sampling frequency of the subsequent smart ring sensor is adjusted by matching the similarity of the real-time data points of the smart ring sensor. Compared with the fixed sampling frequency, when the known data and multiple instructions are similar, the number of subsequent sensor data points can be increased to make the user's gesture recognition more accurate. i Afterwards, the first sampling frequency α=5ms of the smart ring sensor is optimized, and the optimized second sampling frequency is

[0085] After obtaining the real-time sensor data of the smart ring after the optimized sampling frequency, continue to identify the gesture commands through the above-mentioned similarity calculation method. In this process, more accurate data can be obtained through the optimized sampling frequency, thereby obtaining accurate gesture recognition results.

[0086] The gesture monitoring method based on the smart ring of the present application uses the numerical continuity deviation between the real-time newly added data points and the previous data points, and the change in the similarity between the newly added data points and the previous data points in the process of matching the gesture instruction starting data points with the known gesture instructions, to jointly determine whether the newly added data points are used as an optimization factor for a starting data point. Compared with matching each newly added data point as a starting data point for gesture instructions, it can save a lot of cloud computing server occupancy while ensuring the accuracy of gesture instruction recognition. By adjusting the sampling frequency of the subsequent smart ring sensor based on the similarity matching results of the real-time data points of the smart ring sensor, compared with the fixed sampling frequency, when the known data has similarities with multiple instructions, the number of subsequent sensor data points can be increased to make the user's gesture recognition more accurate.

[0087] See also Figure 2 , Figure 2 This is a structural diagram of an embodiment of a gesture monitoring system based on a smart ring of the present invention, which specifically includes: a collection module 21, an optimization factor calculation module 22, a similarity calculation module 23 and an optimization module 24.

[0088] The acquisition module 21 is used to acquire sensor data corresponding to the gesture in real time using a sensor at a first sampling frequency to obtain current data.

[0089] The optimization factor calculation module 22 is used to determine the optimization factor of the current data based on the change between the current data and the reference data before the current data; the reference data before the current data is a preset number of sensor data that are temporally continuous with the current data.

[0090] In one embodiment, the optimization factor calculation module 22 is also used to: calculate the outlier degree of the current data based on the reference data; calculate the first average similarity between the data sequence with the jth data in the reference data as the starting data and the known gesture instruction sequence before obtaining the current data; calculate the second average similarity between the data sequence with the jth data in the reference data as the starting data and the known gesture instruction sequence after obtaining the current data; determine the optimization factor of the current data based on the outlier degree and the difference between the first average similarity and the second average similarity.

[0091] The similarity calculation module 23 is used to calculate the similarity between the current data and a preset gesture instruction sequence based on the optimization factor.

[0092] In one embodiment, the similarity calculation module 23 is also used to: calculate the initial similarity between a data sequence starting with the current data and a preset gesture instruction sequence; process the initial similarity based on the optimization factor to obtain the similarity between the current data and the preset gesture instruction sequence.

[0093] The optimization module 24 is used to determine a sampling frequency optimization factor of the sensor based on the similarity, and optimize the first sampling frequency of the sensor using the sampling frequency optimization factor, so that the sensor collects sensor data corresponding to the gesture in real time at a second sampling frequency.

[0094] In one embodiment, the optimization module 24 is also used to: calculate a first degree of similarity between a data sequence starting with the current data and the most similar gesture instruction data in a preset gesture instruction sequence, calculate a second degree of similarity between a data sequence starting with the current data and the second most similar gesture instruction data in the preset gesture instruction sequence; and determine a sampling frequency optimization factor of the sensor based on the difference between the first degree of similarity and the second degree of similarity.

[0095] The gesture monitoring system based on the smart ring of this embodiment can implement the gesture monitoring method based on the smart ring described above.

[0096] The above are only implementation methods of the present application, and are not intended to limit the patent scope of the present application. Any equivalent structure or equivalent process transformation made using the contents of the present application specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A gesture monitoring method based on a smart ring, characterized in that: include: Using the sensor to collect sensor data corresponding to the gesture in real time at a first sampling frequency to obtain current data; determining an optimization factor for current data based on a change between the current data and reference data before the current data; The reference data before the current data is a preset number of sensor data that are continuous in time with the current data; Calculating the similarity between the current data and a preset gesture instruction sequence based on the optimization factor; Determine a sampling frequency optimization factor of the sensor based on the similarity, and optimize the first sampling frequency of the sensor using the sampling frequency optimization factor, so that the sensor collects sensor data corresponding to the gesture in real time at a second sampling frequency; Determine an optimization factor for the current data based on a change between the current data and reference data before the current data, including: Calculate the outlier degree of current data based on reference data; Calculate the first average similarity between the data sequence with the jth data in the reference data as the starting data and the known gesture instruction sequence before obtaining the current data; After obtaining the current data, calculate the second average similarity between the data sequence with the jth data in the reference data as the starting data and the known gesture instruction sequence; determining an optimization factor for current data based on the outlier degree and a difference between the first average similarity and the second average similarity; The optimization factor of the current data is calculated using the following formula: Among them, COF K＝10 (p i ) represents the current data p i The outlier factor obtained through reference data, the number of reference data is K, the value is 10, Norm(COF K＝10 (p i )) represents the outlier factor COF K＝10 (p i ) is the degree of outlier obtained after linear normalization, △ε ji represents the difference between the first average similarity and the second average similarity.

2. The gesture monitoring method based on the smart ring according to claim 1 is characterized in that: Calculating the similarity between the current data and a preset gesture instruction sequence based on the optimization factor includes: Calculate the initial similarity between the data sequence starting with the current data and the preset gesture command sequence; The initial similarity is processed based on the optimization factor to obtain the similarity between the current data and the preset gesture instruction sequence.

3. The gesture monitoring method based on the smart ring according to claim 2 is characterized in that: The similarity between the current data and the preset gesture command sequence is calculated using the following formula: in, is the optimization factor for the current data, Indicates calculating the data sequence starting with the current data With the preset gesture command sequence T c The distance between them indicates the similarity between them, Norm() indicates linear normalization, ε i,c Indicates the similarity between the current data and the preset gesture command sequence.

4. The gesture monitoring method based on a smart ring according to claim 1, characterized in that: Determining a sampling frequency optimization factor of the sensor based on the similarity includes: Calculate a first similarity between a data sequence starting with the current data and the most similar gesture instruction data in a preset gesture instruction sequence, and calculate a second similarity between a data sequence starting with the current data and the second most similar gesture instruction data in a preset gesture instruction sequence; A sampling frequency optimization factor of the sensor is determined based on the difference between the first degree of similarity and the second degree of similarity.

5. The gesture monitoring system based on the smart ring is characterized by: include: A collection module, used to collect sensor data corresponding to the gesture in real time using a sensor at a first sampling frequency to obtain current data; An optimization factor calculation module, used to determine the optimization factor of current data based on the change between the current data and the reference data before the current data; The reference data before the current data is a preset number of sensor data that are continuous in time with the current data; A similarity calculation module, used for calculating the similarity between the current data and a preset gesture instruction sequence based on the optimization factor; an optimization module, configured to determine a sampling frequency optimization factor of the sensor based on the similarity, and optimize the first sampling frequency of the sensor using the sampling frequency optimization factor, so that the sensor collects sensor data corresponding to the gesture in real time at a second sampling frequency; Determine an optimization factor for the current data based on a change between the current data and reference data before the current data, including: Calculate the outlier degree of current data based on reference data; Calculate the first average similarity between the data sequence with the jth data in the reference data as the starting data and the known gesture instruction sequence before obtaining the current data; After obtaining the current data, calculate the second average similarity between the data sequence with the jth data in the reference data as the starting data and the known gesture instruction sequence; determining an optimization factor for current data based on the outlier degree and a difference between the first average similarity and the second average similarity; The optimization factor of the current data is calculated using the following formula: Among them, COF K＝10 (p i ) represents the current data p i The outlier factor obtained through reference data, the number of reference data is K, the value is 10, Norm(COF K＝10 (p i )) represents the outlier factor COF K＝10 (p i ) is the degree of outlier obtained after linear normalization, △ε ji represents the difference between the first average similarity and the second average similarity.

6. The gesture monitoring system based on the smart ring according to claim 5, characterized in that: The optimization factor calculation module is also used for: Calculate the outlier degree of current data based on reference data; Calculate the first average similarity between the data sequence with the jth data in the reference data as the starting data and the known gesture instruction sequence before obtaining the current data; After obtaining the current data, calculate the second average similarity between the data sequence with the jth data in the reference data as the starting data and the known gesture instruction sequence; An optimization factor for current data is determined based on the outlier degree and a difference between the first average similarity and the second average similarity.

7. The gesture monitoring system based on the smart ring according to claim 6, characterized in that: The similarity calculation module is also used for: Calculate the initial similarity between the data sequence starting with the current data and the preset gesture command sequence; The initial similarity is processed based on the optimization factor to obtain the similarity between the current data and the preset gesture instruction sequence.

8. The gesture monitoring system based on a smart ring according to claim 5, characterized in that: The optimization module is also used to: Calculate a first similarity between a data sequence starting with the current data and the most similar gesture instruction data in a preset gesture instruction sequence, and calculate a second similarity between a data sequence starting with the current data and the second most similar gesture instruction data in a preset gesture instruction sequence; A sampling frequency optimization factor of the sensor is determined based on the difference between the first degree of similarity and the second degree of similarity.

Images