Electric toothbrush oral area recognition method, electric toothbrush, and program product

Abstract

This disclosure provides a method for oral cavity region recognition using an electric toothbrush, an electric toothbrush, and a related software product. The method includes: synchronously acquiring sensor data from a master and slave IMU; determining the linear acceleration of the master and slave IMUs to obtain the linear acceleration correlation, and calculating the difference between the two angular velocity data to obtain the angular velocity difference; determining a dynamic reliability factor based on both; extracting relative attitude deviation features, axial pulse features, horizontal oscillation main direction features, and time-frequency features of the master and slave IMUs respectively; weighting the relative attitude deviation features and axial pulse features using the dynamic reliability factor, and fusing them with the time-frequency features and horizontal oscillation main direction features to obtain fused features; constructing a fused feature sequence based on the fused features within a third preset time period, and performing oral cavity region recognition accordingly, outputting the recognition result. This method is accurate, has strong anti-interference capabilities, and can achieve precise recognition of oral cavity regions.

Description

Technical Field

[0001] This disclosure relates to the field of electric toothbrush control, specifically to an electric toothbrush oral cavity area recognition method, an electric toothbrush, and a program product. Background Technology

[0002] With rising consumer demand and increased health awareness, the intelligentization of electric toothbrushes has become an inevitable trend. One of the core functions of an electric toothbrush is to automatically identify and record the oral cavity areas cleaned by the brush head, thereby assessing the comprehensiveness and effectiveness of brushing and guiding users to improve their cleaning habits. Currently, most smart toothbrushes on the market use a single inertial measurement unit (IMU) solution. By analyzing the patterns corresponding to the acceleration and angular velocity of the handle, pattern matching or machine learning algorithms are used to infer the approximate area cleaned by the brush head.

[0003] However, this single-point motion sensing-based approach has fundamental flaws, resulting in recognition performance falling far short of the required accuracy, especially when distinguishing different sides of teeth, where it almost completely fails. The primary problem with single-point motion sensing-based approaches lies in the ambiguity and low discriminative power of signal features. The oral cavity is a small and complex three-dimensional space, and the brushing motion in different areas (especially the maxilla and mandible, and the front and back teeth) shows very subtle differences in single IMU readings. These subtle differences are easily masked by the randomness of the user's grip, habitual wrist tremors, and high-frequency noise generated by the toothbrush colliding with soft tissue, making the classification model prone to confusion and resulting in unstable and unreliable recognition results.

[0004] A deeper deficiency lies in the fact that single-IMU solutions lose the pose information of the toothbrush as a rigid tool with a defined length and direction in three-dimensional space. For example, when cleaning the upper posterior teeth, the entire toothbrush needs to tilt backward and upward; while when cleaning the lower anterior teeth, the toothbrush tilts forward and downward. More importantly, there are significant differences in internal and external rotation of the wrist when cleaning the outer (labial / buccal) and inner (lingual) sides of the teeth. These spatial pose changes, rich in positional information, cannot be directly and robustly parsed from the composite motion data of a single point on the handle. Existing technologies attempt to compensate for these deficiencies by increasing sensor dimensions or employing more complex deep learning models, but these often fall into "black box" operation, incurring high computational costs, and failing to fundamentally address the problem of poor model generalization ability caused by individual physiological and behavioral differences. Therefore, developing a method that can directly and effectively perceive the spatial pose of the toothbrush during brushing, and on this basis achieve high-precision and robust region recognition, especially the ability to distinguish between inner and outer sides, remains a long-standing technical challenge in this field. Summary of the Invention

[0005] This disclosure provides a method for recognizing oral cavity areas with an electric toothbrush, an electric toothbrush, and a related product.

[0006] According to one aspect of this disclosure, a method for oral cavity region recognition using an electric toothbrush is provided, comprising: synchronously acquiring sensor data from a main IMU and a slave IMU mounted along the long axis of the electric toothbrush handle during operation, wherein the sensor data includes acceleration data and angular velocity data; determining a linear acceleration of the main IMU as a first linear acceleration based on the acceleration data of the main IMU, and determining a linear acceleration of the slave IMU as a second linear acceleration based on the acceleration data of the slave IMU; calculating the correlation between the first linear acceleration and the second linear acceleration to obtain a linear acceleration correlation; calculating the difference between the angular velocity data of the main IMU and the slave IMU to obtain an angular velocity difference; and based on... The linear acceleration correlation and the angular velocity difference are used to determine the dynamic reliability factor; feature extraction is performed based on sensor data from the master IMU and slave IMU to obtain the relative attitude deviation feature, axial pulse feature, and horizontal swing main direction feature between the master IMU and slave IMU; time-frequency features are extracted based on sensor data from the master IMU; the relative attitude deviation feature and axial pulse feature are weighted by the dynamic reliability factor, and the weighted result is fused with the time-frequency feature and the horizontal swing main direction feature to obtain fused features; and a fused feature sequence is constructed based on the fused features within a third preset time period, and oral cavity region recognition is performed based on the fused feature sequence to obtain oral cavity region recognition results.

[0007] According to at least one embodiment of the electric toothbrush oral cavity region recognition method of the present disclosure, determining the linear acceleration of the main IMU as a first linear acceleration based on the acceleration data of the main IMU includes: performing low-pass filtering on the acceleration data of the main IMU to obtain first filtered acceleration data; determining the gravity component of the gravity vector in the main IMU coordinate system based on the real-time attitude of the main IMU and a first rotation matrix for aligning the main IMU coordinate system to the world coordinate system; and subtracting the gravity component from the first filtered acceleration data to obtain the first linear acceleration.

[0008] The electric toothbrush oral cavity region recognition method according to at least one embodiment of the present disclosure further includes: determining a first rotation matrix, comprising: in a static state, acquiring acceleration data of the main IMU within a first preset time period and calculating the average value as a first average acceleration value; and calculating the first rotation matrix based on the first average acceleration value and the gravity vector using the following formula: ,in, Represents the gravity vector; Represents the first rotation matrix; This represents the average value of the first acceleration; This indicates the modulo operation.

[0009] According to at least one embodiment of the electric toothbrush oral cavity region recognition method of this disclosure, the correlation between a first linear acceleration and a second linear acceleration is calculated to obtain the linear acceleration correlation, including: constructing a first linear acceleration model sequence based on the model of the first linear acceleration within a second preset time period; constructing a second linear acceleration model sequence based on the model of the second linear acceleration within a second preset time period; and calculating the Pearson correlation coefficient between the first linear acceleration model sequence and the second linear acceleration model sequence as the linear acceleration correlation.

[0010] According to at least one embodiment of the electric toothbrush oral region recognition method of this disclosure, the dynamic reliability factor is calculated by the following formula:

[0011] in, This represents the dynamic reliability factor at the k-th sampling time. and All of these represent positive adjustment parameters; This represents the linear acceleration correlation at the k-th sampling time. This represents the difference in angular velocity at the k-th sampling time.

[0012] According to at least one embodiment of the electric toothbrush oral cavity region recognition method of this disclosure, the extraction process of the relative posture deviation feature includes: multiplying the real-time posture of the slave IMU by the inverse of the real-time posture of the master IMU to obtain a relative rotation posture, wherein the relative rotation posture represents the real-time dynamic rotation relationship of the slave IMU relative to the master IMU; converting the relative rotation posture into a relative rotation matrix; transposing a second rotation matrix used to align the master IMU coordinate system to the slave IMU coordinate system and multiplying it with the relative rotation matrix to obtain a relative posture deviation matrix; converting the relative posture deviation matrix into an axis-angle representation; and constructing a relative posture deviation feature vector characterizing the spatial tilt orientation and tilt degree of the toothbrush based on the rotation angle and unit rotation axis in the axis-angle representation, as the relative posture deviation feature.

[0013] The electric toothbrush oral cavity region recognition method according to at least one embodiment of the present disclosure further includes: determining a second rotation matrix, comprising: in a static state, acquiring acceleration data of a primary IMU within a first preset time period and calculating an average value as a first acceleration average value; acquiring acceleration data of a secondary IMU within the first preset time period and calculating an average value as a second acceleration average value; and calculating the second rotation matrix based on the first acceleration average value and the second acceleration average value using the following formula: ,in, This represents the average value of the second acceleration; This represents the average value of the first acceleration; This represents the second rotation matrix.

[0014] According to at least one embodiment of the electric toothbrush oral cavity region recognition method of this disclosure, the extraction process of the axial pulse feature includes: projecting the first linear acceleration onto the major axis direction of the electric toothbrush in the world coordinate system to obtain the axial pulse feature.

[0015] According to at least one embodiment of the electric toothbrush oral cavity region recognition method of this disclosure, the extraction process of the horizontal swing principal direction feature includes: projecting the first linear acceleration onto the horizontal plane of the world coordinate system to obtain a two-dimensional horizontal motion vector; and calculating the principal direction angle of the two-dimensional horizontal motion vector through a bivariate arctangent function as the horizontal swing principal direction feature.

[0016] According to another aspect of this disclosure, an electric toothbrush is provided, comprising: a memory storing execution instructions; and a processor executing the execution instructions stored in the memory, causing the processor to perform an electric toothbrush oral cavity region recognition method according to any embodiment of this disclosure.

[0017] According to another aspect of this disclosure, a readable storage medium is provided, wherein executable instructions are stored therein, which, when executed by a processor, are used to implement the electric toothbrush oral cavity region recognition method of any embodiment of this disclosure.

[0018] According to another aspect of this disclosure, a computer program product is provided, including a computer program that, when executed by a processor, implements the electric toothbrush oral cavity region recognition method of any embodiment of this disclosure.

[0019] This disclosure has the following beneficial effects: (1) Innovatively achieve accurate identification of the inner and outer sides of teeth: Through the spatial posture features and axial motion features calculated by the dual IMU array, the unique wrist rotation and toothbrush tilt posture when cleaning the tongue side can be effectively captured. In the pure motion sensor solution, the "tongue side area" can be reliably identified as an independent area, solving the long-standing technical blind spot.

[0020] (2) Significant improvement in recognition accuracy and robustness: This disclosure upgrades the recognition logic from "single-point motion analysis" to "spatial geometric analysis", and the extracted features have clear physical meaning and strong discriminative power. The introduction of dynamic credibility factor enables the system to have intelligent anti-interference capability, which can automatically identify and suppress noise data caused by unconventional wrist movements or collisions, ensuring the stable performance of the system in real and complex usage scenarios.

[0021] (3) Excellent personalization adaptability: The initial calibration step can incorporate the user's unique grip posture and oral geometry differences into the system benchmark, so that the core algorithm is built on the user's personalized coordinate system. The dynamic weight mechanism further adapts to the user's varying brushing force and frequency, realizing a user-friendly "ready to use" experience without the need for users to perform cumbersome training operations.

[0022] (4) The algorithm is efficient and easy to integrate into products: the overall algorithm process is clear and the computational complexity is controllable; the feature dimensions are carefully designed to ensure the amount of feature information and avoid data redundancy. With the lightweight network model, it has low requirements for the computing power and memory of the embedded MCU, making it very suitable for integration and mass production in cost-sensitive consumer electronics products. Attached Figure Description

[0023] The accompanying drawings illustrate exemplary embodiments of the present disclosure and, together with the description thereof, serve to explain the principles of the present disclosure. These drawings are included to provide a further understanding of the present disclosure and are incorporated in and constitute a part of this specification.

[0024] Figure 1 This is a flowchart illustrating an electric toothbrush oral cavity region recognition method according to one embodiment of the present disclosure.

[0025] Figure 2 This is a schematic diagram of the installation method of the master IMU and slave IMU according to one embodiment of the present disclosure.

[0026] Figure 3 This is a flowchart illustrating the method corresponding to step S120 of one embodiment of the present disclosure.

[0027] Figure 4 This is a flowchart illustrating a first rotation matrix determination method according to one embodiment of the present disclosure.

[0028] Figure 5 This is a flowchart illustrating the method corresponding to step S130 of one embodiment of the present disclosure.

[0029] Figure 6 This is a flowchart illustrating a relative posture deviation feature extraction method according to one embodiment of the present disclosure.

[0030] Figure 7 This is a flowchart illustrating a second rotation matrix determination method according to one embodiment of the present disclosure.

[0031] Figure 8 This is a flowchart illustrating a method for extracting the main direction feature of horizontal swing according to one embodiment of the present disclosure.

[0032] Figure 9This is a schematic structural block diagram of an electric toothbrush using a processor-based hardware implementation according to one embodiment of the present disclosure. Detailed Implementation

[0033] The present disclosure will now be described in further detail with reference to the accompanying drawings and examples. It should be understood that the specific examples described herein are for illustrative purposes only and are not intended to limit the scope of the disclosure. Furthermore, it should be noted that, for ease of description, only the parts relevant to the present disclosure are shown in the accompanying drawings.

[0034] It should be noted that, where there is no conflict, the embodiments and features described in this disclosure can be combined with each other. The technical solutions of this disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0035] To facilitate description and make the technical solutions of this disclosure easier to understand, the terminology of this disclosure will be explained before describing the technical solutions of this disclosure.

[0036] An IMU (Inertial Measurement Unit) is a sensor device that can measure the three-axis acceleration and three-axis angular velocity of an object, and then calculate the object's spatial attitude and motion state.

[0037] The sensitive axis direction of an IMU refers to the orientation of the three orthogonal coordinate axes (usually X, Y, and Z axes) inside the IMU used to sense acceleration and angular velocity. It is the core reference for the IMU to accurately collect motion data in the corresponding direction.

[0038] The coordinate system of the electric toothbrush body refers to the local coordinate system that moves synchronously with the movement of the electric toothbrush. It is usually defined by the major axis of the electric toothbrush handle as the X-axis, the Y-axis as the axis perpendicular to the major axis and parallel to the end face of the brush head, and the Z-axis as the axis perpendicular to the end face of the brush head.

[0039] Figure 1 A schematic flowchart illustrating the overall process of an electric toothbrush oral cavity region recognition method according to one embodiment of this disclosure is shown. Figure 1 The method shown includes steps S110 to S140. This method can be performed by an electric toothbrush.

[0040] In step S110, sensor data from the main IMU and slave IMU mounted along the long axis of the electric toothbrush handle are collected synchronously during the operation of the electric toothbrush.

[0041] As one possible implementation, the master IMU and slave IMU can be two IMUs rigidly mounted along the long axis of the electric toothbrush handle (e.g., from the tail of the handle towards the brush head). The IMU closer to the brush head is designated as the master IMU, and the IMU farther from the brush head (closer to the tail of the handle) is designated as the slave IMU. The physical distance between the master and slave IMUs is fixed (exemplarily, it can be set to 60 mm). The master and slave IMUs can be mounted as follows: Figure 2 As shown. It should be noted that during installation, the sensitive axes of the master IMU and slave IMU should be aligned with the X, Y, and Z axes of the coordinate system of the electric toothbrush body as much as possible, so that the collected acceleration and angular velocity data can directly reflect the motion state of the electric toothbrush itself.

[0042] As one possible implementation, during synchronous acquisition, the data acquisition operations of the master IMU and slave IMU can be triggered synchronously at a fixed sampling rate (e.g., 200 Hz) so that time-aligned sensor data can be obtained at each sampling moment.

[0043] The sensor data includes acceleration and angular velocity data, and both acceleration and angular velocity data contain data along three coordinate axes. These three coordinate axes correspond to the three sensing axes of the primary IMU and the secondary IMU, respectively.

[0044] In step S120, the linear acceleration of the main IMU is determined based on the acceleration data of the main IMU as the first linear acceleration, and the linear acceleration of the slave IMU is determined based on the acceleration data of the slave IMU as the second linear acceleration.

[0045] Regarding step S120, in some embodiments of this disclosure, it may include, for example... Figure 3 Steps S1201 to S1203 are shown.

[0046] In step S1201, the acceleration data of the main IMU is low-pass filtered to obtain the first filtered acceleration data.

[0047] Low-pass filtering can filter out high-frequency interference signals (such as vibration, environmental noise, etc.). For example, the cutoff frequency of the low-pass filter can be set to 5 Hz.

[0048] In step S1202, the gravity component of the gravity vector in the main IMU coordinate system is determined based on the real-time attitude of the main IMU and the first rotation matrix used to align the main IMU coordinate system to the world coordinate system.

[0049] Real-time attitude acquisition is a function inherent to the IMU chip itself (obtained by integrating gyroscope angular velocity data) and can be automatically estimated by the main IMU chip.

[0050] Based on the real-time attitude of the main IMU, the gravity vector (with a fixed direction) in the world coordinate system can be transformed to the main IMU coordinate system by combining the first rotation matrix and performing coordinate transformation calculations, thereby determining the gravity component of the gravity vector in the main IMU coordinate system.

[0051] In step S1203, the gravity component is subtracted from the first filtered acceleration data to obtain the first linear acceleration.

[0052] Compared to the acceleration data from the main IMU, the first linear acceleration can more purely reflect the motion of the brush head.

[0053] Based on the acceleration data from the IMU, the linear acceleration from the IMU is determined, which is used as the second linear acceleration method. Figure 3 The implementation methods shown are similar and will not be described in detail here.

[0054] As one possible implementation, before determining the gravity component of the gravity vector in the main IMU coordinate system based on the real-time attitude of the main IMU and a first rotation matrix used to align the main IMU coordinate system to the world coordinate system, the method further includes: determining the first rotation matrix. Figure 4 A flowchart illustrating a first rotation matrix determination method according to an embodiment of this disclosure is shown. Figure 4 The method shown includes steps S210 to S220.

[0055] In step S210, under static conditions, the acceleration data of the main IMU within a first preset time period is collected and the average value is calculated as the first average acceleration value.

[0056] The first preset duration can be set according to the actual application and is not limited here. For example, the first preset duration is 2 seconds.

[0057] As one possible implementation, before calculating the average value of the acceleration data from the main IMU, median filtering is performed on the acquired acceleration data from the main IMU to remove outliers, impulse noise, etc., and to avoid outliers affecting the accuracy of subsequent average value calculation.

[0058] Considering the projection of the average first acceleration value onto the gravity vector in the main IMU coordinate system under a stationary state, in step S220, the first rotation matrix is ​​calculated based on the average first acceleration value and the gravity vector using the following formula:

[0059] in, Represents the gravity vector; Represents the first rotation matrix; This represents the average value of the first acceleration; This represents the modulo operation. For example, .

[0060] In step S130, the correlation between the first linear acceleration and the second linear acceleration is calculated to obtain the linear acceleration correlation.

[0061] Regarding step S130, in some embodiments of this disclosure, it may include, for example... Figure 5 Steps S1301 to S1302 are shown.

[0062] In step S1301, a first linear acceleration mode sequence is constructed based on the mode of the first linear acceleration within a second preset time period, and a second linear acceleration mode sequence is constructed based on the mode of the second linear acceleration within a second preset time period.

[0063] As one possible implementation, within the second preset duration, the first linear acceleration corresponds to a fixed time interval, such as 0.4 seconds.

[0064] In step S1302, the Pearson correlation coefficient between the first linear acceleration mode sequence and the second linear acceleration mode sequence is calculated as the linear acceleration correlation degree.

[0065] As one possible implementation, the range of values ​​for the linear acceleration correlation is: The closer the value is to 1, the more synchronized the motion patterns sensed by the master IMU and the slave IMU are on the waveform.

[0066] In step S140, the difference between the angular velocity data of the master IMU and the slave IMU is calculated to obtain the angular velocity difference.

[0067] As one possible implementation, the difference between the angular velocity data of the master IMU and the slave IMU is calculated to obtain the angular velocity difference. This includes calculating the difference between the angular velocity data of the master IMU and the slave IMU. The L2 norm of the difference is used as the angular velocity difference. The angular velocity difference reflects the difference in rotational motion sensed by the master IMU and the slave IMU. For ideal rigid body motion, the angular velocity difference should be close to 0 (zero).

[0068] In step S150, the dynamic reliability factor is determined based on the linear acceleration correlation and angular velocity difference.

[0069] As one possible implementation method, the dynamic credibility factor is calculated using the following formula:

[0070] in, This represents the dynamic reliability factor at the k-th sampling time. When the master IMU and slave IMU are highly synchronized in motion and rotate in unison, A value close to 1 indicates that the data from the IMU has high reliability; when the motion disturbances are asynchronous, A value close to 0 indicates that the data from the IMU has low reliability, and reliance on information from the IMU can be reduced. and All of these represent positive adjustment parameters, for example, . This represents the linear acceleration correlation at the k-th sampling time. This represents the difference in angular velocity at the k-th sampling time.

[0071] In step S160, feature extraction is performed based on the sensor data of the master IMU and the slave IMU to obtain the relative attitude deviation features, axial pulse features, and horizontal swing main direction features between the master IMU and the slave IMU.

[0072] Figure 6 A flowchart illustrating a relative attitude deviation feature extraction method according to one embodiment of this disclosure is shown. Figure 6 The method shown includes steps S310 to S350.

[0073] In step S310, the relative rotation attitude is obtained by inversely multiplying the real-time attitude of the slave IMU with the real-time attitude of the master IMU. The relative rotation attitude represents the real-time dynamic rotation relationship of the slave IMU relative to the master IMU.

[0074] As one possible implementation, real-time attitude can be represented as attitude quaternions. Attitude quaternions are a common way for IMUs to represent object attitude (avoiding gimbal lock issues compared to Euler angles). Currently, mainstream IMU chips (such as common six-axis and nine-axis IMUs) all support directly outputting attitude quaternions through their own data format conversion.

[0075] For example, the formula for calculating the relative rotational attitude can be expressed as:

[0076] in, This represents the relative rotational attitude at the k-th sampling time. This represents the real-time attitude of the IMU at the k-th sampling time. This represents the inverse of the real-time attitude of the primary IMU at the k-th sampling time; This is represented as quaternion multiplication.

[0077] In step S320, the relative rotation attitude is converted into a relative rotation matrix.

[0078] As one possible implementation, in the process of converting relative rotation attitude into relative rotation matrix, the quaternion can be converted into the corresponding 3×3 rotation matrix as the relative rotation matrix using the conversion formula between quaternions and rotation matrices.

[0079] In step S330, the second rotation matrix used to align the master IMU coordinate system to the slave IMU coordinate system is transposed and multiplied with the relative rotation matrix to obtain the relative attitude deviation matrix.

[0080] For example, the formula for calculating the relative attitude deviation matrix can be expressed as:

[0081] in, This represents the relative attitude deviation matrix at the k-th sampling time. This represents the relative rotation matrix at the k-th sampling time; the superscript T indicates transpose. This represents the second rotation matrix.

[0082] In step S340, the relative attitude deviation matrix is ​​converted into an axis-angle representation.

[0083] As one possible implementation, the axis angle representation includes a rotation angle and a unit rotation axis.

[0084] In step S350, a relative attitude deviation feature vector is constructed based on the rotation angle and unit rotation axis in the axis angle representation to characterize the spatial tilt orientation and tilt degree of the toothbrush, serving as the relative attitude deviation feature.

[0085] For example, the relative attitude deviation feature vector can be represented as ,in, Indicates the rotation angle. Indicates the unit axis of rotation.

[0086] The unit rotation axis in the relative posture deviation feature vector can characterize the orientation of the electric toothbrush tilt (including front / back / left / right), and the rotation angle can characterize the degree of tilt. The relative posture deviation feature is the core of distinguishing between upper / lower jaw and front / back teeth.

[0087] As one possible implementation, before transposing the second rotation matrix used to align the primary IMU coordinate system to the secondary IMU coordinate system, the method further includes: determining the second rotation matrix. Figure 7 A flowchart illustrating a second rotation matrix determination method according to one embodiment of this disclosure is shown. Figure 7 The method shown includes steps S410 to S420.

[0088] In step S410, under static conditions, the acceleration data of the main IMU within a first preset time period is collected and the average value is calculated as the first average acceleration value. The acceleration data of the slave IMU within the first preset time period is collected and the average value is calculated as the second average acceleration value.

[0089] The calculation process for the second average acceleration is similar to step S210, and will not be repeated here.

[0090] In step S420, the second rotation matrix is ​​calculated based on the first average acceleration and the second average acceleration using the following formula:

[0091] in, This represents the average value of the second acceleration; This represents the average value of the first acceleration; This represents the second rotation matrix.

[0092] As one possible implementation, the static state can be achieved by the user holding the electric toothbrush in their habitual grip position in front of their mouth. Based on this, the second rotation matrix encapsulates the fixed relative direction between the master IMU and the slave IMU under the user's habitual grip position, serving as a "zero-position" reference for detecting dynamic posture changes.

[0093] Understandably, the first rotation matrix and the second rotation matrix can be determined (calibrated) and stored in the electric toothbrush when the user first uses it, and do not need to be determined again for subsequent uses.

[0094] As one possible implementation, the axial pulse feature extraction process includes: projecting the first linear acceleration onto the major axis direction of the electric toothbrush in the world coordinate system to obtain the axial pulse feature. For example, the formula for calculating the axial pulse feature can be expressed as:

[0095] in, This represents the axial pulse characteristics at the k-th sampling time. This represents the first linear acceleration at the k-th sampling time. This indicates the direction of the major axis of the electric toothbrush in the world coordinate system at the k-th sampling time.

[0096] Axial pulse characteristics are sensitive to the "push-pull" motion when cleaning the occlusal surface or inner side of posterior teeth, which helps to distinguish between the anterior and posterior tooth areas and assists in judging lingual movements.

[0097] Figure 8 A flowchart illustrating a method for extracting the principal direction feature of horizontal oscillation according to an embodiment of this disclosure is shown. Figure 8The method shown includes steps S510 to S520.

[0098] In step S510, the first linear acceleration is projected onto the horizontal plane of the world coordinate system to obtain a two-dimensional horizontal motion vector.

[0099] In step S520, the principal direction angle of the two-dimensional horizontal motion vector is calculated using a bivariate arctangent function, and this angle is used as the principal direction feature of the horizontal swing. For example, the formula for calculating the principal direction feature of the horizontal swing can be expressed as:

[0100] in, This represents the principal direction of the horizontal oscillation at the k-th sampling time. Represents the bivariate arctangent function; and These represent the values ​​of the two-dimensional horizontal motion vector in the X-axis and Y-axis directions at the k-th sampling time, respectively.

[0101] The main direction of horizontal oscillation can be used to clearly distinguish the left and right oral cavity regions.

[0102] In step S170, time-frequency features are extracted based on the sensor data from the main IMU.

[0103] As one possible implementation, the extraction of time-frequency features can be achieved using existing short-time signal processing algorithms. Specifically, within a second preset time period, the sensor data of the main IMU (e.g., the magnitude of the first linear acceleration or the magnitude of the angular velocity) can be framed and windowed, using a sliding window of fixed length and fixed step size to extract continuous signal segments. For each frame, a short-time energy calculation algorithm and a zero-crossing rate statistical algorithm are used for feature extraction. The short-time energy can be obtained by calculating and normalizing the sum of squares of the amplitudes at each sampling point within the frame, and is used to characterize the impact intensity and energy distribution of the brushing motion. The zero-crossing rate can be obtained by counting the number of times adjacent sampling points cross the zero level or the sign change of the frame mean within the frame, and is used to characterize the vibration frequency and rhythmic characteristics of the brushing motion.

[0104] In step S180, the relative attitude deviation feature and axial pulse feature are weighted by the dynamic confidence factor, and the weighted result is fused with the time-frequency feature and the horizontal swing main direction feature to obtain the fused feature.

[0105] For example, the formula for calculating the fused features can be expressed as:

[0106] in, This represents the fusion feature at the k-th sampling time. This represents the time-frequency characteristics at the k-th sampling time. This represents the dynamic reliability factor at the k-th sampling time. This represents the relative attitude deviation characteristics at the k-th sampling time. This represents the axial pulse characteristics at the k-th sampling time. This represents the principal direction characteristic of the horizontal oscillation at the k-th sampling time.

[0107] In the calculation of the above fusion features, the relative attitude deviation feature and the axial pulse feature are modulated by a dynamic reliability factor. This is only true when the data is reliable. High values ​​indicate that these strong spatial discriminative features (relative attitude deviation features and axial pulse features) are only fully utilized. When the data is unreliable ( If the value is low, it will automatically rely on more robust basic features (time-frequency features) and horizontal oscillation main direction features, thereby improving accuracy while ensuring stability in extreme cases.

[0108] In step S190, a fusion feature sequence is constructed based on the fusion features within a third preset time period. Oral region identification is then performed based on the fusion feature sequence to obtain the oral region identification result.

[0109] The third preset duration can be set by those skilled in the art according to specific applications, and is not limited here.

[0110] As one possible implementation, oral cavity region recognition can employ a pre-trained lightweight temporal classification model. In one example, the network structure of the lightweight temporal classification model can be built based on an existing cascaded architecture of a one-dimensional causal convolutional network and gated linear units (GLUs), using a sliding window of length T (T=1.5s) as input to capture temporal patterns from local to global. The output layer of the lightweight temporal classification model can be set to 9 neurons, each corresponding to a logical value of one of the 9 oral cavity regions. After model computation, the original region sequence (i.e., the oral cavity region recognition result sequence corresponding to each sampling time) is output. The logical values ​​of the 9 oral cavity regions can be converted into a probability distribution using the Softmax function (normalized exponential function). For example, the 9 regions may include: R1: left maxillary anterior teeth, R2: right maxillary anterior teeth, R3: left maxillary posterior teeth, R4: right maxillary posterior teeth, R5: left mandibular anterior teeth, R6: right mandibular anterior teeth, R7: left mandibular posterior teeth, R8: right mandibular posterior teeth, and R9: lingual region.

[0111] As a further implementation, to eliminate the impact of transient jitter on the region identification results and improve the stability and accuracy of the identification results, post-processing operations can be performed on the original region sequence output by the lightweight temporal classification model. Specifically, majority voting or dwell time filtering can be used to optimize the original region sequence.

[0112] This disclosure has the following beneficial effects: (1) Innovatively achieve accurate identification of the inner and outer sides of teeth: Through the spatial posture features and axial motion features calculated by the dual IMU array, the unique wrist rotation and toothbrush tilt posture when cleaning the tongue side can be effectively captured. In the pure motion sensor solution, the "tongue side area" can be reliably identified as an independent area, solving the long-standing technical blind spot.

[0113] (2) Significant improvement in recognition accuracy and robustness: This disclosure upgrades the recognition logic from "single-point motion analysis" to "spatial geometric analysis", and the extracted features have clear physical meaning and strong discriminative power. The introduction of dynamic credibility factor enables the system to have intelligent anti-interference capability, which can automatically identify and suppress noise data caused by unconventional wrist movements or collisions, ensuring the stable performance of the system in real and complex usage scenarios.

[0114] (3) Excellent personalization adaptability: The initial calibration step can incorporate the user's unique grip posture and oral geometry differences into the system benchmark, so that the core algorithm is built on the user's personalized coordinate system. The dynamic weight mechanism further adapts to the user's varying brushing force and frequency, realizing a user-friendly "ready to use" experience without the need for users to perform cumbersome training operations.

[0115] (4) The algorithm is efficient and easy to integrate into products: the overall algorithm process is clear and the computational complexity is controllable; the feature dimensions are carefully designed to ensure the amount of feature information and avoid data redundancy. With the lightweight network model, it has low requirements for the computing power and memory of the embedded MCU, making it very suitable for integration and mass production in cost-sensitive consumer electronics products.

[0116] The following is a specific example illustrating the oral cavity region recognition method of the electric toothbrush disclosed herein.

[0117] 1. When the user uses the electric toothbrush for the first time, the initial calibration is performed to determine the first rotation matrix and the second rotation matrix.

[0118] 2. After the user begins their daily brushing routine, they first brush the outer surface of the upper right posterior teeth. At this time, the electric toothbrush tilts to the right and rear, and the wrist makes regular left-right swinging motions. The sensor data acquired by the main IMU and the secondary IMU are highly synchronized, resulting in a dynamic reliability factor of 0.95. The extracted relative posture deviation feature is a vector pointing to the right and rear, with a relatively large magnitude; the axial pulse feature presents a medium-intensity periodic pulse; the horizontal swing direction feature indicates the swing direction is to the right. After feature fusion, the oral cavity region identification result obtained by the classification model is "R4: Right maxillary posterior teeth region".

[0119] 3. Next, the user flips the brush head to the inside (lingual) side of the same area for cleaning. At this time, there is a noticeable inward rotation of the wrist, and the tilt angle of the toothbrush changes significantly. Therefore, the vector direction corresponding to the relative posture deviation feature may shift more towards the inside and downward; the axial pulse feature may weaken or change pattern. Since the action remains stable, the dynamic confidence factor remains around 0.85. Based on these changed fusion features, the classification model updates the recognition result to "R9: Lingual Area".

[0120] 4. When a user quickly moves the toothbrush from right to left, there is a period of irregular, aimless waving motion. The sensor data collected by the main IMU and the slave IMU lose synchronization for a short time, causing the dynamic reliability factor to plummet to 0.2. At this point, the region probability distribution output by the classification model becomes smoother, or the recognition result of the previous region can be maintained through post-processing rules, effectively avoiding meaningless region jumps and misjudgments during the movement.

[0121] According to a further embodiment of this disclosure, an electric toothbrush is also provided. Figure 9This diagram illustrates a schematic block diagram of an electric toothbrush using a processor-based hardware implementation according to an embodiment of the present disclosure. The hardware structure of the electric toothbrush of this disclosure can be implemented using a bus architecture. The bus architecture can include any number of interconnect buses and bridges, depending on the specific application and overall design constraints of the hardware. Bus 1100 connects various circuits including one or more processors 1200, memory 1300, and / or hardware modules. Bus 1100 can also connect various other circuits 1400 such as peripheral devices, voltage regulators, power management circuits, external antennas, etc. Bus 1100 can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Component (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, only one connecting line is used in this figure, but this does not indicate that there is only one bus or one type of bus. The memory 1300 stores a computer program. When the processor 1200 executes the computer program, the processor 1200 can perform the following processes: Synchronously acquire sensor data from the main IMU and slave IMU mounted along the long axis of the electric toothbrush handle during operation. The sensor data includes acceleration data and angular velocity data. Based on the acceleration data of the main IMU, determine the linear acceleration of the main IMU as the first linear acceleration. Based on the acceleration data of the slave IMU, determine the linear acceleration of the slave IMU as the second linear acceleration. Calculate the correlation between the first linear acceleration and the second linear acceleration to obtain the linear acceleration correlation. Calculate the difference between the angular velocity data of the main IMU and the slave IMU to obtain the angular velocity difference. Determine the dynamic reliability factor based on the linear acceleration correlation and the angular velocity difference. Perform feature extraction based on the sensor data of the main IMU and the slave IMU to obtain the relative attitude deviation features, axial pulse features, and horizontal oscillation main direction features between the main IMU and the slave IMU. Extract time-frequency features based on the sensor data of the main IMU. The relative posture deviation feature and axial pulse feature are weighted by the dynamic confidence factor, and the weighted result is fused with the time-frequency feature and the horizontal oscillation main direction feature to obtain the fused feature. A fused feature sequence is constructed based on the fused feature within a third preset time period, and oral cavity region recognition is performed based on the fused feature sequence to obtain the oral cavity region recognition result.

[0122] This disclosure also provides a readable storage medium storing a computer program that, when executed by a processor, is used to implement the methods described above. A "readable storage medium" can be any means capable of containing, storing, communicating, propagating, or transmitting a program for use by or in conjunction with an instruction execution system, apparatus, or device. More specific examples of a readable storage medium include: an electrical connection with one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and programmable read-only memory (EPROM or flash memory), fiber optic devices, and portable read-only memory (CDROM), etc.

[0123] This disclosure also provides a computer program product, the methods of which can be implemented wholly or partially through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented wholly or partially as a computer program product. The computer program product includes one or more computer programs or instructions. When the computer program or instructions are loaded and executed, all or part of the processes or functions of this disclosure are performed.

[0124] Computer programs or instructions can be stored in a readable storage medium or transferred from one readable storage medium to another. For example, the computer program or instructions can be transferred from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means. The readable storage medium can be any available medium capable of access, or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium, such as a floppy disk, hard disk, or magnetic tape; an optical medium, such as a digital video optical disc; or a semiconductor medium, such as a solid-state drive. The computer-readable storage medium can be a volatile or non-volatile storage medium, or it can include both volatile and non-volatile types of storage media.

[0125] Those skilled in the art will understand that embodiments of this disclosure can be provided as methods, systems, or computer program products. Therefore, this disclosure can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this disclosure can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0126] This disclosure is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus, and computer program products according to this disclosure. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0127] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0128] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0129] In the description of this specification, the references to terms such as "one embodiment / mode," "some embodiments / modes," "example," "specific example," or "some examples," etc., refer to specific features, structures, or characteristics described in connection with that embodiment / mode or example, which are included in at least one embodiment / mode or example of this disclosure. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment / mode or example. Moreover, the specific features, structures, or characteristics described may be combined in any suitable manner in one or more embodiments / modes or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments / modes or examples described in this specification, as well as the features of different embodiments / modes or examples.

[0130] Those skilled in the art should understand that the above embodiments are merely for illustrating the present disclosure and are not intended to limit the scope of the disclosure. Those skilled in the art can make other changes or modifications based on the above disclosure, and these changes or modifications still fall within the scope of the present disclosure.

Claims

1. A method for recognizing oral cavity regions using an electric toothbrush, characterized in that, include: The sensor data of the main IMU and slave IMU installed along the long axis of the electric toothbrush handle are collected synchronously during the operation of the electric toothbrush. The sensor data includes acceleration data and angular velocity data. Based on the acceleration data from the main IMU, the linear acceleration of the main IMU is determined as the first linear acceleration. Based on the acceleration data from the slave IMU, the linear acceleration of the slave IMU is determined as the second linear acceleration. Calculate the correlation between the first linear acceleration and the second linear acceleration to obtain the linear acceleration correlation. Calculate the difference between the angular velocity data of the master IMU and the slave IMU to obtain the angular velocity difference. The dynamic reliability factor is determined based on the linear acceleration correlation and the angular velocity difference. Feature extraction is performed based on sensor data from the master IMU and slave IMU to obtain the relative attitude deviation features, axial pulse features, and horizontal sway main direction features between the master IMU and slave IMU. Extract time-frequency features from sensor data based on the main IMU; The relative attitude deviation feature and axial pulse feature are weighted by the dynamic confidence factor, and the weighted result is fused with the time-frequency feature and the horizontal swing main direction feature to obtain the fused feature. as well as A fusion feature sequence is constructed based on the fusion features within a third preset time period. Oral region recognition is performed based on the fusion feature sequence to obtain oral region recognition results.

2. The method for identifying oral cavity regions with an electric toothbrush as described in claim 1, characterized in that, Based on the acceleration data from the main IMU, the linear acceleration of the main IMU is determined as the first linear acceleration, including: The acceleration data from the main IMU is low-pass filtered to obtain the first filtered acceleration data; and Based on the real-time attitude of the primary IMU and the first rotation matrix used to align the primary IMU coordinate system to the world coordinate system, determine the gravity components of the gravity vector in the primary IMU coordinate system; and The first linear acceleration is obtained by subtracting the gravity component from the first filtered acceleration data.

3. The method for identifying oral cavity regions with an electric toothbrush as described in claim 2, characterized in that, Also includes: Determine the first rotation matrix, including: In a static state, the acceleration data of the main IMU within a first preset time period is collected and the average value is calculated as the first average acceleration value; Based on the first average acceleration and the gravity vector, the first rotation matrix is ​​calculated using the following formula: in, Represents the gravity vector; Represents the first rotation matrix; This represents the average value of the first acceleration; This indicates the modulo operation.

4. The method for identifying oral cavity regions with an electric toothbrush as described in claim 1, characterized in that, Calculate the correlation between the first linear acceleration and the second linear acceleration to obtain the linear acceleration correlation, including: Based on the magnitude of the first linear acceleration within a second preset time period, a first linear acceleration magnitude sequence is constructed; based on the magnitude of the second linear acceleration within a second preset time period, a second linear acceleration magnitude sequence is constructed. Calculate the Pearson correlation coefficient between the first linear acceleration mode sequence and the second linear acceleration mode sequence, and use it as the linear acceleration correlation.

5. The method for identifying oral cavity regions with an electric toothbrush as described in claim 1, characterized in that, The dynamic credibility factor is calculated using the following formula: in, This represents the dynamic reliability factor at the k-th sampling time. and All of these represent positive adjustment parameters; This represents the linear acceleration correlation at the k-th sampling time. This represents the difference in angular velocity at the k-th sampling time.

6. The method for identifying oral cavity regions with an electric toothbrush as described in claim 1, characterized in that, The extraction process of the relative attitude deviation features includes: The relative rotation attitude is obtained by multiplying the real-time attitude of the slave IMU by the inverse of the real-time attitude of the master IMU. The relative rotation attitude represents the real-time dynamic rotation relationship of the slave IMU relative to the master IMU. Convert the relative rotation attitude into a relative rotation matrix; After transposing the second rotation matrix used to align the primary IMU coordinate system to the secondary IMU coordinate system, multiply it with the relative rotation matrix to obtain the relative attitude deviation matrix; The relative attitude deviation matrix is ​​converted into an axis-angle representation; and Based on the rotation angle and unit rotation axis in the axis angle representation, a relative posture deviation feature vector is constructed to characterize the spatial tilt orientation and tilt degree of the toothbrush, serving as the relative posture deviation feature.

7. The method for identifying oral cavity regions with an electric toothbrush as described in claim 1, characterized in that, The process of extracting the axial pulse feature includes: projecting the first linear acceleration onto the major axis of the electric toothbrush in the world coordinate system to obtain the axial pulse feature.

8. The method for identifying oral cavity regions with an electric toothbrush as described in claim 1, characterized in that, The extraction process of the main direction feature of the horizontal swing includes: Projecting the first linear acceleration onto the horizontal plane of the world coordinate system yields a two-dimensional horizontal motion vector; and The principal direction angle of the two-dimensional horizontal motion vector is calculated using a bivariate arctangent function, and is used as the principal direction feature of the horizontal swing.

9. An electric toothbrush, characterized in that, include: The memory stores execution instructions; as well as A processor that executes the execution instructions stored in the memory, causing the processor to perform the electric toothbrush oral cavity region recognition method according to any one of claims 1 to 8.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the electric toothbrush oral cavity area recognition method according to any one of claims 1 to 8.

Images