Gesture mode switching method and system for radar body detector

Abstract

The application relates to the technical field of radar body detectors, and particularly discloses a radar body detector gesture mode switching method and system, wherein the method comprises the following steps: acquiring a radar echo signal, and extracting motion parameter information of a user according to the radar echo signal; generating a motion state index sequence according to the motion parameter information; analyzing the motion state index sequence, detecting a signal segment with a radar feature mode similar to a preset gesture and greater than a first preset threshold, and marking the signal segment as a potential gesture signal; and based on context intention judgment, analyzing the potential gesture signal to obtain an intention judgment score, and judging whether the potential gesture signal is a mode switching instruction; the method effectively distinguishes between a conscious mode switching instruction of a user and an unconscious background motion, and is adapted to gesture variation caused by changes in the position and posture of the user, so that the false rejection rate is significantly reduced, and the gesture interaction accuracy and reliability are improved.

Description

Technical Field

[0001] This application relates to the field of radar body detector technology, and more specifically, to a method and system for switching gesture modes in a radar body detector. Background Technology

[0002] Radar body detectors (also known as radar human body detectors) are advanced health and wellness devices that provide unobtrusive, continuous monitoring of vital signs and activity levels through fixed installation in the user's daily environment. The devices are typically deployed in specific locations in bedrooms or living rooms, utilizing millimeter-wave radar modules to transmit and receive echo signals, enabling real-time monitoring of indicators such as heart rate, respiration, sleep quality, and fall detection. Simultaneously, this technology supports gesture recognition, allowing users to execute mode-switching commands through simple actions, such as pausing monitoring or switching to activity mode, enhancing ease of interaction. This intelligent monitoring solution is driving the health management field towards contactless and automated development, meeting users' demands for efficient and comfortable health services, and demonstrating broad market application prospects.

[0003] However, in healthcare scenarios, users' activity locations and postures are highly uncertain. Users may move around in different areas of a room, such as beside the bed, at a desk, or in the center of the room. When users execute preset gesture commands in non-ideal locations, radar echo signals change significantly due to distance, angle, direction, or partial obstruction, resulting in gesture features that differ greatly from the standard model. Existing recognition algorithms struggle to accurately match these varied gestures. Furthermore, during continuous monitoring, devices capture various non-instructional body activities, such as turning over in sleep, stretching limbs, or scratching the body when awake. The radar echo features of these unconscious movements may show accidental similarities to preset gesture commands. Because existing systems lack a deep understanding mechanism for user intent, they cannot effectively distinguish between background movement and genuine commands, easily leading to misjudgments. For example, unconscious movements may be identified as mode switching commands, unexpectedly interrupting monitoring or switching to an inappropriate mode. These problems reduce the accuracy and reliability of gesture interaction, affecting user experience and the continuity of health monitoring, and limiting the practical application of contactless interaction technology.

[0004] There is currently no effective technical solution to the above problems. Summary of the Invention

[0005] The purpose of this application is to provide a method and system for switching gesture modes in a radar body detector, so as to distinguish between the user's conscious command gestures and unconscious background movements, reduce the misjudgment rate of gesture recognition, improve the accuracy and reliability of gesture interaction, improve user experience, and ensure the continuity of health monitoring.

[0006] In a first aspect, this application provides 1. a method for switching gesture modes in a radar body detector, used to determine whether a user's gesture is a mode switching command, characterized in that the method includes the following steps:

[0007] S1. Acquire radar echo signals and extract user motion parameter information based on the radar echo signals;

[0008] S2. Based on the motion parameter information, generate a sequence of motion state indicators with a first preset window length arranged in a temporal order;

[0009] S3. Analyze the motion state index sequence, detect signal segments whose similarity to the radar feature pattern of the preset gesture is greater than a first preset threshold, and mark the signal segments as potential gesture signals;

[0010] S4. Based on the contextual intent judgment, analyze the potential gesture signals to obtain the intent judgment score;

[0011] S5. Based on the intention, determine the score and whether the potential gesture signal is a mode switching instruction.

[0012] The aforementioned radar body detector gesture mode switching method, wherein step S1 includes:

[0013] S11. Acquire radar echo signals;

[0014] S12. Perform multi-target separation processing on the radar echo signal to obtain target echo signals of multiple moving targets;

[0015] S13. Filter the target echo signal to filter out interference signals that have a similarity to a preset non-user motion feature pattern greater than a second preset threshold.

[0016] S14. Extract the motion parameter information based on the filtered target echo signal.

[0017] The aforementioned radar body detector gesture mode switching method, wherein step S13 includes:

[0018] S131. Extract motion feature parameters based on the target echo signal;

[0019] S132. Compare the motion feature parameters with a preset non-user motion feature pattern to determine the similarity between the target echo signal and the non-user motion feature pattern.

[0020] S133. Determine whether the similarity is greater than the second preset threshold. If the similarity is greater than the second preset threshold, then the target echo signal is determined to be an interference signal and filtered out.

[0021] The aforementioned radar body detector gesture mode switching method, wherein step S2 includes:

[0022] S21. Perform time-domain smoothing on the motion parameter information to obtain a time-domain smoothed sequence of motion parameter information with a time length equal to the length of the first preset window.

[0023] S22. Based on the second preset window length, the motion parameter information sequence is split and recombined into multiple motion parameter segments;

[0024] S23. Extract features from each motion parameter segment to obtain one or more of the user's overall motion energy, limb activity amplitude, position change rate and motion trajectory dispersion, as motion state indicators.

[0025] S24. Arrange the motion state indicators in chronological order to generate the motion state indicator sequence.

[0026] The aforementioned radar body detector gesture mode switching method, wherein step S3 includes:

[0027] S31. Extract the motion state index sequence based on the third preset window length to obtain multiple signal segments;

[0028] S32. Extract the changing trends and statistical characteristics of motion state indicators for each signal segment;

[0029] S33. Based on the changing trend and the statistical features, compare the radar feature patterns of the preset gestures to obtain a similarity score;

[0030] S34. Determine whether the similarity score of each signal segment is greater than the first preset threshold.

[0031] S35. Mark the signal segments with similarity scores greater than the first preset threshold as the potential gesture signals.

[0032] The aforementioned radar body detector gesture mode switching method, wherein the radar feature mode includes multiple reference vectors representing different gestures, composed of reference change trends and reference statistical features, and step S33 includes:

[0033] S331. Construct a multi-dimensional vector based on the changing trends and statistical characteristics of motion state indicators of each signal segment;

[0034] S332. Calculate and obtain the initial similarity score based on the spatial distance between the multidimensional vector of each signal segment and the reference vector of each radar feature pattern.

[0035] S333. Filter and obtain the maximum similarity score corresponding to each signal segment, and use it as the similarity score.

[0036] The aforementioned radar body detector gesture mode switching method, wherein step S4 includes:

[0037] S41. Based on the motion state index sequence, analyze the motion state index before the occurrence of the potential gesture signal, and determine the pre-gesture intention score of the potential gesture signal.

[0038] S42. Based on the motion state index sequence, analyze the motion state index after the execution of the potential gesture signal, and determine the subsequent gesture intention score of the potential gesture signal;

[0039] S43. Extract the motion features of the potential gesture signal, and compare the motion features with preset intentional gesture feature patterns and unconscious background motion feature patterns to determine the gesture matching intention score of the potential gesture signal.

[0040] S44. Calculate the intent judgment score based on the pre-gesture intent score, the post-gesture intent score, and the gesture matching intent score.

[0041] The aforementioned radar body detector gesture mode switching method, wherein step S43 includes:

[0042] S431. Extract the motion features of the potential gesture signal;

[0043] S432. The action feature is compared with a preset intentional gesture feature pattern to obtain a first similarity, wherein the first similarity is the similarity between the potential gesture signal and the intentional gesture feature pattern.

[0044] S433. The motion features are compared with a preset unconscious background motion feature pattern to obtain a second similarity, wherein the second similarity is the similarity between the potential gesture signal and the unconscious background motion feature pattern.

[0045] S434. Calculate the gesture matching intent score based on the first similarity and the second similarity.

[0046] The aforementioned radar body detector gesture mode switching method, wherein step S5 includes:

[0047] S51. Compare the intent judgment score with a preset instruction confirmation threshold to determine whether the potential gesture signal meets the conditions for a mode switching instruction.

[0048] S52. If the potential gesture signal meets the conditions of the mode switching instruction, then the potential gesture signal is determined as the mode switching instruction, and the corresponding mode switching operation is executed.

[0049] Secondly, this application also provides a radar body detector gesture mode switching system for determining whether a user's gesture is a mode switching command, the system comprising:

[0050] The acquisition module is used to acquire radar echo signals and extract the user's motion parameter information based on the radar echo signals.

[0051] The sequence generation module is used to generate a sequence of motion state indicators with a first preset window length arranged in a temporal order based on the motion parameter information.

[0052] The sequence analysis module is used to analyze the motion state index sequence, detect signal segments whose similarity to the radar feature pattern of a preset gesture is greater than a first preset threshold, and mark the signal segments as potential gesture signals.

[0053] The intent integration module is used to analyze the potential gesture signals based on contextual intent judgment to obtain an intent judgment score;

[0054] The intent determination module is used to determine whether the potential gesture signal is a mode switching instruction based on the intent determination score.

[0055] As can be seen from the above, this application provides a gesture mode switching method and system for a radar body detector. The method of this application combines gesture feature pattern matching with a context-based intent judgment mechanism to effectively distinguish between the user's conscious mode switching command and unconscious background movement, and adapts to gesture variations caused by changes in user position and posture, thereby achieving the effect of significantly reducing the misjudgment rate and improving the accuracy and reliability of gesture interaction. Attached Figure Description

[0056] Figure 1 A flowchart of the gesture mode switching method for a radar body detector provided in this application embodiment.

[0057] Figure 2 This is a schematic diagram of the radar body detector gesture mode switching system provided in an embodiment of this application.

[0058] Attached reference numerals: 201, Acquisition module; 202, Sequence generation module; 203, Sequence analysis module; 204, Intent integration module; 205, Intent determination module. Detailed Implementation

[0059] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0060] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0061] Firstly, please refer to Figure 1 This application provides a method for switching gesture modes in a radar body detector, used to determine whether a user's gesture is a mode switching command. The method includes the following steps:

[0062] S1. Acquire radar echo signals and extract user motion parameter information based on radar echo signals;

[0063] S2. Based on the motion parameter information, generate a sequence of motion state indicators with a first preset window length arranged in time sequence;

[0064] S3. Analyze the motion state index sequence, detect signal segments whose similarity to the radar feature pattern of the preset gesture is greater than the first preset threshold, and mark the signal segments as potential gesture signals.

[0065] S4. Based on contextual intent judgment, analyze potential gesture signals to obtain intent judgment scores;

[0066] S5. Determine the score based on the intent and determine whether the potential gesture signal is a mode switching command.

[0067] Specifically, acquiring radar echo signals and extracting motion parameter information refers to transmitting and receiving electromagnetic waves through radar equipment, capturing signals reflected from the user's body, and parsing quantitative data describing the user's motion state from these signals. Millimeter-wave radar modules can be used to acquire these signals. Motion parameter information can include the user's distance, speed, angle, Doppler shift, or target echo intensity, etc., which are used to establish the basic data for gesture recognition, converting physical motion into processable digital information and providing raw input for subsequent analysis.

[0068] More specifically, the motion state index sequence is a data sequence that is organized in chronological order by extracting quantitative indicators (i.e., motion state indicators) from motion parameter information that can characterize the overall or local motion state of a user. Motion state indicators are used to capture the dynamic characteristics and trends of user motion, transforming instantaneous parameters into more representative temporal features, and providing more stable and meaningful input for gesture pattern matching.

[0069] More specifically, the detection of potential gesture signals can be achieved by using pattern recognition algorithms, machine learning classifiers, or template matching techniques to achieve similarity detection. The radar feature pattern of the preset gesture can be composed of multiple reference vectors representing different gestures, which can be used to initially screen out signals that may be user gestures, eliminate a large number of irrelevant background movements, and improve the efficiency of subsequent processing.

[0070] More specifically, contextual intent judgment analysis can be performed by analyzing the motion stability before the potential gesture signal appears, the motion coherence after execution, and the degree of matching between the gesture action features and the intentional gesture pattern or unconscious background motion pattern to calculate the intent judgment score. This is used to introduce a deeper judgment of the user's intent, distinguish between conscious command gestures and unconscious background motion, and thus solve the problem of misjudgment.

[0071] Specifically, the method of this application first acquires radar echo signals and extracts the user's motion parameter information from these signals, which serves as the foundation for all subsequent analyses. Next, based on this motion parameter information, a motion state index sequence with a specific time length is generated. This time-series processing method captures the dynamic characteristics of the user's movement, providing more comprehensive information for gesture recognition. Subsequently, the system analyzes this motion state index sequence, comparing it with preset gesture radar feature patterns to initially filter out signal segments with similarity reaching a preset threshold, and marking these signal segments as potential gesture signals. This step completes the initial recognition of possible gestures. However, to address the problem of misjudgment of background motion, this method further introduces a contextual intent judgment mechanism. Based on the motion state index sequence, the system deeply analyzes the motion state before and after the appearance of the potential gesture signal, extracts the action features of the potential gesture signal itself, and compares it with intentional gesture feature patterns and unconscious background motion feature patterns, thereby comprehensively calculating an intent judgment score. This score quantifies the user's true intent in performing the action. Finally, based on this intent judgment score, the system makes a final decision to determine whether the potential gesture signal is indeed a mode switching command. Through this progressive process that combines feature matching and intent judgment, the method of this application can effectively distinguish between conscious command gestures and unconscious background movements, and can achieve accurate gesture recognition even in complex environments where the user's position and posture are uncertain.

[0072] The method of this application combines gesture feature pattern matching with a context-based intent judgment mechanism to effectively distinguish between conscious mode switching commands and unconscious background movements, and adapts to gesture variations caused by changes in user position and posture, thereby significantly reducing the misjudgment rate and improving the accuracy and reliability of gesture interaction.

[0073] In some preferred embodiments, step S1 includes:

[0074] S11. Acquire radar echo signals;

[0075] S12. Perform multi-target separation processing on the radar echo signal to obtain the target echo signals of multiple moving targets;

[0076] S13. Filter the target echo signal to filter out interference signals that are more similar to the preset non-user motion feature pattern than the second preset threshold.

[0077] S14. Extract motion parameter information based on the filtered target echo signal.

[0078] Specifically, after obtaining the radar echo signal in step S11, the radar echo signal can be preprocessed. Preprocessing can include denoising and Doppler processing. Denoising refers to eliminating random noise and clutter in the radar echo signal through algorithms to improve the signal-to-noise ratio. This can be achieved using wavelet transform, Kalman filtering, or mean filtering. Doppler processing refers to extracting the Doppler frequency shift information caused by target motion from the radar echo signal to obtain the target's radial velocity. This can be achieved using fast Fourier transform (FFT), short-time Fourier transform (STFT), or Chirp-Z transform.

[0079] More specifically, multi-target separation processing refers to distinguishing and identifying the echo signals of multiple overlapping moving targets to obtain the echo signal of each independent moving target. This can be achieved by using clustering algorithms (such as DBSCAN and K-means) to analyze the range-Doppler map, or by using multi-target tracking algorithms (such as multi-hypothesis tracking and joint probability data association). The target echo signals of the separated multiple moving targets actually correspond to multiple moving parts of the user. Filtering processing refers to screening the signal to remove signal components that do not conform to a specific pattern. This can be achieved using digital filters (such as Butterworth filters and Chebyshev filters), adaptive filtering, or pattern matching algorithms. Preset non-user motion feature patterns refer to a predefined set of motion signal features that do not belong to the user's active gestures. These can include patterns such as environmental clutter, device noise, non-target object movement, or unconscious background movement by the user. These patterns can be modeled through historical data analysis or expert experience. The second preset threshold is a critical value used to judge the similarity between the target echo signal and the preset non-user motion feature pattern. When the similarity exceeds this value, the signal is judged as interference. It can be set according to the system performance requirements and the balance between false positive rate and false negative rate.

[0080] Specifically, step S11 involves preprocessing the acquired raw radar echo signal, a step fundamental to all subsequent analyses. Noise reduction effectively suppresses random interference, improving signal purity. Doppler processing identifies motion components and distinguishes moving targets from stationary backgrounds, laying the foundation for subsequent motion analysis. In real-world environments, radar may simultaneously receive echoes from different moving parts of the user (e.g., hands, arms) or other moving objects. Step S12's multi-target separation ensures that the echo signal of each independent moving target can be individually identified and extracted, avoiding signal aliasing interference with subsequent analysis and allowing the system to focus on the signal from the user's specific moving parts. Subsequently, step S13 filters the separated target echo signals. Even after multi-target separation, the signal may still contain interference signals unrelated to the user's gestures but possessing motion characteristics, such as environmental clutter or subtle, unconscious movements by the user. By comparing these signals with preset non-user motion feature patterns and filtering out interference signals with similarity exceeding a certain threshold, the target echo signals are further purified, ensuring that the signals used for final analysis are highly correlated, thereby significantly reducing the possibility of misjudgment due to non-commanded movements. Finally, step S14 extracts the user's motion parameter information based on the highly purified and accurate target echo signals after the above preprocessing, multi-target separation, and filtering. This series of refined processing steps ensures that the extracted motion parameter information can accurately reflect the true characteristics of the user's gestures, thus providing high-quality input for subsequent steps (such as generating motion state index sequences, detecting gesture feature patterns, and performing contextual intent judgment).

[0081] Through the above processing, the method of this application can effectively solve the problems of noise, multi-target aliasing and interference signals of non-user motion feature patterns in the original radar echo signal. By performing refined preprocessing, multi-target separation and filtering on the radar echo signal, the accuracy of the extracted motion parameter information can be significantly improved, thereby enhancing the reliability of subsequent gesture recognition and intent judgment, reducing the occurrence of misjudgment, improving user experience and ensuring the continuity of health monitoring.

[0082] In some preferred embodiments, step S13 includes:

[0083] S131. Extract motion feature parameters based on the target echo signal;

[0084] S132. Compare the motion characteristic parameters with the preset non-user motion characteristic pattern to determine the similarity between the target echo signal and the non-user motion characteristic pattern.

[0085] S133. Determine whether the similarity is greater than the second preset threshold. If the similarity is greater than the second preset threshold, then the target echo signal is determined to be an interference signal and filtered out.

[0086] Specifically, motion characteristic parameters can include the frequency distribution, energy distribution, and time series variation characteristics of the signal. Motion characteristic parameters refer to quantitative indicators extracted from radar echo signals that can characterize the dynamic characteristics of the signal. The frequency distribution of the signal can be obtained by signal processing techniques such as Fourier transform and wavelet analysis, the energy distribution of the signal can be obtained by calculating the root mean square value, peak value, or integral, and the time series variation characteristics can be obtained by analyzing the variation trend of the amplitude, phase, or Doppler frequency shift of the signal at different time points.

[0087] More specifically, the preset non-user motion feature patterns can be trained and learned from known interference signal samples using supervised learning algorithms, or the frequency, energy, time series, and other features of different types of interference signals can be summarized and quantified through expert experience, statistical analysis, and other methods to form feature vectors or pattern libraries for comparison. Similarity can be calculated using mathematical methods such as Euclidean distance, cosine similarity, and correlation coefficient.

[0088] Specifically, step S131 involves in-depth analysis of the target echo signal obtained after multi-target separation processing, extracting motion feature parameters across multiple dimensions. Subsequently, step S132 compares the extracted motion feature parameters with pre-established non-user motion feature patterns. These non-user motion feature patterns cover common interference sources in healthcare scenarios, such as environmental clutter, equipment noise, and motion patterns of non-target objects. Through this targeted comparison, the system can quantify the similarity between the target echo signal and these known interference patterns. Simultaneously, during the comparison process, the system also considers the similarity between the target echo signal and the non-user motion feature patterns, which helps avoid misjudging subtle but effective user movements as interference. Finally, step S133 compares the calculated similarity with a preset second threshold. If the similarity is greater than this threshold, it indicates a high degree of similarity between the target echo signal and the non-user motion feature pattern; in this case, the signal is explicitly identified as interference and filtered out.

[0089] Through the meticulous feature extraction, pattern comparison, and threshold judgment described above, the method of this application can effectively distinguish and suppress complex and diverse interference signals in health maintenance scenarios, such as subtle environmental clutter, noise generated by the equipment itself, and the movement of non-target objects. This accurate identification and filtering of interference signals ensures the purity and reliability of the motion parameter information extracted from the target echo signal.

[0090] In some preferred embodiments, step S2 includes:

[0091] S21. Perform time-domain smoothing on the motion parameter information to obtain a time-domain smoothed sequence of motion parameter information with a time length of the first preset window.

[0092] S22. Based on the second preset window length, the motion parameter information sequence is split and recombined into multiple motion parameter segments;

[0093] S23. Extract features from each motion parameter segment to obtain one or more of the user's overall motion energy, limb activity amplitude, position change rate and motion trajectory dispersion, as motion state indicators.

[0094] S24. Arrange the motion state indicators in chronological order to generate a motion state indicator sequence.

[0095] Specifically, time-domain smoothing refers to filtering time-series data to reduce instantaneous fluctuations and noise interference. Algorithms such as moving average filtering, exponential smoothing filtering, or Kalman filtering can be used. The first preset window length refers to the time span used to define the motion parameter information sequence or the final motion state index sequence, and can be set according to the typical duration of the gesture or the system's processing capacity. The second preset window length refers to the length used to divide the motion parameter information sequence into smaller, analyzable time periods. It is smaller than the first preset window length and can be determined based on the precision of the local gesture features or computational efficiency.

[0096] More specifically, a motion parameter segment refers to a time segment extracted from a continuous sequence of motion parameter information. It may contain raw or preprocessed motion data such as motion trajectory, velocity, and acceleration within a specific time window.

[0097] More specifically, overall user kinetic energy refers to the quantitative representation of a user's kinetic intensity or activity level within a specific time period. Limb range of motion refers to the quantitative representation of a user's limb range of motion or extension within a specific time period. Rate of position change refers to how quickly a user's position changes over time. Motion trajectory dispersion refers to the stability or complexity of a user's movement path.

[0098] Specifically, step S21 employs temporal smoothing to make the original motion parameter data more stable and continuous, thus forming a pre-purified motion parameter information sequence with a first preset window length. Subsequently, step S22 splits and reassembles the smoothed motion parameter information sequence into multiple motion parameter segments based on a second preset window length. This subdivision operation allows the system to capture motion features from different temporal granularities or local time windows, laying the foundation for extracting local dynamic features related to specific gestures. By dividing the continuous motion sequence into smaller, manageable segments, more flexible and refined feature extraction is possible. Next, step S23 extracts features from these split motion parameter segments, extracting one or more of the following as motion state indicators: overall user motion energy, limb range of motion, rate of position change, and motion trajectory dispersion. These features have physical meaning and discriminative power, comprehensively describing the user's motion state, such as motion intensity, limb extension range, motion speed, and motion path stability. By extracting these robust features, the essential characteristics of gestures can be effectively captured, and sensitivity to minor fluctuations in the original data can be reduced. Finally, step S24 rearranges the motion state indicators extracted from each motion parameter segment according to their original temporal order, forming a complete, temporally sequenced sequence of motion state indicators. This sequence, the result of multi-layer processing and feature extraction, accurately describes the user's motion state changes within a specific time window in a structured, high-dimensional form. Through the above processing flow, the method of this application can generate a high-quality, highly expressive sequence of motion state indicators. This sequence, as input for gesture radar feature pattern matching, significantly improves the accuracy and reliability of gesture recognition.

[0099] In some preferred embodiments, the motion parameter segment includes motion trajectory and velocity;

[0100] Step S23, the process of extracting the user's overall motion energy, includes:

[0101] The displacement is extracted based on the motion trajectory of the motion parameter segment, and the user's overall motion energy is calculated by combining the corresponding velocity. The user's overall motion energy is the sum of the square of the displacement and the square of the velocity.

[0102] Step S23, the process of extracting the range of motion of the limbs, includes:

[0103] Extracting limb activity amplitude based on motion trajectory using motion parameter segments;

[0104] Step S23, the process of extracting the rate of position change includes:

[0105] The rate of change of position over time is calculated based on the motion trajectory of the motion parameter segment, and is used as the position change rate.

[0106] Step S23, the process of extracting the dispersion of the motion trajectory includes:

[0107] The average position is determined based on the motion trajectory of the motion parameter segment, and the degree of deviation of the motion trajectory of the motion parameter segment from the average position is calculated as the motion trajectory dispersion.

[0108] Specifically, the motion trajectory refers to the path of a user's limbs in space that changes over time. It can be obtained using radar target tracking algorithms by calculating the position of the target echo signal and connecting continuous points. Speed ​​refers to the rate and direction of the user's limb movement.

[0109] Specifically, when extracting features from motion parameter segments, the solution uses the motion trajectory to extract displacement for the user's overall motion energy, and combines this with the corresponding velocity. The intensity of the user's motion over a specific time period is comprehensively quantified by calculating the sum of the squares of the displacement and the squares of the velocity. For limb range of motion, the solution directly determines the range of motion of the limbs in space based on the motion trajectory. When extracting the rate of change of position, the solution obtains this by calculating the rate of change of the motion trajectory over time, ensuring a close correlation between the rate and the actual motion path. Furthermore, to measure the stability or concentration of motion, the solution determines the average position based on the motion trajectory and calculates the degree of deviation of the trajectory from this average position, using this as the motion trajectory dispersion. In this way, the solution concretizes the abstract motion parameter segments into data containing motion trajectory and velocity, and provides a clear calculation method for extracting various motion state indicators. This makes the motion state indicators extracted from radar echo signals more accurate and have stronger discriminative power, effectively addressing the problems of radar echo signals being easily affected by environmental factors and the possibility that unconscious user movements may accidentally resemble preset gesture commands.

[0110] In some preferred embodiments, step S3 includes:

[0111] S31. Extract multiple signal segments by truncating the motion state index sequence based on the third preset window length;

[0112] S32. Extract the changing trends and statistical characteristics of motion state indicators for each signal segment;

[0113] S33. Based on the changing trends and statistical characteristics, compare the radar feature patterns of preset gestures to obtain similarity scores;

[0114] S34. Determine whether the similarity score of each signal segment is greater than the first preset threshold.

[0115] S35. Mark signal segments with similarity scores greater than the first preset threshold as potential gesture signals.

[0116] Specifically, the third preset window length refers to the time or number of data points used to extract the motion state index sequence to obtain the signal segment; it is less than the first preset window length but greater than the second preset window length. The change trend of the motion state index refers to the dynamic characteristics of the motion state index changing over time, which can be characterized using methods such as difference, derivative, slope, or dynamic time warping path. Statistical characteristics refer to the overall distribution and central tendency of the motion state index within a specific time window, which can be characterized using methods such as mean, variance, standard deviation, and median. The radar feature pattern of the preset gesture refers to the pre-stored feature set used to represent specific gesture actions. The similarity score is a quantified value of the degree of matching between the features of the current signal segment and the radar feature pattern of the preset gesture, which can be calculated using methods such as Euclidean distance, cosine similarity, correlation coefficient, the reciprocal of the dynamic time warping distance, or the probability value output by the classifier.

[0117] Specifically, step S31 extracts and acquires multiple signal segments from a continuous sequence of motion state indicators, based on a third preset window length. This segmentation process decomposes the user's overall motion into smaller, more easily identifiable action units, providing clear and independent analysis objects for subsequent feature extraction and gesture comparison. Next, step S32 extracts the changing trends and statistical characteristics of the motion state indicators for each signal segment. These high-level features are a deep abstraction and generalization of the original motion parameter information; the changing trends capture the dynamic characteristics of the gesture, while the statistical characteristics reflect the overall distribution and stability of the gesture within a specific time period. Compared to directly using the original motion parameters, these features are more robust to noise and subtle individual differences, and can more accurately characterize the essential features of the gesture, thus effectively addressing the problem of significant changes in radar echo signals when users perform gestures in non-ideal positions. Subsequently, step S33 compares the extracted trend and statistical features with the radar feature pattern of the preset gesture to obtain a similarity score. Utilizing these more representative and robust features for comparison allows the system to more accurately assess the matching degree between the current movement and the preset gesture pattern, effectively overcoming gesture feature variations caused by user posture, distance, or partial occlusion, improving the accurate matching ability for varied gesture actions, and reducing misjudgments caused by accidental similarities between unconscious background movements and preset gestures. Further, step S34 determines whether the similarity score of each signal segment is greater than a first preset threshold, providing a quantitative standard for screening potential gestures and helping to filter out random or non-instructional actions with low matching degrees to the gesture pattern. Finally, step S34 marks signal segments with similarity scores greater than the first preset threshold as potential gesture signals. This marking process clearly identifies the signal segments that have undergone preliminary screening and feature comparison, serving as input for subsequent contextual intent judgment.

[0118] Through the above processing, the method of this application can effectively address the problem of gesture feature variation caused by the uncertainty of user activity position and posture, and improve the accurate matching ability of varied gesture actions. At the same time, this application can perform refined analysis and feature extraction on motion state index sequences, effectively distinguish between background motion and real commands, and reduce misjudgments caused by accidental similarity between unconscious actions and preset gestures.

[0119] In some preferred embodiments, the radar feature pattern includes multiple reference vectors representing different gestures, composed of reference change trends and reference statistical features. Step S33 includes:

[0120] S331. Construct a multi-dimensional vector based on the changing trends and statistical characteristics of motion state indicators of each signal segment;

[0121] S332. Calculate and obtain the initial similarity score based on the spatial distance between the multidimensional vector of each signal segment and the reference vector of each radar feature pattern.

[0122] S333. Filter and obtain the maximum similarity score corresponding to each signal segment, and use it as the similarity score.

[0123] Specifically, spatial distance refers to a quantitative measure of the similarity between two vectors in a multidimensional feature space, which can be achieved using various distance measurement methods such as Euclidean distance, Manhattan distance, cosine similarity, or Mahalanobis distance. The initial similarity score refers to the quantified similarity value obtained by comparing the multidimensional vector of the signal segment to be identified with a single reference vector in the radar feature pattern. It can be the reciprocal of the distance, a negative exponential function of the distance, or a normalized distance value.

[0124] Specifically, step S331 constructs a multi-dimensional vector based on the changing trends and statistical characteristics of motion state indicators of each signal segment. By integrating the changing trends and statistical characteristics of motion state indicators extracted from potential gesture signal segments, a unified multi-dimensional vector is formed, transforming complex temporal motion information into quantifiable numerical features. This allows the features of different gestures to be compared in a unified mathematical space, providing standardized input for subsequent similarity calculations. Next, step S332 quantifies the similarity between the multi-dimensional vector of the signal segment to be identified and each reference vector in the preset radar feature pattern by calculating the spatial distance between them. The smaller the spatial distance, the closer they are in the feature space, i.e., the higher the similarity. This distance-based similarity calculation method can effectively measure the matching degree between the signal segment and the standard gesture pattern. Even if there are subtle variations in the gesture, they can be reflected by the distance changes in the feature space, thereby improving the ability to recognize variant gestures. Finally, step S333 selects the maximum similarity score corresponding to each signal segment as the final similarity score. Since a potential gesture signal segment may have varying degrees of similarity to multiple preset reference vectors, the final similarity score is obtained by selecting the signal segment with the highest similarity among all reference vectors. This ensures that the system can identify the preset gesture pattern that best matches the current signal segment, even when the user's gesture is not standard or multiple similar gestures exist, the most likely matching result can be selected. Through this synergistic effect, this solution effectively solves the misjudgment problem caused by user gesture variations and background motion interference, improving the accuracy and reliability of gesture recognition.

[0125] In some preferred embodiments, step S4 includes:

[0126] S41. Based on the motion state index sequence, analyze the motion state index before the occurrence of potential gesture signals to determine the pre-gesture intention score of potential gesture signals.

[0127] S42. Based on the motion state index sequence, analyze the motion state index after the execution of the potential gesture signal, and determine the post-gesture intention score of the potential gesture signal.

[0128] S43. Extract the motion features of the potential gesture signal, and compare the motion features with the preset intentional gesture feature pattern and unconscious background motion feature pattern to determine the gesture matching intention score of the potential gesture signal.

[0129] S44. Calculate the intent judgment score based on the pre-gesture intent score, the post-gesture intent score, and the gesture matching intent score.

[0130] Specifically, the pre-gesture intent score is used to evaluate the stability or readiness characteristics of the user's motion state before the potential gesture signal appears; the post-gesture intent score is used to evaluate the recovery or stability characteristics of the user's motion state after the potential gesture signal is executed. Motion features refer to the set of parameters that describe the kinematic or dynamic characteristics of the potential gesture signal itself.

[0131] More specifically, intentional gesture feature patterns refer to a set of typical features of specific gesture commands consciously executed by a user, pre-trained or defined. These can be constructed using expert experience definitions, statistical models trained on user demonstration data, deep learning models, or based on keyframe feature vectors. Unconscious background motion feature patterns refer to a set of typical features of unconscious background activities (such as rolling over, stretching, scratching), pre-trained or defined. These can be constructed using statistical distributions extracted from large amounts of unconscious activity data, anomaly detection models, or typical patterns obtained through cluster analysis. Gesture matching intent score is used to assess the similarity or dissimilarity between the motion features of a potential gesture signal and the intentional and unconscious background motion patterns. Intent judgment score is a comprehensive quantitative indicator used as the final criterion for comprehensively evaluating whether a potential gesture signal is a conscious command.

[0132] Specifically, this solution constructs a multi-dimensional intent judgment mechanism to analyze detected potential gesture signals, distinguishing between conscious mode-switching commands issued by the user and unconscious background movements. The core of this mechanism lies in comprehensively considering three time and feature dimensions: before the gesture, after the gesture, and the gesture itself. Step S41 assesses whether the user is in a static or stable state before the gesture, providing a preliminary judgment on whether the gesture is intentional. Step S42 further verifies the gesture's intent by observing whether the user returns to a stable state after the gesture. Step S43 uses bidirectional comparison of motion features with pre-set intentional gesture feature patterns and unconscious background movement feature patterns to determine whether the gesture is more inclined towards conscious command actions or unconscious background interference, thus obtaining a gesture matching intent score. Finally, step S44 comprehensively calculates the pre-gesture intent score, post-gesture intent score, and gesture matching intent score to obtain an intent judgment score. This multi-dimensional fusion judgment method combines contextual information before and after the gesture, as well as the inherent characteristics of the gesture itself, making the final intent judgment more accurate and reliable. The method of this application improves the accuracy and reliability of gesture recognition by introducing multi-dimensional context and gesture feature analysis to deeply filter and judge these potential signals, thereby effectively solving the problem of misjudging unconscious background movements as mode switching commands.

[0133] In some preferred embodiments, step S41 includes:

[0134] S411. Based on the motion state index sequence, obtain the statistical characteristics of the motion state index of the signal segment before the potential gesture signal appears, and calculate the stability degree before the potential gesture signal appears based on the statistical characteristics and the preset statistical threshold.

[0135] S412. Calculate the pre-gesture intention score based on the degree of stability before the potential gesture signal appears.

[0136] Specifically, the preset statistical threshold is used to compare with the statistical characteristics of motion state indicators to determine the stability of the user's motion state. This can be determined based on empirical data, machine learning model training results, or expert knowledge. The degree of stability refers to the quantified degree of smoothness or fluctuation in the user's motion state before a potential gesture occurs, based on the comparison between the statistical characteristics of motion state indicators and the preset statistical threshold.

[0137] Specifically, the method in this application calculates the stability of user movement by comparing statistical characteristics with preset statistical thresholds. Higher stability indicates a more stable state before the user executes a potential gesture, thus assigning a higher pre-gesture intent score to that gesture. This mechanism ensures that the pre-gesture intent score accurately reflects the user's preparation state before the gesture is executed. This solution is closely integrated with the pre-processing solution. This integration enables the system to more robustly determine user intent in complex and changing user environments, reducing misjudgments and improving the accuracy of gesture interaction and user experience.

[0138] Similarly, step S42 can be designed as follows:

[0139] S421. Based on the motion state index sequence, obtain the statistical characteristics of the motion state index of the signal segment after the execution of the potential gesture signal, and calculate the stability of the potential gesture signal after execution based on the statistical characteristics and the preset statistical threshold.

[0140] S422. Calculate the gesture intent score based on the stability of the potential gesture signal after execution.

[0141] In some preferred embodiments, step S43 includes:

[0142] S431. Extract the motion features of potential gesture signals;

[0143] S432. Compare the action features with the preset intentional gesture feature pattern to obtain the first similarity, where the first similarity is the similarity between the potential gesture signal and the intentional gesture feature pattern.

[0144] S433. Compare the motion features with the preset unconscious background motion feature pattern to obtain the second similarity. The second similarity is the similarity between the potential gesture signal and the unconscious background motion feature pattern.

[0145] S434. Calculate the gesture matching intent score based on the first similarity and the second similarity.

[0146] Specifically, motion characteristics refer to quantitative indicators that describe the motion characteristics of a gesture, which may include peak velocity, duration, and displacement in the main direction.

[0147] More specifically, the first similarity refers to the degree of matching between the motion features of the potential gesture signal and the intentional gesture feature pattern, while the second similarity refers to the degree of matching between the motion features of the potential gesture signal and the unconscious background motion feature pattern.

[0148] Specifically, peak velocity reflects the force and instantaneous intensity of the gesture execution, duration characterizes the completeness and coherence of the gesture, and main directional displacement reveals the main trajectory and intention direction of the gesture. By integrating these features, the dynamic characteristics of the potential gesture signal can be comprehensively and meticulously described, laying the foundation for subsequent accurate comparison. Subsequently, the extracted motion features are compared with two preset modes. The first similarity directly quantifies the degree of matching between the potential gesture signal and the expected mode-switching command gesture, used to identify the real command. The second similarity is used to evaluate the similarity between the potential gesture signal and everyday unconscious movements, thereby identifying and suppressing background interference. By calculating the similarity with intentional gestures and unconscious background movements separately, the system can obtain a more comprehensive perspective on the nature of the potential gesture signal, avoiding the one-sidedness that may result from a single comparison. Finally, based on the first and second similarities, a gesture matching intention score is calculated, where the first similarity is positively correlated with the gesture matching intention score, and the second similarity is negatively correlated with the gesture matching intention score. This comprehensive calculation method ensures that the score not only reflects the similarity with the command gesture but also considers the difference with the background movement.

[0149] This multi-dimensional, context-aware intent judgment mechanism significantly improves the accuracy and reliability of radar body detectors in determining whether a user's gesture is a mode switching command. By more accurately identifying intentional gestures and effectively suppressing interference from unconscious background movements, the method in this application overcomes the misjudgment problem caused by accidental feature similarity in existing technologies, thereby improving the accuracy of gesture interaction and user experience, and ensuring the continuity of health monitoring.

[0150] In some preferred embodiments, step S5 includes:

[0151] S51. Compare the intent judgment score with the preset instruction confirmation threshold to determine whether the potential gesture signal meets the conditions for the mode switching instruction.

[0152] S52. If the potential gesture signal meets the conditions of the mode switching instruction, then the potential gesture signal is determined as the mode switching instruction, and the corresponding mode switching operation is executed.

[0153] Specifically, step S51 compares the intent judgment score with a pre-set instruction confirmation threshold, providing the system with a clear decision-making basis and avoiding ambiguous judgments caused by a lack of standards in complex environments. In this way, the system can effectively distinguish unconscious background movements that are accidentally similar to preset gestures. Step S52 explicitly confirms a potential gesture signal as a mode switching instruction when it meets the conditions for such an instruction. Once the instruction is confirmed, the system immediately executes the corresponding mode switching operation, such as pausing the radar body detector's monitoring function or switching it to active mode. This series of operations ensures that only gestures that truly match the user's intent and have high confidence are responded to by the system, thus avoiding the misidentification of non-instructional actions as mode switching instructions and effectively solving the problem of user experience and monitoring continuity being affected by misjudgments in existing technologies. By introducing clear quantitative judgment standards and a strict confirmation mechanism, this solution significantly improves the accuracy and reliability of gesture interaction.

[0154] Secondly, please refer to Figure 2 Some embodiments of this application also provide a radar body detector gesture mode switching system for determining whether a user's gesture is a mode switching command. The system includes:

[0155] The acquisition module 201 is used to acquire radar echo signals and extract the user's motion parameter information based on the radar echo signals.

[0156] The sequence generation module 202 is used to generate a sequence of motion state indicators with a first preset window length arranged in a temporal order based on motion parameter information.

[0157] The sequence analysis module 203 is used to analyze the motion state index sequence, detect signal segments whose similarity to the radar feature pattern of the preset gesture is greater than a first preset threshold, and mark the signal segments as potential gesture signals.

[0158] The intent integration module 204 is used to analyze potential gesture signals based on contextual intent judgment to obtain an intent judgment score;

[0159] The intent determination module 205 is used to determine whether a potential gesture signal is a mode switching instruction based on the intent determination score.

[0160] The system of this application combines gesture feature pattern matching with a context-based intent judgment mechanism to effectively distinguish between conscious mode switching commands and unconscious background movements, and adapts to gesture variations caused by changes in user position and posture, thereby significantly reducing the misjudgment rate and improving the accuracy and reliability of gesture interaction.

[0161] Furthermore, the units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0162] Furthermore, the functional modules in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0163] In this document, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, without necessarily requiring or implying any such actual relationship or order between these entities or operations.

[0164] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for switching gesture modes in a radar body detector, used to determine whether a user's gesture is a mode switching command, characterized in that, The method includes the following steps: S1. Acquire radar echo signals and extract user motion parameter information based on the radar echo signals; S2. Based on the motion parameter information, generate a sequence of motion state indicators with a first preset window length arranged in a temporal order; S3. Analyze the motion state index sequence, detect signal segments whose similarity to the radar feature pattern of the preset gesture is greater than a first preset threshold, and mark the signal segments as potential gesture signals; S4. Based on the contextual intent judgment, analyze the potential gesture signals to obtain the intent judgment score; S5. Determine the score based on the intent, and determine whether the potential gesture signal is a mode switching command; Step S4 includes: S41. Based on the motion state index sequence, analyze the motion state index before the occurrence of the potential gesture signal, and determine the pre-gesture intention score of the potential gesture signal. S42. Based on the motion state index sequence, analyze the motion state index after the execution of the potential gesture signal, and determine the subsequent gesture intention score of the potential gesture signal; S43. Extract the motion features of the potential gesture signal, and compare the motion features with preset intentional gesture feature patterns and unconscious background motion feature patterns to determine the gesture matching intention score of the potential gesture signal. S44. Calculate the intent judgment score based on the pre-gesture intent score, the subsequent gesture intent score, and the gesture matching intent score; Step S41 includes: S411. Based on the motion state index sequence, obtain the statistical characteristics of the motion state index of the signal segment before the potential gesture signal appears, and calculate the stability degree before the potential gesture signal appears based on the statistical characteristics and the preset statistical threshold. S412. Calculate the pre-gesture intention score based on the degree of stability before the potential gesture signal appears.

2. The gesture mode switching method for a radar body detector according to claim 1, characterized in that, Step S1 includes: S11. Acquire radar echo signals; S12. Perform multi-target separation processing on the radar echo signal to obtain target echo signals of multiple moving targets; S13. Filter the target echo signal to filter out interference signals that have a similarity to a preset non-user motion feature pattern greater than a second preset threshold. S14. Extract the motion parameter information based on the filtered target echo signal.

3. The gesture mode switching method for a radar body detector according to claim 2, characterized in that, Step S13 includes: S131. Extract motion feature parameters based on the target echo signal; S132. Compare the motion feature parameters with a preset non-user motion feature pattern to determine the similarity between the target echo signal and the non-user motion feature pattern. S133. Determine whether the similarity is greater than the second preset threshold. If the similarity is greater than the second preset threshold, then the target echo signal is determined to be an interference signal and filtered out.

4. The gesture mode switching method for a radar body detector according to claim 1, characterized in that, Step S2 includes: S21. Perform time-domain smoothing on the motion parameter information to obtain a time-domain smoothed sequence of motion parameter information with a time length equal to the length of the first preset window. S22. Based on the second preset window length, the motion parameter information sequence is split and recombined into multiple motion parameter segments; S23. Extract features from each motion parameter segment to obtain one or more of the user's overall motion energy, limb activity amplitude, position change rate and motion trajectory dispersion, as motion state indicators. S24. Arrange the motion state indicators in chronological order to generate the motion state indicator sequence.

5. The gesture mode switching method for a radar body detector according to claim 1, characterized in that, Step S3 includes: S31. Extract the motion state index sequence based on the third preset window length to obtain multiple signal segments; S32. Extract the changing trends and statistical characteristics of motion state indicators for each signal segment; S33. Based on the changing trend and the statistical features, compare the radar feature patterns of the preset gestures to obtain a similarity score; S34. Determine whether the similarity score of each signal segment is greater than the first preset threshold. S35. Mark the signal segments with similarity scores greater than the first preset threshold as the potential gesture signals.

6. The gesture mode switching method for a radar body detector according to claim 5, characterized in that, The radar feature pattern includes multiple reference vectors representing different gestures, composed of reference change trends and reference statistical features. Step S33 includes: S331. Construct a multi-dimensional vector based on the changing trends and statistical characteristics of motion state indicators of each signal segment; S332. Calculate and obtain the initial similarity score based on the spatial distance between the multidimensional vector of each signal segment and the reference vector of each radar feature pattern. S333. Filter and obtain the maximum similarity score corresponding to each signal segment, and use it as the similarity score.

7. The gesture mode switching method for a radar body detector according to claim 1, characterized in that, Step S43 includes: S431. Extract the motion features of the potential gesture signal; S432. The action feature is compared with a preset intentional gesture feature pattern to obtain a first similarity, wherein the first similarity is the similarity between the potential gesture signal and the intentional gesture feature pattern. S433. The action features are compared with a preset unconscious background motion feature pattern to obtain a second similarity, wherein the second similarity is the similarity between the potential gesture signal and the unconscious background motion feature pattern. S434. Calculate the gesture matching intent score based on the first similarity and the second similarity.

8. The gesture mode switching method for a radar body detector according to claim 1, characterized in that, Step S5 includes: S51. Compare the intent judgment score with a preset instruction confirmation threshold to determine whether the potential gesture signal meets the conditions for a mode switching instruction. S52. If the potential gesture signal meets the conditions of the mode switching instruction, then the potential gesture signal is determined as the mode switching instruction, and the corresponding mode switching operation is executed.

9. A radar body detector gesture mode switching system, used to determine whether a user's gesture is a mode switching command, characterized in that, The system is used to perform the radar body detector gesture mode switching method as described in any one of claims 1-8, the system comprising: The acquisition module is used to acquire radar echo signals and extract the user's motion parameter information based on the radar echo signals. The sequence generation module is used to generate a sequence of motion state indicators with a first preset window length arranged in a temporal order based on the motion parameter information. The sequence analysis module is used to analyze the motion state index sequence, detect signal segments whose similarity to the radar feature pattern of a preset gesture is greater than a first preset threshold, and mark the signal segments as potential gesture signals. The intent integration module is used to analyze the potential gesture signals based on contextual intent judgment to obtain an intent judgment score; The intent determination module is used to determine whether the potential gesture signal is a mode switching instruction based on the intent determination score.

Images