Chewing action recognition method, dietary behavior monitoring earphone and storage medium

Abstract

The invention relates to the technical field of diet monitoring, in particular to a chewing action recognition method, a diet behavior monitoring earphone and a storage medium. According to the chewing action recognition method, a target sensing signal needs to be received from a dietary behavior monitoring earphone in a target terminal; wherein the target behavior characteristics are obtained in the dietary behavior monitoring earphone through the following steps: performing vibration sensing on a target object to obtain a sensing data stream, and performing noise frequency band screening on the sensing data stream to obtain a target sensing signal; performing feature extraction on the target sensing signal to obtain a chewing rhythm feature, a chewing force feature and a chewing time value feature of the target object; and performing chewing behavior analysis on the chewing rhythm feature, the chewing force feature and the chewing time value feature to determine a chewing analysis result of the target object. In this way, the chewing action of the monitored object can be accurately recognized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of diet monitoring, in particular to a chewing action recognition method, a diet behavior monitoring earphone and a storage medium. BACKGROUND

[0002] The existing diet monitoring technology has some limitations in processing chewing action recognition. The commonly used amplitude threshold method combined with general frequency domain analysis, although simple to implement, is difficult to stably distinguish chewing from other similar actions, resulting in low recognition accuracy and frequent false triggering, such as misjudging loud speaking as chewing or missing real eating behavior.

[0003] The traditional fixed threshold method has the problem of being not flexible enough. Some existing technologies use absolute numerical judgment, which is difficult to adapt to individual differences of different users and different eating scenes, resulting in poor adaptability and insufficient robustness. At present, how to accurately recognize the chewing action of the monitored object is a problem to be solved in the industry. SUMMARY

[0004] The present application aims to at least solve one of the technical problems existing in the prior art. To this end, the present application proposes a chewing action recognition method, a diet behavior monitoring earphone and a storage medium, which can accurately recognize the chewing action of the monitored object.

[0005] The chewing action recognition method according to the first aspect of the present application is applied to a target terminal, comprising: Receiving a target sensing signal from a diet behavior monitoring earphone; wherein the target sensing signal is obtained in the diet behavior monitoring earphone by the following steps: performing vibration sensing on a target object to obtain a sensing data stream, and excluding noise frequency bands from the sensing data stream to obtain the target sensing signal; Performing feature extraction on the target sensing signal to obtain chewing rhythm features, chewing intensity features and chewing time value features of the target object; Performing chewing behavior analysis on the chewing rhythm features, the chewing intensity features and the chewing time value features to determine a chewing analysis result of the target object.

[0006] According to some embodiments of the present application, the feature extraction on the target sensing signal to obtain the chewing rhythm features, the chewing intensity features and the chewing time value features of the target object comprises: Generating a target vibration sensing frequency spectrum based on the target sensing signal; Performing frequency spectrum dynamic analysis on the target vibration sensing frequency spectrum to determine the chewing rhythm features of the target sensing signal; performing spectral amplitude analysis on the target vibration perception spectrum to determine the chewing intensity feature of the target perception signal; performing spectral time value analysis on the target vibration perception spectrum to determine the chewing time value feature of the target perception signal.

[0007] According to some embodiments of the present application, the performing spectral dynamic analysis on the target vibration perception spectrum to determine the chewing rhythm feature of the target perception signal comprises: dividing the vibration perception spectrum based on a predetermined first division interval to obtain a plurality of dynamic analysis frames; extracting corresponding vibration main frequency data from the plurality of dynamic analysis frames; performing dynamic feature analysis based on the vibration main frequency data of a plurality of consecutive dynamic analysis frames to obtain a main frequency data dynamic feature as the chewing rhythm feature.

[0008] According to some embodiments of the present application, the performing spectral amplitude analysis on the target vibration perception spectrum to determine the chewing intensity feature of the target perception signal comprises: selecting a target spectrum segment from the target vibration perception spectrum; dividing the target spectrum segment based on a predetermined second division interval to obtain a plurality of spectrum sub-segments; performing amplitude analysis on each of the spectrum sub-segments to obtain a corresponding local amplitude; performing amplitude variation calculation on the local amplitudes of a plurality of consecutive spectrum sub-segments to obtain an amplitude variation coefficient; generating the chewing intensity feature based on the amplitude variation coefficient and the local amplitudes of the spectrum sub-segments.

[0009] According to some embodiments of the present application, the performing spectral time value analysis on the target vibration perception spectrum to determine the chewing time value feature of the target perception signal comprises: selecting a target spectrum segment from the target vibration perception spectrum; dividing the target spectrum segment into a chewing occurrence sub-segment and a chewing stop sub-segment based on a preset chewing spectrum discrimination condition; performing time value analysis according to the distribution features of the chewing occurrence sub-segment and the chewing stop sub-segment in the target spectrum segment to determine the chewing time value feature of the target perception signal.

[0010] According to some embodiments of the present application, the performing chewing behavior analysis on the chewing rhythm feature, the chewing intensity feature, and the chewing time value feature to determine the chewing analysis result of the target object comprises: perform rhythm index analysis based on the chewing rhythm feature to obtain an average chewing frequency and a chewing frequency stability; perform force index analysis based on the chewing force feature to obtain an average chewing energy intensity and a coefficient of variation in amplitude; perform time value index analysis based on the chewing time value feature to obtain a total chewing time, a proportion of effective chewing time, and a number of chewing interruptions; perform feature fusion processing on the average chewing frequency, the chewing frequency stability, the average chewing energy intensity, the coefficient of variation in amplitude, the total chewing time, the proportion of effective chewing time, and the number of chewing interruptions to obtain a target chewing feature vector; perform chewing mode recognition based on the target chewing feature vector to determine the chewing analysis result of the target object.

[0011] According to some embodiments of the present application, the chewing mode recognition based on the target chewing feature vector to determine the chewing analysis result of the target object comprises: obtain a diet behavior mode library; wherein the diet behavior mode library stores corresponding chewing mode representation vectors for various diet behavior modes; compare the target chewing feature vector with each chewing mode representation vector to select a hit mode representation vector from the diet behavior mode library; generate the chewing analysis result based on the diet behavior mode corresponding to the hit mode representation vector.

[0012] According to the chewing action recognition method of the second aspect embodiment of the present application, applied to a diet behavior monitoring earphone, comprising: perform vibration sensing on a target object to obtain a sensing data stream; perform noise frequency band screening on the sensing data stream to obtain a target sensing signal; perform feature extraction on the target sensing signal to obtain a target behavior feature of the target object; send the target behavior feature to a target terminal to perform feature extraction on the target sensing signal through the target terminal to obtain a chewing rhythm feature, a chewing force feature, and a chewing time value feature of the target object, and perform chewing behavior analysis on the chewing rhythm feature, the chewing force feature, and the chewing time value feature to determine a chewing analysis result of the target object.

[0013] In a third aspect, an embodiment of the present application provides a dietary behavior monitoring earphone, comprising a memory and a processor, the memory stores a computer program, and the processor implements the chewing action recognition method according to any one of the embodiments of the first aspect of the present application when executing the computer program.

[0014] In a fourth aspect, an embodiment of the present application provides a computer readable storage medium, the storage medium stores a program, and the program is executed by a processor to implement the chewing action recognition method according to any one of the embodiments of the first aspect of the present application.

[0015] According to the chewing action recognition method, the dietary behavior monitoring earphone and the storage medium provided by the embodiments of the present application, at least the following beneficial effects are achieved: The chewing action recognition method provided by the present application needs to receive a target sensing signal from the dietary behavior monitoring earphone in the target terminal; wherein, the target behavior feature is obtained in the dietary behavior monitoring earphone by the following steps: vibration sensing is performed on the target object to obtain a sensing data stream, noise frequency bands are screened out for the sensing data stream to obtain the target sensing signal; feature extraction is performed on the target sensing signal to obtain the chewing rhythm feature, the chewing force feature and the chewing time value feature of the target object; and chewing behavior analysis is performed on the chewing rhythm feature, the chewing force feature and the chewing time value feature to determine the chewing analysis result of the target object. In this way, the chewing action of the monitoring object can be accurately identified.

[0016] Additional aspects and advantages of the present application will be in part apparent and in part pointed out hereinafter in the description of embodiments of the present application. BRIEF DESCRIPTION OF DRAWINGS

[0017] The above and / or additional aspects and advantages of the present application will become apparent and be readily appreciated from the following description of embodiments, taken in conjunction with the accompanying drawings, in which: Figure 1 A flowchart of a chewing action recognition method provided by an embodiment of the present application is shown; Figure 2 Another flowchart of a chewing action recognition method provided by an embodiment of the present application is shown; Figure 3 Another flowchart of a chewing action recognition method provided by an embodiment of the present application is shown; Figure 4A The change of the angular velocity data of the target sensing data of all axes over time is shown; Figure 4B The change of the angular velocity data of the target sensing data on the X-axis over time is shown; Figure 4C The change of the angular velocity data of the target sensing data on the Y-axis over time is shown; Figure 4D The angular velocity data of the target perception data on the Z-axis over time is shown; Figure 4E Another flowchart of the chewing action recognition method provided by the embodiment of the present application is shown; Figure 5 Another flowchart of the chewing action recognition method provided by the embodiment of the present application is shown; Figure 6 Another flowchart of the chewing action recognition method provided by the embodiment of the present application is shown; Figure 7 Another flowchart of the chewing action recognition method provided by the embodiment of the present application is shown; Figure 8 A hardware structure schematic diagram of the diet behavior monitoring earphone provided by the embodiment of the present application is shown; Figure 9 Another hardware structure schematic diagram of the diet behavior monitoring earphone provided by the embodiment of the present application is shown. DETAILED DESCRIPTION

[0018] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the drawings, wherein the same or similar notations represent the same or similar elements or elements with the same or similar functions throughout. The embodiments described below by referring to the drawings are exemplary and are only used to explain the present application, and cannot be understood as a limitation of the present application.

[0019] In the description of the present application, several means one or more, the meaning of multiple is two or more, greater than, less than, more than, etc. are understood as not including the number, above, below, within, etc. are understood as including the number. If it is described that the first, the second is only used to distinguish the technical features for the purpose, and cannot be understood as indicating or implying the relative importance or implicitly indicating the number of indicated technical features or implicitly indicating the sequence of indicated technical features.

[0020] In the description of the present application, it is understood that the orientation description, such as up, down, left, right, front, back, etc. indicates the orientation or position relationship based on the orientation or position relationship shown in the drawings, which is only for the convenience of describing the present application and simplifying the description, and is not intended to indicate or imply that the device or element indicated must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation of the present application.

[0021] In the description of the specification, the description of the terms "one embodiment", "some embodiments", "exemplary embodiment", "example", "specific example", or "some examples" and the like means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the specification, the exemplary description of the above terms does not necessarily mean the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0022] In the description of the present application, it should be noted that, unless otherwise explicitly limited, the words such as setting, installing, connecting and the like should be understood in a broad sense, and those skilled in the art can reasonably determine the specific meaning of the above words in the present application in combination with the specific content of the technical solution. In addition, the identification of the specific steps in the following does not represent the limitation of the order and execution logic of the steps, and the execution order and execution logic between the steps should be understood and inferred with reference to the content expressed in the embodiments.

[0023] The existing diet monitoring technology has some limitations in processing chewing action recognition. The traditional method mainly relies on inertial sensor to monitor large amplitude body activity, and its algorithm design is aimed at walking, running and other signals with strong signals and simple patterns, but chewing is a small action of the head, and the signal characteristics are completely different. This mismatch causes the existing scheme to be difficult to accurately capture the chewing signal, and it is more difficult to effectively cope with the interference brought by daily head activities such as speaking, nodding and shaking.

[0024] The amplitude threshold method commonly used at present combined with general frequency domain analysis, although simple to implement, is difficult to stably distinguish chewing from other similar actions, resulting in low recognition accuracy and frequent false triggering, for example, misjudging loud speaking as chewing, or missing the real eating behavior. The traditional fixed threshold method has the problem of being not flexible enough. Some existing technologies use absolute numerical judgment, which is difficult to adapt to individual differences of different users and different eating scenes, resulting in poor adaptability and insufficient robustness.

[0025] At present, how to accurately identify the chewing action of the monitored object is a problem to be solved in the industry.

[0026] The present application aims to at least solve one of the technical problems existing in the prior art. To this end, the present application proposes a chewing action recognition method, a diet behavior monitoring earphone and a storage medium, which can accurately identify the chewing action of the monitored object.

[0027] Reference Figure 1 According to the chewing action recognition method of the present application, the diet behavior monitoring earphone can include: Step S101, vibration sensing is performed on the target object to obtain a sensing data stream; Step S102, noise frequency bands are screened out for the sensing data stream to obtain a target sensing signal; Step S103, feature extraction is performed on the target sensing signal to obtain a target behavior feature of the target object; Step S104, the target behavior feature is sent to a target terminal, so that feature extraction is performed on the target sensing signal by the target terminal to obtain a chewing rhythm feature, a chewing strength feature, and a chewing time value feature of the target object, and chewing behavior analysis is performed on the chewing rhythm feature, the chewing strength feature, and the chewing time value feature to determine a chewing analysis result of the target object.

[0028] According to the chewing action recognition method of the embodiments of the present application, the chewing action recognition method is applied to a target terminal and can include the following steps: Step S105, the target sensing signal is received from the diet behavior monitoring earphone, and feature extraction is performed on the target sensing signal to obtain a chewing rhythm feature, a chewing strength feature, and a chewing time value feature of the target object; Step S106, chewing behavior analysis is performed on the chewing rhythm feature, the chewing strength feature, and the chewing time value feature to determine a chewing analysis result of the target object.

[0029] The chewing action recognition method of the embodiments of the present application aims to realize accurate recognition of chewing actions through a diet behavior monitoring earphone, thereby solving the problems of low recognition accuracy and insufficient adaptability in the prior art.

[0030] In some embodiments, step S101, vibration sensing is performed on the target object to obtain a sensing data stream; It should be noted that the diet behavior monitoring earphone is used to first perform vibration sensing on the target object (i.e., the user) to obtain an original sensing data stream. This process uses the high-sensitivity sensor built into the earphone to capture the tiny vibration signals in the ear region, and these signals contain the mechanical vibration features generated by the chewing action. Since the earphone is in close contact with the human body, it can effectively filter out environmental noise, thereby preliminarily ensuring the purity and stability of the signals. This initial step lays a foundation for subsequent signal processing and ensures the quality of the input data.

[0031] In some embodiments, step S102, noise frequency bands are screened out for the sensing data stream to obtain a target sensing signal; It should be noted that next, the embodiments of the present application screen out the noise frequency band from the perception data stream to obtain the target perception signal. This step is the key to solving the noise interference problem in traditional technology. By analyzing the spectral characteristics of the perception data stream, those frequency bands irrelevant to the chewing action are identified and removed, such as electronic noise in the high frequency band or background vibration in the low frequency band. This frequency band screening method can effectively reduce the false trigger rate and avoid misjudging the vibration caused by loud talking and other non-chewing actions as chewing signals, thereby improving the quality and reliability of the signal. This process not only improves the signal-to-noise ratio of the signal, but also provides a clearer signal basis for subsequent feature extraction.

[0032] In step S103 of some embodiments, feature extraction is performed on the target perception signal to obtain target behavior features of the target object; It should be noted that feature extraction is performed on the target perception signal to obtain target behavior features of the target object. This step is the core of the entire method, which extracts features related to chewing behavior from the signal to provide key information for subsequent analysis. The feature extraction process not only includes simple amplitude analysis, but also covers more complex frequency and time characteristic analysis, such as chewing rhythm, chewing intensity and chewing duration features. These features can fully reflect the essential characteristics of the chewing action and lay a foundation for subsequent accurate recognition. For example, the chewing rhythm feature can be determined by analyzing the periodic changes of the signal, the chewing intensity feature is measured by the amplitude change of the signal, and the chewing duration feature is evaluated by the duration of the signal.

[0033] In step S104 of some embodiments, the target behavior features are sent to the target terminal to perform feature extraction on the target perception signal through the target terminal to obtain chewing rhythm features, chewing intensity features and chewing duration features of the target object, and perform chewing behavior analysis on the chewing rhythm features, chewing intensity features and chewing duration features to determine the chewing analysis result of the target object; It should be noted that the extracted target behavior features are sent to the target terminal. After receiving these features, the target terminal further performs chewing behavior analysis to determine the chewing analysis result of the target object. The terminal device has stronger computing and data analysis capabilities and can perform deep processing on the extracted features, including but not limited to pattern recognition, classification analysis and health assessment, etc. Through the analysis of the terminal device, the embodiments of the present application can accurately judge whether the user's chewing behavior is normal or whether there are bad habits, such as chewing too fast, chewing insufficiently or eating discontinuously, etc. This process not only improves the accuracy of recognition, but also can be self-adapted according to different users and scenes, providing strong support for diet behavior monitoring and health management.

[0034] Reference Figure 2According to some embodiments of the present application, in step S104, the chewing behavior analysis is performed on the chewing rhythm feature, the chewing force feature, and the chewing time value feature to determine the chewing analysis result of the target object, which can include: In step S201, rhythm index analysis is performed based on the chewing rhythm feature to obtain the average chewing frequency and the chewing frequency stability; In step S202, force index analysis is performed based on the chewing force feature to obtain the average chewing energy intensity and the amplitude variation coefficient; In step S203, time value index analysis is performed based on the chewing time value feature to obtain the total chewing time, the effective chewing time proportion, and the number of chewing interruptions; In step S204, feature fusion processing is performed on the average chewing frequency, the chewing frequency stability, the average chewing energy intensity, the amplitude variation coefficient, the total chewing time, the effective chewing time proportion, and the number of chewing interruptions to obtain a target chewing feature vector; In step S205, chewing mode recognition is performed based on the target chewing feature vector to determine the chewing analysis result of the target object.

[0035] In the embodiments of the present application, the core step of the chewing behavior analysis is to deeply analyze and fuse the extracted chewing rhythm feature, chewing force feature, and chewing time value feature, thereby determining the chewing analysis result of the target object.

[0036] In step S201 of some embodiments, rhythm index analysis is performed based on the chewing rhythm feature to obtain the average chewing frequency and the chewing frequency stability; It should be noted that from the rhythm index analysis, the average chewing frequency and the chewing frequency stability are calculated based on the chewing rhythm feature in the embodiments of the present application. The average chewing frequency reflects the number of chewing times per minute of the user, which is a key indicator for measuring eating speed; and the chewing frequency stability describes the regularity of the chewing action, and a stable frequency usually means a more uniform chewing process. These two indicators provide important information about the chewing rhythm for subsequent behavior analysis.

[0037] In step S202 of some embodiments, force index analysis is performed based on the chewing force feature to obtain the average chewing energy intensity and the amplitude variation coefficient; It should be noted that the average chewing energy intensity and the amplitude variation coefficient are obtained by analyzing the chewing force feature in the embodiments of the present application. The average chewing energy intensity quantifies the force degree of the user during chewing, reflecting the intensity of the chewing action; and the amplitude variation coefficient measures the uniformity of the chewing force, and a lower variation coefficient indicates that the user uses force more consistently during the chewing process. These two indicators can reveal whether the user exerts excessive or insufficient force during chewing, and whether there is a problem of uneven force, providing strong support for evaluating the sufficiency and health of chewing.

[0038] In step S203 of some embodiments, time value indicators are analyzed based on the chewing time value characteristics to obtain the total chewing time, the effective chewing time proportion, and the number of chewing interruptions. It should be noted that the embodiments of the present application analyze the chewing time value characteristics to obtain the total chewing time, the effective chewing time proportion, and the number of chewing interruptions. The total chewing time directly reflects the total time required by the user to complete a meal; the effective chewing time proportion shows the proportion of time actually used for chewing by the user, and the higher this proportion, the more focused the eating process is; and the number of chewing interruptions records the number of pauses in the eating process due to drinking water, talking, or distraction. These time value indicators can comprehensively reflect the eating continuity and concentration of the user, providing an important basis for identifying unhealthy eating habits.

[0039] In step S204 of some embodiments, feature fusion processing is performed on the average chewing frequency, the chewing frequency stability, the average chewing energy intensity, the amplitude coefficient of variation, the total chewing time, the effective chewing time proportion, and the number of chewing interruptions to obtain a target chewing feature vector. It should be noted that after obtaining the above-mentioned multiple key indicators, the embodiments of the present application perform feature fusion processing to integrate the average chewing frequency, the chewing frequency stability, the average chewing energy intensity, the amplitude coefficient of variation, the total chewing time, the effective chewing time proportion, and the number of chewing interruptions into a target chewing feature vector. This fusion process is not simply a data stacking, but through a specific algorithm or model, the features of different dimensions are weighted and normalized to generate a feature vector that can comprehensively describe the user's chewing behavior. This vector integrates the information of rhythm, intensity, and time value, providing a unified input format for subsequent chewing pattern recognition.

[0040] In step S205 of some embodiments, chewing pattern recognition is performed based on the target chewing feature vector to determine the chewing analysis result of the target object.

[0041] It should be noted that based on the target chewing feature vector, the embodiments of the present application map the feature vector to a specific chewing behavior pattern through a pre-set classification model or rule engine. This process can identify whether the user has unhealthy chewing habits, such as fast eating, insufficient chewing, uneven force, or discontinuous eating, etc. Through pattern recognition, the embodiments of the present application can provide accurate analysis results and personalized improvement suggestions for the user, helping the user to optimize eating behavior and improve health level. The entire analysis process from multi-dimensional feature analysis to fusion and then to pattern recognition forms a complete technical chain, ensuring the accuracy and reliability of the analysis results.

[0042] Reference Figure 3According to some embodiments of the present application, the step S205 of performing chewing pattern recognition based on the target chewing feature vector to determine the chewing analysis result of the target object can include: In step S301, a dietary behavior pattern library is obtained, wherein the dietary behavior pattern library stores corresponding chewing pattern representation vectors for various dietary behavior patterns. In step S302, the target chewing feature vector is compared with each chewing pattern representation vector to select a hit pattern representation vector from the dietary behavior pattern library. In step S303, a chewing analysis result is generated based on the dietary behavior pattern corresponding to the hit pattern representation vector.

[0043] In some embodiments of the present application, the chewing pattern recognition is performed based on the target chewing feature vector, and this process is a key link for determining the chewing analysis result of the target object.

[0044] In step S301 of some embodiments, a dietary behavior pattern library is obtained, wherein the dietary behavior pattern library stores corresponding chewing pattern representation vectors for various dietary behavior patterns. It should be noted that the embodiments of the present application need to obtain a dietary behavior pattern library, and this pattern library is the basis of the entire recognition process. The dietary behavior pattern library stores chewing pattern representation vectors corresponding to various dietary behavior patterns. These representation vectors are obtained by analyzing and summarizing a large amount of sample data, and they can accurately reflect the chewing characteristics under different dietary behavior patterns. For example, the representation vector of the fast eating pattern may have a high average chewing frequency and a low effective chewing time ratio; and the representation vector of the insufficient chewing pattern may have a high amplitude coefficient of variation and a short total chewing time. These representation vectors provide standards and references for subsequent pattern comparison.

[0045] In step S302 of some embodiments, the target chewing feature vector is compared with each chewing pattern representation vector to select a hit pattern representation vector from the dietary behavior pattern library. It should be noted that after obtaining the dietary behavior pattern library, the target chewing feature vector is compared with each chewing pattern representation vector in the library. This comparison process is the core of recognition, which determines the most matching pattern by calculating the similarity or distance between the target feature vector and each representation vector in the library. For example, Euclidean distance, cosine similarity or other statistical methods can be used to quantify the difference between two vectors. The target vector is compared with each representation vector in the library one by one to find the vector with the smallest difference or the highest similarity. This process not only needs to consider the matching degree of a single feature, but also needs to comprehensively evaluate the overall consistency between multiple features. For example, even if a certain feature is very close to a certain pattern in the library, if other features have a large difference from the representation vector of the pattern, this pattern may not be selected as the hit pattern. Through this comprehensive comparison, the application embodiment can select the hit pattern representation vector that best matches the target chewing feature vector from the dietary behavior pattern library.

[0046] In step S303 of some embodiments, a chewing analysis result is generated based on the dietary behavior pattern corresponding to the hit pattern representation vector.

[0047] It should be noted that based on the dietary behavior pattern corresponding to the hit pattern representation vector, the application embodiment generates a chewing analysis result. This process is to convert the recognized pattern into a specific analysis conclusion, providing detailed information about the user's chewing behavior. For example, if the dietary behavior pattern corresponding to the hit pattern representation vector is "fast eating and insufficient chewing", the analysis result will clearly indicate that the user has the problem of fast eating and insufficient chewing during eating. This result not only includes a description of the current chewing behavior, but can also further provide health suggestions, such as suggesting the user to slow down the eating speed, increase the number of chewing, etc. Through this process from feature vector to pattern comparison to result generation, the application embodiment can provide accurate and instructive chewing analysis results for users, helping them understand their eating habits and take appropriate improvement measures.

[0048] It should be understood that the entire chewing pattern recognition process of the application is a systematic analysis chain, from obtaining the dietary behavior pattern library to feature vector comparison, and finally to the generation of the analysis result. Each step is closely connected and dependent on each other. This process not only solves the problem of low recognition accuracy in the prior art, but also provides personalized dietary behavior evaluation and suggestions for users through detailed feature analysis and pattern matching, thereby achieving technical progress in the field of dietary monitoring.

[0049] In step S105 of some embodiments, a target perception signal is received from a dietary behavior monitoring earphone, and feature extraction is performed on the target perception signal to obtain chewing rhythm features, chewing force features and chewing time value features of a target object. It is necessary to note that the feature extraction of the target perception signal to obtain the chewing rhythm feature, the chewing force feature and the chewing time value feature is a key link in the whole chewing action recognition and analysis process. This process not only provides a precise data basis for subsequent behavior analysis, but also improves the accuracy and adaptability of the recognition system, so that it can effectively cope with complex and variable actual use scenarios.

[0050] It is worth noting that the feature extraction of the target perception signal to obtain the chewing rhythm feature, the chewing force feature and the chewing time value feature is a key link in the whole chewing action recognition and analysis process. This process not only provides a precise data basis for subsequent behavior analysis, but also improves the accuracy and adaptability of the recognition system, so that it can effectively cope with complex and variable actual use scenarios.

[0051] Firstly, the extraction of the chewing rhythm feature can reflect the frequency and regularity of the user's chewing action. By analyzing the periodic changes in the perception signal, the embodiments of the present application can determine the number of chews per minute and the stability of the chewing action. For example, a stable chewing rhythm usually means that the user is more focused and chewed fully during eating, while frequent rhythm changes may indicate that the user is distracted or unevenly chewed during eating. The extraction of such rhythm features provides a direct basis for identifying unhealthy habits such as fast eating and insufficient chewing, and also lays a foundation for subsequent behavior analysis and health assessment.

[0052] Secondly, the extraction of the chewing force feature can reveal the force used by the user during chewing. By analyzing the amplitude changes of the perception signal, the embodiments of the present application can quantify the force of each chew and the uniformity of force changes. The strength of the chewing force directly affects the grinding effect of food, and thus affects the burden on the digestive system. For example, insufficient force may result in insufficient grinding of food, increasing the difficulty of gastrointestinal digestion; while excessive force may cause unnecessary pressure on the teeth and jaw joints. By extracting the chewing force feature, the embodiments of the present application can identify potential problems such as uneven or excessive force, and provide targeted improvement suggestions to help users establish healthy chewing habits.

[0053] Finally, the extraction of chewing time value features can comprehensively reflect the time characteristics of the user's eating process. By analyzing the duration of the perception signal, the embodiments of the present application can determine the total duration of a single meal, the proportion of effective chewing time, and the number of interruptions during the eating process. These parameters can intuitively reflect the user's eating speed, eating continuity, and eating concentration. For example, a shorter total duration may indicate that the user eats too fast, and frequent interruptions may suggest that the user is distracted or frequently drinks water during the eating process. By analyzing these time value features, the embodiments of the present application can identify unhealthy habits such as eating discontinuity and eating too fast, and provide personalized intervention measures for the user to help them adjust their eating rhythm and improve their eating quality.

[0054] In summary, the extraction of chewing rhythm features, chewing intensity features, and chewing time value features provides multi-dimensional data support for the recognition and analysis of chewing actions. These features not only accurately describe the user's chewing behavior, but also reveal potential unhealthy habits and their impact on health. Through comprehensive analysis of these features, the embodiments of the present application can provide personalized dietary recommendations and health intervention measures for the user, thereby effectively promoting the user to establish healthy eating habits and improve overall health. This process fully embodies the importance of fine feature extraction in complex behavior recognition and provides technical support for the development of dietary behavior monitoring technology.

[0055] Referring to Figure 4A , the target perception data is shown in all axes. Figure 4A The gyroscopic three-axis angular velocity data over time is reflected, which can be used to analyze chewing rhythm features, chewing intensity features, and chewing time value features.

[0056] For the analysis of chewing rhythm features, it can be observed that the Z-axis angular velocity in the figure shows periodic fluctuations during chewing. The frequency of these fluctuations can reflect the rhythm of chewing, i.e., the number of chews per unit time. By calculating the average period of these fluctuations, the average chewing frequency can be obtained, which is a key indicator of eating speed. At the same time, by analyzing the consistency of these fluctuations, the stability of the chewing frequency can be evaluated, i.e., whether the chewing action is uniform.

[0057] In terms of chewing intensity features, the amplitude changes of the Z-axis angular velocity in the figure can provide relevant information. During chewing, the amplitude of the angular velocity can reflect the intensity of chewing, with a larger amplitude generally indicating a larger chewing intensity, and a smaller amplitude possibly indicating a smaller chewing intensity. By calculating the average value and coefficient of variation of these amplitudes, the average level and consistency of the chewing intensity can be evaluated, thereby identifying whether the chewing intensity is uniform and whether there are problems of excessive or insufficient intensity.

[0058] As for the time value characteristics during chewing, they can be determined by analyzing the time periods when chewing occurs and stops. By identifying the distribution of chewing actions on the time axis, parameters such as the total duration of chewing, the proportion of effective chewing time, and the number of interruptions during chewing can be calculated. These parameters can reflect the continuity and concentration of eating, for example, the total duration can reflect the duration of a single meal, the proportion of effective chewing time can reflect the proportion of chewing in the eating process, and the number of interruptions during chewing can reveal the frequency of pauses during eating.

[0059] Referring to Figure 4B , it specifically shows the change of angular velocity data of the target perception data on the X-axis over time. X-axis angular velocity data analysis: the graph shows that the angular velocity fluctuates slightly throughout the time period, with small amplitude changes. This may mean that the movement of the head in the X-axis direction is less or more stable. If the X-axis represents horizontal movement, this may indicate that the horizontal shaking of the head during eating is small, which may not be related to chewing movements, or it may indicate that the head remains relatively stable during eating.

[0060] Referring to Figure 4C , it specifically shows the change of angular velocity data of the target perception data on the Y-axis over time. Y-axis angular velocity data analysis: similar to the X-axis, the angular velocity of the Y-axis also changes relatively smoothly with small fluctuations. This may indicate that the movement in the Y-axis direction, which may be the vertical direction or another horizontal direction, is also less. This stability may help more accurately analyze the Z-axis data related to chewing, as it reduces potential interference from movements in other directions on the Z-axis data.

[0061] Referring to Figure 4D , it specifically shows the change of angular velocity data of the target perception data on the Z-axis over time. Z-axis angular velocity data analysis: these data show obvious fluctuations during chewing, which is crucial for analyzing chewing characteristics. The Z-axis has large fluctuations, and there are obvious peaks in some time periods, which may correspond to chewing movements. By analyzing the frequency of these peaks, the rhythm characteristics of chewing can be determined, such as the number of chews per minute. At the same time, the size of the peaks can reflect the force characteristics of chewing, i.e. the degree of force during chewing. In addition, by observing the duration and interval of the Z-axis angular velocity, the time value characteristics of chewing can be analyzed, such as the total duration of chewing and the number of interruptions.

[0062] Notably, the apparent fluctuations in the Z-axis data suggest that movements in this axis are closely related to chewing behavior, making the Z-axis data particularly important for analyzing chewing rhythm, intensity, and duration characteristics. The data from the X-axis and Y-axis also provide valuable information and are crucial for a comprehensive understanding of head movements and assessing the accuracy of eating behavior. While the fluctuations in the X-axis and Y-axis data are less pronounced, they provide information about the movement of the head in the horizontal and possibly vertical directions. These data help analyze the comprehensiveness of head movements, ensuring that all relevant movement dimensions are considered when assessing chewing behavior. When analyzing Z-axis data, the data from the X-axis and Y-axis can help identify and exclude non-chewing-related movement interference. For example, if the data from the X-axis or Y-axis shows abnormal fluctuations during a certain time period, it may indicate that the Z-axis fluctuations during that time period are not entirely caused by chewing but are influenced by other head movements. Additionally, the data from the X-axis and Y-axis can assist in identifying specific behavioral patterns. For example, slight shaking or tilting of the head may be associated with certain specific actions during eating, such as tilting the head to better chew food. The identification of these patterns can increase the depth of understanding of eating behavior. Based on this, the integrity of multi-axis data is very important when performing any type of movement analysis. In the case where data from all axes are considered, the accuracy and reliability of the analysis results can be further improved.

[0063] It should be understood that by integrating the data from the three axes, chewing behavior can be more accurately identified and analyzed. The stability of the X-axis and Y-axis helps confirm that fluctuations in the Z-axis are mainly caused by chewing. Through detailed analysis of the Z-axis data, chewing rhythm, intensity, and duration characteristics can be extracted, which are important for understanding eating behavior and identifying possible unhealthy eating habits. For example, if the Z-axis data shows frequent high-intensity fluctuations, it may indicate a habit of rapid and forceful chewing, which can put pressure on the digestive system. Conversely, if the fluctuations are relatively smooth and long in duration, it may indicate slow and sufficient chewing, which is generally considered a healthier eating habit.

[0064] Reference Figure 4E According to some embodiments of the present application, step S105, which performs feature extraction on the target perception signal to obtain the chewing rhythm characteristics, chewing intensity characteristics, and chewing duration characteristics of the target object, can include: Step S401, generating a target vibration perception spectrum based on the target perception signal; Step S402, performing spectral dynamic analysis on the target vibration perception spectrum to determine the chewing rhythm characteristics of the target perception signal; Step S403, performing spectral amplitude analysis on the target vibration perception spectrum to determine the chewing intensity characteristics of the target perception signal; Step S404, performing spectrum time value analysis on the target vibration perception spectrum to determine the chewing time value feature of the target perception signal.

[0065] In some embodiments of the present application, the feature extraction process for the target perception signal is a key link of the chewing action recognition. Through this process, the chewing rhythm feature, chewing intensity feature and chewing time value feature of the target object can be obtained.

[0066] Step S401 of some embodiments, based on the target perception signal, generates a target vibration perception spectrum; It should be noted that the present application starts from generating a target vibration perception spectrum, which is the result of frequency domain conversion of the original perception signal. By converting the time domain perception signal into a spectrum, the present application can more clearly identify different frequency components in the signal and their corresponding energy distribution. This spectrum conversion provides a basis for subsequent feature extraction, enabling the system to separate specific frequency features related to chewing action from complex signals.

[0067] Step S402 of some embodiments, performing spectrum dynamic analysis on the target vibration perception spectrum to determine the chewing rhythm feature of the target perception signal; It should be noted that the present application performs spectrum dynamic analysis on the target vibration perception spectrum to determine the chewing rhythm feature of the target perception signal. This analysis process mainly focuses on the variation of energy in the spectrum over time, especially those dynamic changes related to the periodicity of chewing action. For example, chewing action produces vibrations with a certain frequency, and this frequency corresponds to the opening and closing speed of the lower jaw. By analyzing the spectrum dynamics, the present application can identify this periodic change and quantify it as a chewing rhythm feature, such as the average chewing frequency and the stability of the chewing frequency. These features reflect the speed and regularity of the user's chewing and are important indicators for assessing the health of chewing behavior.

[0068] Reference Figure 5 According to some embodiments of the present application, step S402 performs spectrum dynamic analysis on the target vibration perception spectrum to determine the chewing rhythm feature of the target perception signal, which can include: Step S501, dividing the vibration perception spectrum based on a pre-determined first division interval to obtain a plurality of dynamic analysis frames; Step S502, extracting corresponding vibration main frequency data from the plurality of dynamic analysis frames; Step S503, performing dynamic feature analysis based on the vibration main frequency data of the plurality of consecutive dynamic analysis frames to obtain a main frequency data dynamic feature as the chewing rhythm feature.

[0069] In some embodiments of the present application, the process of performing spectral dynamic analysis on the target vibration perception spectrum to determine the chewing rhythm features is a fine signal processing step.

[0070] In step S501 of some embodiments, the vibration perception spectrum is divided based on a predetermined first division interval to obtain a plurality of dynamic analysis frames. It should be noted that the vibration perception spectrum is divided based on a predetermined first division interval. This division method divides the continuous spectrum data into a plurality of dynamic analysis frames with fixed time length. The selection of the first division interval is crucial because it directly determines the degree of signal details that each analysis frame can capture. If the interval is too long, the rapid changes of the chewing action may be missed; if the interval is too short, excessive noise will be introduced, increasing the computational complexity. Therefore, the determination of the first division interval needs to match the typical frequency characteristics of the chewing action, so that each dynamic analysis frame can contain sufficient information to reflect the periodic changes of the chewing action.

[0071] In step S502 of some embodiments, corresponding vibration main frequency data is extracted from the plurality of dynamic analysis frames. It should be noted that after dividing the vibration perception spectrum into a plurality of dynamic analysis frames, the corresponding vibration main frequency data is extracted from each dynamic analysis frame in the embodiments of the present application. The main frequency data refers to the dominant frequency component within each analysis frame, which usually corresponds to the main vibration frequency generated by the chewing action. This frequency component is a direct reflection of the chewing action, because the opening and closing movement of the lower jaw during chewing produces a vibration signal with a certain frequency. By extracting the main frequency data from each dynamic analysis frame, the embodiments of the present application can capture the frequency characteristics of the chewing action at different time points, thereby providing basic data for further dynamic feature analysis.

[0072] In step S503 of some embodiments, dynamic feature analysis is performed based on the vibration main frequency data of a plurality of consecutive dynamic analysis frames to obtain main frequency data dynamic features as chewing rhythm features.

[0073] It should be noted that the embodiments of the present application perform dynamic feature analysis based on the vibration main frequency data of a plurality of consecutive dynamic analysis frames. This process is a time series analysis of the extracted main frequency data to identify the variation of the main frequency data over time. This variation directly reflects the rhythm features of the chewing action, such as the speed and regularity of chewing. By analyzing the dynamic changes of the main frequency data, the embodiments of the present application can calculate the average chewing frequency, i.e. the number of chewing actions per unit time, and the stability of the chewing frequency, i.e. whether the chewing action is uniform and consistent. These dynamic features as chewing rhythm features can provide key information for subsequent behavior analysis, helping to identify whether the user has bad habits such as fast eating and irregular chewing.

[0074] It should be understood that the entire spectrum dynamic analysis process is a systematic signal processing chain, from the division of the spectrum to the extraction of the main frequency data, and then to the analysis of the dynamic characteristics, each step is closely connected and dependent on each other. This process not only solves the problem of difficult stable differentiation of chewing action in the prior art, but also provides precise chewing behavior evaluation and health advice for users through refined feature extraction. Through this step-by-step analysis from spectrum to feature, the embodiments of the present application can effectively identify the rhythm characteristics of the chewing action, thereby achieving technical progress in the field of diet behavior monitoring.

[0075] In step S403 of some embodiments, a spectrum amplitude analysis is performed on the target vibration perception spectrum to determine the chewing intensity characteristics of the target perception signal; It should be noted that the embodiments of the present application perform spectrum amplitude analysis on the target vibration perception spectrum to determine the chewing intensity characteristics of the target perception signal. This analysis process focuses on the energy intensity of different frequency components in the spectrum, especially those in the frequency range related to chewing intensity. For example, a larger chewing intensity will produce a higher energy vibration signal, while a smaller intensity will correspond to a lower energy. By analyzing the spectrum amplitude, the embodiments of the present application can quantify the chewing intensity, such as the average chewing energy intensity and the amplitude variation coefficient. These features not only reflect the force degree of the user when chewing, but also reveal whether the force is uniform, which is crucial for evaluating the sufficiency and health of chewing.

[0076] Reference Figure 6 According to some embodiments of the present application, step S403 of performing spectrum amplitude analysis on the target vibration perception spectrum to determine the chewing intensity characteristics of the target perception signal can include: Step S601, selecting a target spectrum segment from the target vibration perception spectrum; Step S602, dividing the target spectrum segment based on a predetermined second division interval to obtain a plurality of spectrum sub-segments; Step S603, performing amplitude analysis on each spectrum sub-segment to obtain the corresponding local amplitude; Step S604, performing amplitude variation calculation on the local amplitudes of the plurality of consecutive spectrum sub-segments to obtain the amplitude variation coefficient; Step S605, generating the chewing intensity characteristics based on the amplitude variation coefficient and the local amplitudes of the spectrum sub-segments.

[0077] In some embodiments of the present application, the process of performing spectrum amplitude analysis on the target vibration perception spectrum to determine the chewing intensity characteristics is a step-by-step refined signal processing procedure.

[0078] Step S601 of some embodiments selects a target spectrum segment from the target vibration perception spectrum; It should be noted that the target spectrum segment is selected from the target vibration perception spectrum. Since the vibration signal generated by the chewing action has a specific frequency range, the embodiments of the present application need to screen out the spectrum segment directly related to chewing from the entire spectrum. The selection of this segment is based on prior knowledge of the frequency characteristics of the chewing action, ensuring that subsequent analysis is focused on the frequency interval most relevant to chewing behavior, thereby improving the accuracy and efficiency of the analysis.

[0079] Step S602 of some embodiments divides the target spectrum segment based on a predetermined second division interval to obtain a plurality of spectrum sub-segments; It should be noted that the embodiments of the present application divide the segment based on a predetermined second division interval to obtain a plurality of spectrum sub-segments. This division method further subdivides the target spectrum segment into smaller time windows in order to analyze the signals within each sub-segment more carefully. The selection of the second division interval is also crucial, as it needs to be determined according to the typical duration of the chewing action to ensure that each sub-segment can capture the signal characteristics of a complete cycle or part of a cycle of the chewing action. This division method enables the system to monitor the intensity changes of the chewing action more finely.

[0080] Step S603 of some embodiments performs amplitude analysis for each spectrum sub-segment to obtain a corresponding local amplitude; It should be noted that amplitude analysis is performed for each spectrum sub-segment, and the embodiments of the present application calculate the local amplitude within each sub-segment. The local amplitude reflects the intensity of the signal within each sub-segment and is a key indicator of chewing intensity. By analyzing the amplitude of each sub-segment, the embodiments of the present application can capture the intensity changes of the chewing action at different time points. For example, a larger local amplitude may correspond to a more forceful chewing action, while a smaller amplitude may indicate a lighter chewing intensity. This process provides basic data for subsequent amplitude variation calculation, enabling the system to further quantify the variation characteristics of chewing intensity.

[0081] Step S604 of some embodiments performs amplitude variation calculation on the local amplitudes of a plurality of consecutive spectrum sub-segments to obtain an amplitude variation coefficient; It should be noted that the embodiments of the present application calculate the amplitude variation coefficient based on the local amplitudes of multiple consecutive spectral sub-segments. The amplitude variation coefficient is an important indicator for measuring the uniformity of chewing intensity changes. By calculating the ratio of the standard deviation to the average value of the local amplitude of each sub-segment, the embodiments of the present application can quantify the fluctuation degree of chewing intensity in the time series. A lower amplitude variation coefficient indicates that the chewing intensity is more uniform, while a higher variation coefficient may indicate that the chewing intensity changes significantly, for example, uneven force or intermittent excessive force during chewing. This indicator provides an important basis for evaluating the health of chewing behavior.

[0082] In step S605 of some embodiments, a chewing intensity feature is generated based on the amplitude variation coefficient and the local amplitudes of the spectral sub-segments.

[0083] It should be noted that the embodiments of the present application generate a chewing intensity feature based on the amplitude variation coefficient and the local amplitudes of the spectral sub-segments. This feature combines the absolute intensity of chewing intensity (reflected by the local amplitude) and the uniformity of intensity changes (reflected by the amplitude variation coefficient). The chewing intensity feature not only quantifies the force degree of the user during chewing, but also reveals whether the force is uniform, thereby providing comprehensive force-related information for subsequent behavior analysis and health assessment. For example, if the local amplitude is high but the amplitude variation coefficient is low, it may indicate that the user chews with a large force but the force is uniform; while if the local amplitude is low but the amplitude variation coefficient is high, it may indicate that the user chews with a light force and the force is not uniform. Through this comprehensive analysis, the embodiments of the present application can provide detailed evaluation of chewing intensity for the user, helping them optimize their chewing habits and improve their dietary health level.

[0084] It should be understood that the entire spectral amplitude analysis process is a systematic signal processing chain, from the selection of the target spectral segment to the division of the sub-segments, and then to the calculation of the local amplitude and the generation of the amplitude variation coefficient. Each step is closely connected and dependent on each other. This process not only solves the problem of accurately quantifying chewing intensity in the prior art, but also provides precise chewing behavior evaluation and health advice for users through detailed feature extraction. Through this step-by-step analysis from spectrum to feature, the embodiments of the present application can effectively identify the intensity features of the chewing action, thereby achieving technical progress in the field of dietary behavior monitoring.

[0085] In step S404 of some embodiments, spectral time value analysis is performed on the target vibration perception spectrum to determine the chewing time value feature of the target perception signal.

[0086] It should be noted that the embodiments of the present application perform spectral time value analysis on the target vibration perception spectrum to determine the chewing time value characteristics of the target perception signal. This analysis process mainly focuses on the time characteristics of the energy duration in the spectrum, especially the time length information related to the chewing action. For example, by analyzing the time period of energy duration in the spectrum, the embodiments of the present application can determine the total duration of the chewing action, the proportion of effective chewing time, and the number of chewing interruptions. These time value characteristics reflect the continuity and concentration of the user's eating, and are important basis for evaluating whether the eating habits are healthy.

[0087] Referring to Figure 7 According to some embodiments of the present application, the step S404 of performing spectral time value analysis on the target vibration perception spectrum to determine the chewing time value characteristics of the target perception signal can include: Step S701, selecting a target spectrum segment from the target vibration perception spectrum; Step S702, dividing the target spectrum segment into a chewing occurrence sub-segment and a chewing stop sub-segment based on a preset chewing spectrum discrimination condition; Step S703, performing time value analysis according to the distribution characteristics of the chewing occurrence sub-segment and the chewing stop sub-segment in the target spectrum segment to determine the chewing time value characteristics of the target perception signal.

[0088] In some embodiments of the present application, the process of performing spectral time value analysis on the target vibration perception spectrum is an important part of chewing action recognition, and the purpose is to determine the chewing time value characteristics of the target perception signal by analyzing the spectrum data.

[0089] Step S701 of some embodiments, selecting a target spectrum segment from the target vibration perception spectrum; It should be noted that the target spectrum segment is selected from the target vibration perception spectrum. Since the vibration signal generated by the chewing action has specific frequency range and energy distribution characteristics in the spectrum, it is necessary to select the part related to the chewing behavior from the entire spectrum. This selection process is based on prior knowledge of the frequency characteristics of the chewing action, ensuring that the subsequent analysis is focused on the frequency spectrum interval most relevant to the chewing behavior, thereby improving the accuracy and efficiency of the analysis.

[0090] Step S702 of some embodiments, dividing the target spectrum segment into a chewing occurrence sub-segment and a chewing stop sub-segment based on a preset chewing spectrum discrimination condition; It should be noted that after selecting the target spectral segment, the chewing spectrum discrimination condition is preset based on the target spectral segment, and the target spectral segment is divided into chewing occurrence sub-segments and chewing stop sub-segments. This division process is achieved by analyzing the energy change and frequency distribution in the spectral segment. Chewing action usually shows energy increase in a specific frequency range in the spectrum, while energy decreases significantly when chewing stops. Therefore, the preset discrimination condition, such as energy threshold or frequency feature, is used to identify which sub-segments correspond to the occurrence of chewing action and which sub-segments correspond to the stop of chewing action. This division not only needs to consider the characteristics of individual sub-segments, but also needs to consider the continuity and correlation between sub-segments to ensure the accuracy and reliability of the division results.

[0091] In step S703 of some embodiments, time value analysis is performed according to the distribution characteristics of the chewing occurrence sub-segments and the chewing stop sub-segments in the target spectral segment to determine the chewing time value characteristics of the target perceptual signal.

[0092] It should be noted that the time value analysis is performed according to the distribution characteristics of the chewing occurrence sub-segments and the chewing stop sub-segments in the target spectral segment. This analysis process mainly determines the chewing time value characteristics of the target perceptual signal by counting the duration of the chewing occurrence sub-segments and the interval time of the chewing stop sub-segments. Specifically, the total duration of the chewing occurrence sub-segments reflects the total length of the chewing action, and the interval time of the chewing stop sub-segments reflects the number of interruptions and interruption times of the chewing action. Through these statistical information, the total length of chewing, the proportion of effective chewing time, and the number of chewing interruptions can be calculated. These parameters not only directly reflect the user's eating speed and eating continuity, but also reveal the user's behavior patterns and potential problems in the eating process.

[0093] It should be understood that the whole spectral time value analysis process is a systematic signal processing chain, from the selection of the target spectral segment to the division of the sub-segments, and then to the calculation of the time value characteristics, each step is closely connected and dependent on each other. This process not only solves the problem of accurately quantifying the chewing time value characteristics in the prior art, but also provides precise chewing behavior evaluation and health suggestions for users through detailed feature extraction. Through this step-by-step analysis from spectrum to feature, the chewing time value characteristics can be effectively identified, thereby achieving technical progress in the field of diet behavior monitoring. This analysis method provides important time dimension information for subsequent behavior analysis and health evaluation, helping users better understand their eating habits and take appropriate improvement measures.

[0094] Through the steps of the above embodiments of the present application, the embodiments of the present application can extract comprehensive and accurate chewing features from the target perception signal, which not only cover the rhythm, intensity and duration of the chewing action, but also reveal the behavior patterns and potential problems of the user in the eating process. The extraction of these features provides a solid data foundation for subsequent chewing behavior analysis and health assessment, enabling the system to provide personalized dietary recommendations and health interventions for the user.

[0095] In step S106 of some embodiments, chewing behavior analysis is performed on the chewing rhythm feature, the chewing intensity feature and the chewing duration feature to determine the chewing analysis result of the target object.

[0096] It should be noted that the embodiments of the present application perform behavior analysis on the extracted chewing features to determine the chewing analysis result of the target object. This analysis process combines multiple features and comprehensively evaluates the user's chewing rhythm, intensity and duration, so that the embodiments of the present application can fully understand the user's eating behavior. For example, by analyzing the chewing rhythm and intensity, the embodiments of the present application can determine whether the user has the problem of insufficient chewing; by analyzing the chewing duration and the number of interruptions, the embodiments of the present application can evaluate the user's eating continuity and concentration. These analysis results not only provide personalized dietary recommendations for the user, but also provide a basis for health monitoring and behavior intervention. Through this phased and multi-level processing method, the method of the present application can effectively solve the problems in the prior art and provide a more accurate, flexible and adaptable chewing action recognition scheme.

[0097] Referring to Figure 8 , Figure 8 The figure illustrates a behind-the-ear structure of the dietary behavior monitoring earphone according to an embodiment of the present application, which is designed to monitor and analyze the user's chewing behavior. Figure 8 In the figure, the housing of the earphone can be seen, which not only provides physical protection, but also accommodates all the internal components of the earphone. Inside the earphone, several key electronic modules are distributed, each with its specific function, working together to achieve the purpose of intelligent monitoring of the earphone.

[0098] Figure 8 In the figure, the position of the vibration sensor is marked. This vibration sensor is the core component for monitoring chewing action, which can capture the tiny vibrations generated when the user's jaw moves. These vibration signals will then be sent to the processor for further analysis and processing. The accurate placement and high sensitivity of the sensor are crucial for capturing the details of the chewing action, thereby ensuring the accuracy of the monitoring data.

[0099] Secondly, the power module provides the necessary power support for the earphone. This usually includes a small battery and possibly a charging circuit, ensuring that the earphone can work continuously for a long time. The design of the power module needs to consider the volume limit of the earphone and the wearing comfort of the user, while also ensuring the stable supply of power to support all the functions of the earphone.

[0100] The processor is the brain of the earphone, which is responsible for processing the data collected by the vibration sensor. The processor executes complex algorithms to analyze features such as chewing rhythm, force, and duration, and converts this information into useful monitoring results. The performance of the processor directly affects the speed and accuracy of data processing, so it needs to balance computing power and energy consumption when designing.

[0101] The memory is used to save the data analyzed by the processor and the firmware or software program of the earphone. This can include temporary storage to buffer data streams and long-term storage to save the user's historical chewing data. The size and read-write speed of the memory will affect the data storage capacity and response speed of the earphone.

[0102] The communication module is responsible for transmitting the data collected and processed by the earphone to the target terminal, such as a smartphone or tablet. Data transmission is achieved through wireless technology, such as Bluetooth. The design of the communication module needs to ensure the stability and security of data transmission, while also considering energy consumption and compatibility issues.

[0103] Reference Figure 9 , illustrates another in-ear structure of the diet behavior monitoring earphone of the embodiments of the present application, which includes a shell, a power module, a processor, a memory, a vibration sensor, and a communication module. The power module is responsible for providing power for the earphone, the processor processes the chewing action data captured by the vibration sensor, the memory is used to save data and programs, the vibration sensor is specifically used to monitor the vibration during chewing, and the communication module is responsible for wirelessly transmitting the processed data to external devices. These components work together to enable the earphone to achieve real-time monitoring and analysis of the user's diet behavior.

[0104] In summary, Figure 8 and Figure 9 demonstrates the various key components inside the diet behavior monitoring earphone and their functions. The precise design and coordinated work of these components enable the earphone to accurately monitor the user's diet actions and effectively transmit data to the target terminal for further analysis.

[0105] The embodiments of the present application also provide a computer program product, which includes a computer program. The processor of the computer device reads and executes the computer program, so that the computer device executes the chewing action recognition method as described above.

[0106] The terms "first", "second", "third", "fourth", and the like in the description of the disclosure and the above drawings, if any, are used to distinguish similar objects, and do not necessarily have to be used to describe a particular order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the disclosure described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "include" and "contain" and any variations thereof are intended to cover non-exclusive inclusion, for example, a process, method, system, product or device containing a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but can include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0107] It should be understood that in the present disclosure, "at least one" means one or more, and "multiple" means two or more. "And / or" is used to describe the relationship between the associated objects, which means that there can be three relationships, for example, "A and / or B" can mean that there are three cases: only A, only B, and A and B at the same time, where A and B can be singular or plural. The character " / " generally represents an "or" relationship between the associated objects. "At least one of the following" or similar expressions means any combination of these items, including any combination of single or multiple items. For example, at least one of a, b or c can mean a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0108] It should be understood that in the description of the embodiments of the present application, the meaning of multiple (or multiple) is two or more, greater than, less than, more than, etc. are not included in the number, above, below, etc. are understood to include the number.

[0109] In several embodiments provided by the present disclosure, it should be understood that the disclosed system, device and method can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of units is only a logical function division, and actual implementation can have another division manner, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the displayed or discussed objects can be indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0110] The units described as separate components may or may not be physically separate, and the components displayed as units may or may not be physical units, that is, may be located in one place, or may be distributed to multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment scheme.

[0111] In addition, each functional unit in various embodiments of the present disclosure can be integrated into one processing unit, or each unit can exist physically, or two or more units can be integrated into one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit.

[0112] If the integrated unit is realized in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present disclosure essentially or the part that contributes to the prior art or the whole or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the various embodiments of the method of the present disclosure. The aforementioned storage medium can include: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0113] It should also be appreciated that the various embodiments provided by the present application can be combined in any way to achieve different technical effects.

[0114] The above is a specific description of the embodiments of the present disclosure, but the present disclosure is not limited to the above embodiments. Those skilled in the art can make various equivalent modifications or replacements without departing from the spirit of the present disclosure, and these equivalent modifications or replacements are all included in the scope defined by the claims of the present disclosure.

Claims

1. A method of recognizing a chewing action, characterized by, Applied to a target terminal, comprising: Receiving a target perception signal from a dietary behavior monitoring earphone; wherein the target perception signal is obtained in the dietary behavior monitoring earphone by the following steps: vibration perception is performed on a target object to obtain a perception data stream, noise frequency bands are screened out for the perception data stream to obtain a target perception signal; Performing feature extraction on the target perception signal to obtain chewing rhythm features, chewing force features and chewing time value features of the target object; Performing chewing behavior analysis on the chewing rhythm features, the chewing force features and the chewing time value features to determine a chewing analysis result of the target object.

2. The method of claim 1, wherein, The feature extraction on the target perception signal to obtain the chewing rhythm features, the chewing force features and the chewing time value features of the target object comprises: Generating a target vibration perception spectrum based on the target perception signal; Performing spectrum dynamic analysis on the target vibration perception spectrum to determine the chewing rhythm features of the target perception signal; Performing spectrum amplitude analysis on the target vibration perception spectrum to determine the chewing force features of the target perception signal; Performing spectrum time value analysis on the target vibration perception spectrum to determine the chewing time value features of the target perception signal.

3. The method of claim 2, wherein, The spectrum dynamic analysis on the target vibration perception spectrum to determine the chewing rhythm features of the target perception signal comprises: Dividing the vibration perception spectrum based on a pre-determined first division interval to obtain a plurality of dynamic analysis frames; Extracting corresponding vibration main frequency data from a plurality of the dynamic analysis frames; Performing dynamic feature analysis based on the vibration main frequency data of a plurality of continuous dynamic analysis frames to obtain main frequency data dynamic features as the chewing rhythm features.

4. The method of claim 2, wherein, The spectrum amplitude analysis on the target vibration perception spectrum to determine the chewing force features of the target perception signal comprises: Selecting a target spectrum segment from the target vibration perception spectrum; Dividing the target spectrum segment based on a pre-determined second division interval to obtain a plurality of spectrum sub-segments; Performing amplitude analysis on each of the spectrum sub-segments to obtain corresponding local amplitudes; Performing amplitude variation calculation on the local amplitudes of a plurality of continuous spectrum sub-segments to obtain an amplitude variation coefficient; Generating the chewing force features based on the amplitude variation coefficient and the local amplitudes of each of the spectrum sub-segments.

5. The method of claim 2, wherein, The spectrum time value analysis on the target vibration perception spectrum to determine the chewing time value features of the target perception signal comprises: Selecting a target spectrum segment from the target vibration perception spectrum; Dividing the target spectrum segment into a chewing occurrence sub-segment and a chewing stop sub-segment based on a pre-set chewing spectrum discrimination condition; Performing time value analysis according to the distribution characteristics of the chewing occurrence sub-segment and the chewing stop sub-segment in the target spectrum segment to determine the chewing time value features of the target perception signal.

6. The method of claim 1, wherein, The chewing behavior analysis is performed on the chewing rhythm feature, the chewing strength feature, and the chewing time value feature to determine a chewing analysis result of the target object, including: Based on the chewing rhythm feature, rhythm index analysis is performed to obtain an average chewing frequency and a chewing frequency stability; Based on the chewing strength feature, strength index analysis is performed to obtain an average chewing energy intensity and an amplitude variation coefficient; Based on the chewing time value feature, time value index analysis is performed to obtain a total chewing time, an effective chewing time proportion, and a chewing interruption number; Feature fusion processing is performed on the average chewing frequency, the chewing frequency stability, the average chewing energy intensity, the amplitude variation coefficient, the total chewing time, the effective chewing time proportion, and the chewing interruption number to obtain a target chewing feature vector; Based on the target chewing feature vector, chewing mode recognition is performed to determine the chewing analysis result of the target object.

7. The method of claim 6, wherein, The chewing mode recognition based on the target chewing feature vector to determine the chewing analysis result of the target object includes: An eating behavior mode library is obtained; the eating behavior mode library stores corresponding chewing mode representation vectors for various eating behavior modes; The target chewing feature vector is compared with each chewing mode representation vector to select a hit mode representation vector from the eating behavior mode library; The hit mode representation vector corresponds to an eating behavior mode, and the chewing analysis result is generated based on the eating behavior mode.

8. A method of recognizing a chewing action, characterized by, An application for a dietary behavior monitoring earphone includes: Vibration sensing of a target object is performed to obtain sensing data flow; Noise frequency bands are screened out from the sensing data flow to obtain a target sensing signal; Feature extraction is performed on the target sensing signal to obtain target behavior features of the target object; The target behavior features are sent to a target terminal to extract features of the target sensing signal through the target terminal to obtain chewing rhythm features, chewing strength features, and chewing time value features of the target object, and to perform chewing behavior analysis on the chewing rhythm features, the chewing strength features, and the chewing time value features to determine a chewing analysis result of the target object.

9. A dietary behavior monitoring earpiece, comprising: It includes: A memory stores a computer program, and a processor executes the computer program to implement the chewing action recognition method of any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The storage medium stores a program, and the program is executed by the processor to implement the chewing action recognition method of any one of claims 1 to 8.

Images