Diet management method based on health data, storage device, and medium

Abstract

The application relates to the technical field of diet management, and discloses a diet management method based on health data, which comprises the following steps: acquiring first camera data, wherein the first camera data comprises a face image; predicting real-time body state data according to the first camera data and generating a first regulation index, wherein the first regulation index comprises parameters corresponding to at least part of the body state data; acquiring second environment data, wherein the second environment data comprises kitchen cooking data; generating second regulation data according to the second environment data; and generating a recommended nutrition list according to the first regulation data and the second regulation data. The diet management method improves the accuracy of diet suggestions.

Description

Technical Field

[0001] This application relates to the field of dietary management technology, and in particular to a dietary management method, storage device and medium based on health data. Background Technology

[0002] With the continuous improvement of people's living standards and the increasing demand for health management, the research and application of dietary nutrition care has gradually become an important direction in the field of public health. In recent years, the rapid development of intelligent health technology and big data analysis has brought new opportunities to dietary nutrition management. Intelligent dietary nutrition management methods based on health data are gradually emerging, and many intelligent devices and applications are beginning to provide personalized dietary recommendations by monitoring users' health status (such as weight, blood sugar, blood lipids, etc.) and eating habits. Currently, related technologies mainly collect users' physiological data and dietary preference data, use algorithms to analyze users' nutritional needs, and thus recommend dietary plans that meet their health needs. However, although existing dietary recommendation systems have improved the accuracy of personalized nutrition management to some extent, many problems still exist, especially in the application and accuracy of food selection, nutrient loss, and dynamic optimization of dietary plans, which remain relatively limited.

[0003] In existing technologies, dietary recommendations are often not accurate enough. Therefore, how to improve the accuracy of dietary recommendations is a technical problem that needs to be solved. Summary of the Invention

[0004] This application provides a dietary management method, storage device, and medium based on health data, which improves the accuracy of dietary recommendations.

[0005] To achieve the above objectives, the main technical solutions adopted in this application include: In a first aspect, embodiments of this application provide a dietary management method based on health data, comprising the following steps: acquiring first camera data, the first camera data including facial images; predicting real-time body state data based on the first camera data and generating a first regulatory indicator, the first regulatory indicator including parameters corresponding to at least some of the body state data; acquiring second environmental data, the second environmental data including kitchen cooking data; generating second regulatory data based on the second environmental data; and generating a recommended nutrition list based on the first regulatory data and the second regulatory data.

[0006] The dietary management method proposed in this application can not only predict body status data based on facial images, thereby helping to determine the user's actual physical condition, but also estimate dietary conditions based on kitchen cooking data, improving the accuracy of predicting body status data. Finally, it combines first camera data and second environmental data to generate dietary recommendations, avoiding the lag bias caused by traditional methods relying on historical data, ensuring that the recommendations are highly matched with real-time health status, and improving the accuracy of dietary recommendations.

[0007] Optionally, acquiring the first camera data includes: Multispectral images are acquired using a kitchen camera, and facial data signals are extracted from the multispectral images. The facial data signal is corrected in real time based on the motion compensation model, and the facial image is reconstructed based on the corrected facial data signal.

[0008] In the above solution, by optimizing facial data acquisition and processing, interference from kitchen fumes and steam can be reduced, physiological signals such as facial capillary pulsation can be clearly extracted, the influence of stray light in the environment can be reduced, and the motion compensation model can be used to correct in real time, accurately offsetting image displacement and local deformation caused by actions such as turning the head and stirring during cooking. Through global motion capture, local deformation correction and smoothing processing, signal stability is ensured, providing key support for subsequent health management.

[0009] Optionally, real-time correction of facial data signals based on a motion compensation model includes: Global motion parameters are determined using optical flow, local deformation corrections are performed based on these global motion parameters to obtain local correction parameters, and time series prediction is performed based on both the global motion parameters and the local deformation correction parameters.

[0010] In the above scheme, optical flow is used to capture global head movements such as shaking and translation, and then targeted corrections are made to local muscle deformations such as speaking and chewing. This achieves dual correction of "global calibration + local refinement," which helps reduce dynamic interference and avoids the inadequacy of a single motion model for complex facial movements. By combining global and local parameters for time-series prediction, the motion state of the next frame can be estimated in advance, reducing delays caused by image blurring or loss of feature points, and further improving the motion compensation response speed.

[0011] Optionally, determining global motion parameters based on optical flow includes: A stable set of facial feature points is extracted from multispectral images, and an iterative algorithm is used to calculate the inter-frame optical flow vector for the selected feature points. Rigid body motion model fitting is performed based on optical flow vectors, and global motion parameters are output.

[0012] The above scheme ensures the robustness of motion estimation, improves the ability to resist environmental interference, and enhances parameter accuracy, providing a reliable benchmark for subsequent local deformation correction. It effectively reduces physiological signal shifts caused by global motion and improves the prediction accuracy of body state data. In other words, through refined feature point processing and motion modeling, the accuracy and reliability of global motion estimation are significantly improved, laying a solid foundation for subsequent signal correction.

[0013] Optionally, local correction parameters can be obtained by performing local deformation correction based on global motion parameters, including: Based on global motion parameters, local dense feature points of the face are extracted, and local feature points of adjacent frames are matched using a neighborhood matching algorithm; Identify regions with nonlinear deformation and perform local pixel remapping on these regions based on piecewise affine transformation.

[0014] In the above scheme, the deficiencies of global motion estimation in local details are effectively made up for by refined local feature processing and nonlinear deformation correction, providing a higher-precision image basis for facial physiological signal extraction.

[0015] Optionally, time series prediction based on global motion parameters and local deformation correction parameters includes: The state transition equation is constructed by using global motion parameters and local deformation correction parameters as state vectors. The motion state is predicted by determining the observed values ​​based on the state transition equation and the image noise level.

[0016] In the above scheme, global motion parameters and local deformation parameters are integrated into a state vector, comprehensively covering the overall head movement and local facial muscle deformation. This avoids the limitations of single-parameter modeling and makes the motion description more closely resemble the actual scene. Dynamic noise processing enhances anti-interference capabilities, with observation noise adjusted in real time according to image noise, effectively filtering errors caused by oil fumes and blurring. Temporal prediction anticipates the state of the next frame, further reducing motion compensation delay and ensuring real-time performance. Ultimately, this provides a temporally continuous, low-noise image foundation for stable extraction of physiological signals, improving the accuracy of body state prediction and contributing to the accuracy of health demand analysis.

[0017] Optionally, obtaining second environment data includes: Obtain the operating parameters of the equipment and environmental parameters in the kitchen, and generate kitchen cooking data based on the operating parameters and environmental parameters.

[0018] In the above solution, generating kitchen cooking data based on equipment operating parameters and environmental parameters can improve the comprehensiveness and practicality of kitchen cooking data, providing precise scenario support for dietary recommendations. By integrating equipment operating parameters (such as smart stove temperature and duration, refrigerator food inventory, and seasoning usage) and environmental parameters (temperature, humidity, and oil fume concentration), the generated kitchen cooking data can accurately reflect the actual state of the kitchen, thereby further improving the accuracy of dietary recommendations.

[0019] Optionally, the recommended nutrition list generated based on the first and second regulatory data includes: Based on the analysis of users' real-time nutritional needs using the first regulatory indicator, and combined with the second regulatory data, food ingredients are matched and screened to build a nutritional adaptation model and generate dynamic adjustment rules.

[0020] In the above scheme, real-time nutritional needs are analyzed based on the first regulatory indicator, which can accurately capture the user's current health status and avoid the lag of traditional recommendations that rely on historical data; dietary recommendations are further optimized by combining the second regulatory data; and dynamic adjustment rules are implemented based on actual cooking conditions to further improve the accuracy of dietary recommendations.

[0021] Secondly, embodiments of this application provide a storage device, which includes a controller, a memory, and a computer program stored in the memory and executable by the controller, wherein when the computer program is executed by the controller, it implements the dietary management method of any embodiment.

[0022] The storage device proposed in this application embodiment can implement the dietary management method of any embodiment, making it easy to be deployed in smart home devices and further improving the ease of use of smart homes.

[0023] Thirdly, embodiments of this application provide a storage medium for computer-readable storage, the storage medium storing one or more programs, which can be executed by one or more processors to implement the dietary management method of any embodiment.

[0024] The storage medium proposed in this application embodiment can implement the dietary management method of any embodiment, thus improving the ease of use of smart homes. Attached Figure Description

[0025] To more clearly illustrate the technical solutions in the specific embodiments of this application or the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0026] Figure 1 This is a schematic diagram of the steps of the dietary management method in some embodiments of this application; Figure 2 This is a schematic diagram of the steps of the dietary management method in some other embodiments of this application; Figure 3 This is a schematic diagram of the steps of the dietary management method in some other embodiments of this application; Figure 4 This is a schematic diagram of the steps of a dietary management method in some other embodiments of this application. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, 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 some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0028] Unless otherwise defined, all technical and scientific terms used in this application have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used in the description of this application is for the purpose of describing particular embodiments only and is not intended to limit the application; the terms "comprising" and "having," and any variations thereof, in the description, claims, and accompanying drawings of this application are intended to cover non-exclusive inclusion. The terms "first," "second," etc., in the description, claims, or accompanying drawings of this application are used to distinguish different objects, not to describe a specific order or hierarchy.

[0029] In this application, the reference to "embodiment" means that a specific feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a mutually exclusive, independent, or alternative embodiment. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described in this application can be combined with other embodiments.

[0030] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installation," "connection," "linking," and "attachment" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal communication between two components. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0031] In this application, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, in this application, the character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0032] In this application, "multiple" refers to two or more (including two), similarly, "multiple sets" refers to two or more (including two sets), and "multiple pieces" refers to two or more (including two pieces).

[0033] With the continuous improvement of people's living standards and the increasing demand for health management, the research and application of dietary nutrition care has gradually become an important direction in the field of public health. In recent years, the rapid development of intelligent health technology and big data analysis has brought new opportunities to dietary nutrition management. Intelligent dietary nutrition management methods based on health data are gradually emerging, and many intelligent devices and applications are beginning to provide personalized dietary recommendations by monitoring users' health status (such as weight, blood sugar, blood lipids, etc.) and eating habits. Currently, related technologies mainly collect users' physiological data and dietary preference data, use algorithms to analyze users' nutritional needs, and thus recommend dietary plans that meet their health needs. However, although existing dietary recommendation systems have improved the accuracy of personalized nutrition management to some extent, many problems still exist, especially in the application and accuracy of food selection, nutrient loss, and dynamic optimization of dietary plans, which remain relatively limited.

[0034] In current technologies, the data between various devices in a smart home are independent of each other, and dietary recommendations are often not accurate enough. Therefore, how to improve the accuracy of dietary recommendations is a technical problem that needs to be solved.

[0035] Therefore, embodiments of this application provide a dietary management method based on health data, which improves the accuracy of dietary recommendations.

[0036] Please refer to Figure 1According to an embodiment of the first aspect of this application, a dietary management method based on health data includes the following steps: S1, acquiring first camera data, the first camera data including facial images; S2, predicting real-time body state data based on the first camera data and generating a first regulatory index, the first regulatory index including parameters corresponding to at least some of the body state data; S3, acquiring second environmental data, the second environmental data including kitchen cooking data, generating second regulatory data based on the second environmental data, and generating a recommended nutrition list based on the first and second regulatory data. This method not only predicts body state data based on facial images, thus helping to determine the user's actual physical condition, but also estimates dietary conditions based on kitchen cooking data, improving the accuracy of predicting body state data. Finally, it combines the first camera data and the second environmental data to generate dietary recommendations, avoiding the lag bias caused by traditional methods relying on historical data, ensuring a high degree of match between the recommendations and real-time health status, and improving the accuracy of dietary recommendations.

[0037] The dietary management method proposed in this application, on the one hand, relies on the first camera data to capture physiological signals in the user's facial image in real time, such as heart rate, blood pressure, and metabolic rate fluctuations, and dynamically generates a first regulatory indicator to accurately anchor the current nutritional needs gap. For example, when blood pressure rises instantaneously, it immediately marks the need for "low sodium and high potassium". This avoids the lag bias caused by the reliance on historical data in traditional methods and ensures that the recommended starting point is highly matched with the real-time health status.

[0038] On the other hand, by integrating information from the entire kitchen cooking scenario through second environmental data, such as ingredient inventory, equipment parameters, and real-time cooking progress, the system can determine the real-time dietary situation, predict physical condition data, and then make accurate dietary recommendations. Ultimately, this achieves precise dietary advice across the entire chain, from demand positioning and scenario adaptation to dynamic optimization.

[0039] In a specific embodiment, this method is applicable to smart kitchens. Specifically, it can be applied to smart devices such as cameras, smoke sensors, and smart terminals. Based on the first camera data and the second environmental data during the cooking process, it intelligently generates dietary suggestions and pushes them to the user on the smart terminal to improve the user experience.

[0040] In other embodiments, please refer to Figure 2 The acquisition of the first camera data includes: S11. Acquire multispectral images through a kitchen camera, and extract facial data signals based on the multispectral images; it is understood that the multispectral images contain near-infrared band data of 700-1000nm.

[0041] Kitchen scenes are often filled with interference from fumes, steam, and dynamic light (such as stove flames). Traditional visible light images are easily obscured or scattered, leading to blurred facial features. The S11, however, employs multispectral imaging including the 700-1000nm near-infrared band. This band of light has strong penetrating power, effectively penetrating particulate matter such as fumes and steam, reducing the impact of stray ambient light on image quality. This ensures clearer and more stable physiological signals (such as pulsation waveforms) extracted from capillaries under the facial skin, providing high-quality raw input for subsequent predictions of body condition data such as heart rate and blood pressure.

[0042] S12. Correct the facial data signal in real time according to the motion compensation model, and reconstruct the facial image based on the corrected facial data signal.

[0043] It is understandable that users inevitably engage in dynamic movements such as stirring, turning their heads, and shaking their limbs while cooking in the kitchen, which can easily lead to inter-frame displacement and local deformation of facial images, directly affecting the accuracy of physiological signal extraction, such as misjudging capillary pulsation as motion noise.

[0044] Real-time correction is achieved through a motion compensation model: first, global motion trajectories are captured using algorithms such as optical flow; then, local muscle deformations are corrected; and finally, time-series predictions are used to smooth signal fluctuations. This correction mechanism helps reduce dynamic interference, ensures the stability of physiological feature points in the reconstructed facial image, avoids distortion of physiological signals caused by motion, and lays the foundation for accurate prediction of body state data.

[0045] In the above solution, by optimizing facial data acquisition and processing, interference from kitchen fumes and steam can be reduced, physiological signals such as facial capillary pulsation can be clearly extracted, the influence of stray light in the environment can be reduced, and the motion compensation model can be used to correct in real time, accurately offsetting image displacement and local deformation caused by actions such as turning the head and stirring during cooking. Through global motion capture, local deformation correction and smoothing processing, signal stability is ensured, providing key support for subsequent health management.

[0046] In other embodiments, please refer to Figure 3 Real-time correction of facial data signals based on a motion compensation model includes: S121. Determine global motion parameters based on optical flow method, and obtain local correction parameters by performing local deformation correction based on global motion parameters; S122. Perform time series prediction based on global motion parameters and local deformation correction parameters.

[0047] In the above scheme, optical flow is used to capture global head movements such as shaking and translation, and then targeted corrections are made to local muscle deformations such as speaking and chewing. This achieves dual correction of "global calibration + local refinement," which helps reduce dynamic interference and avoids the inadequacy of a single motion model for complex facial movements. By combining global and local parameters for time-series prediction, the motion state of the next frame can be estimated in advance, reducing delays caused by image blurring or loss of feature points, and further improving the motion compensation response speed.

[0048] In other embodiments, please refer to Figure 4 Determining global motion parameters based on optical flow includes: S1211. Extract a stable set of facial feature points from the multispectral image, and use an iterative algorithm to calculate the inter-frame optical flow vector for the selected feature points. As an example, extract a stable set of facial feature points from the multispectral image, and preferentially select Harris corner points in rigid areas such as the corners of the eyes, the tip of the nose, and the corners of the mouth to ensure that ≥50 non-redundant feature points are extracted from each frame of the image, and that the feature points are evenly distributed in the facial area, covering areas such as the forehead, cheeks, and jaw.

[0049] Determine the inter-frame optical flow vector for selected feature points, for example, by calculating the inter-frame optical flow vector using the Lucas-Kanade iterative algorithm. Determine the pixel displacement based on the inter-frame optical flow vector, for example, by minimizing the grayscale consistency error function.

[0050] The window size is dynamically adjusted during the determination of pixel displacement (expanded to a 7×7 window when the motion is intense), and an iteration termination threshold is set (stopped when the displacement change is less than the threshold number of pixels) to further reduce the error in optical flow vector calculation and provide high-precision displacement data for subsequent motion modeling.

[0051] S1212. Fitting a rigid body motion model based on optical flow vectors and outputting global motion parameters. It can be understood that the rigid body motion model fitting focuses on eliminating outliers and accurately estimating global parameters.

[0052] As an example, this scheme first uses the RANSAC algorithm to robustly screen optical flow vectors. Multiple candidate motion models are generated through random sampling, and the proportion of optical flow vectors that conform to the motion law within the model is calculated. Outliers caused by oil fume interference and feature point mismatch are eliminated, and ≥80% of the effective optical flow vectors are retained. Then, a rigid body motion model is fitted based on the remaining optical flow vectors. This model includes multiple parameters, including translation, rotation angle and scaling factor. The least squares method is used to solve for the parameter combination that minimizes the error between the predicted displacement and the actual optical flow vector, ensuring that the mean square error is <2 times the square of the pixel. The final output global motion parameters can completely describe the overall swaying, translation, rotation and other motion states of the user's head, providing a benchmark reference for subsequent local deformation correction and further improving the accuracy of global motion compensation.

[0053] In the above scheme, the accuracy and reliability of global motion estimation are significantly improved by refining feature point processing and motion modeling, laying a solid foundation for subsequent signal correction.

[0054] Understandably, selecting Harris corner points in rigid regions such as the corner of the eye, combined with ≥50 non-redundant and evenly distributed feature points, ensures the stability of motion tracking (rigid regions are less affected by muscle interference) while avoiding the loss of local information. By dynamically adjusting the window size and using a strict iterative threshold, optical flow vector errors are further reduced, providing high-precision displacement data for motion analysis.

[0055] By using the RANSAC algorithm to remove outliers such as oil fumes and mismatches, and then fitting a rigid body model using the least squares method, global parameters such as translation, rotation, and scaling are accurately extracted. This ensures the robustness of motion estimation, improves the ability to resist environmental interference, and enhances parameter accuracy, providing a reliable benchmark for subsequent local deformation correction. It effectively reduces physiological signal shifts caused by global motion and improves the prediction accuracy of body state data.

[0056] In other embodiments, the local correction parameters are obtained by performing local deformation correction based on global motion parameters, including: S1213. Extract dense local feature points on the face based on global motion parameters, and match local feature points in adjacent frames using a neighborhood matching algorithm; that is, based on global motion estimation, extract dense local feature points on the face (≥10 feature points per square centimeter), focusing on covering the cheek area with dense capillaries.

[0057] The nearest neighbor matching algorithm matches local feature points in adjacent frames and calculates the affine transformation matrix M of the local region. The elements in the affine transformation matrix M include the scaling, translation, and rotation of the feature points.

[0058] S1214. Determine the regions with nonlinear deformation and perform local pixel remapping on the regions with nonlinear deformation based on piecewise affine transformation.

[0059] Understandably, for regions with nonlinear deformation, such as facial muscle deformation caused by a user speaking or chewing, local pixel remapping of the image is performed based on piecewise affine transformation to correct the capillary pulsation signal shift caused by local deformation, ensuring that the positional error of local feature points is ≤1 pixel.

[0060] In the above scheme, the deficiencies of global motion estimation in local details are effectively made up for by refined local feature processing and nonlinear deformation correction, providing a higher-precision image basis for facial physiological signal extraction.

[0061] Understandably, the accurate extraction and matching of local dense feature points significantly improves the finesse of local motion perception: after global motion parameter calibration, dense feature points of ≥10 per square centimeter (especially focusing on the capillary-rich cheek area) can capture millimeter-level local muscle changes. Combined with the nearest neighbor matching algorithm for accurate alignment of feature points in adjacent frames, an affine transformation matrix M containing scaling, translation, and rotation parameters can be generated, providing fine-grained data support for local deformation modeling and avoiding missed detection of local motion due to sparse feature points.

[0062] Piecewise affine transformation specifically addresses the interference of nonlinear deformation: For non-rigid facial muscle deformations caused by speaking and chewing (such as pulling at the corners of the mouth and bulging of the cheeks), affine transformation is performed separately on independent regions, which can accurately correct the spatial offset of capillary pulsation signals. Compared with a single global transformation, this piecewise processing controls the positional error of local feature points to ≤1 pixel, ensuring that the physiological signals (such as pulsation waveforms) in the capillary distribution area maintain spatial consistency between frames and avoiding misinterpretation of muscle deformation as changes in blood flow signals. This lays a high-fidelity signal foundation for subsequent physiological parameter calculations: In the kitchen scene, users' facial dynamics are frequent (such as changes in facial expressions when stir-frying). Through precise matching and remapping, these steps further reduce signal noise caused by local deformation, making the extracted capillary pulsation signals purer, directly improving the prediction accuracy of parameters such as heart rate and blood pressure, and ultimately providing reliable data support for the accurate mapping of "body state - nutritional needs" in dietary management.

[0063] In other embodiments, time series prediction based on global motion parameters and local deformation correction parameters includes: S1221. Construct the state transition equation by using the global motion parameters and local deformation correction parameters as state vectors: Xk=F⋅X k−1 +W k−1 This includes the state transition equation F based on the assumption of motion continuity, with the translation amount updated according to a uniform velocity model, and W. k-1 X is the process noise (following a Gaussian distribution N(0,Q), where Q is the process noise covariance matrix). k This is the state vector.

[0064] S1222. Determine the observed values ​​and predict the motion state based on the state transition equation and the image noise level.

[0065] This includes using the motion parameters obtained from the Lucas-Kanade optical flow method and feature point matching as the observed value Z. k Construct the observation equation: Z k =H⋅X k +V kWhere H is the observation matrix (mapping the state vector to the observation space), and V k The observation noise (following a Gaussian distribution N(0,R), where R is the observation noise covariance matrix, dynamically adjusted according to the image noise level) is used to predict the motion state of the next frame, reducing the motion estimation delay caused by image blurring or loss of feature points, and further reducing the time response delay of motion compensation.

[0066] In the above scheme, the accurate prediction and dynamic optimization of motion state provide key technical support for real-time motion compensation, which helps to further improve the accuracy of prediction.

[0067] Understandably, the comprehensiveness of the state vector enhances the completeness of motion modeling: it integrates global motion parameters (translation, rotation, scaling) with local deformation correction parameters (affine matrix coefficients) into a state vector X. k It covers both overall head movement and local facial muscle deformation, avoiding the one-sidedness of single-parameter modeling and making the description of motion state more in line with the actual scene.

[0068] Dynamic noise handling enhances the model's robustness: the state transition equations are based on the assumption of motion continuity (e.g., translations are updated according to a uniform model), and process noise W is used to improve the model's robustness. k-1 (Gaussian distribution N(0,Q)) quantifies the uncertainty of abrupt motion changes; the observation equation uses the optical flow method and feature point matching results as the observation value Z. k Combined with dynamically adjusted observation noise V k (N(0,R), R is updated in real time with the image noise level), effectively filtering observation errors caused by oil fumes and blur in the kitchen environment, further improving the robustness of motion parameter estimation.

[0069] Furthermore, temporal prediction significantly reduces real-time latency: through an iterative mechanism of state equation prediction and observation equation update, the system can predict the motion state of the next frame in advance, compensating for computational delays caused by image blurring or loss of feature points, and further improving the temporal response latency of motion compensation. Ultimately, this scheme provides a temporally continuous and lower-noise image foundation for the stable extraction of facial physiological signals (such as capillary pulsation), directly improving the accuracy of subsequent body state prediction and contributing to the accuracy of health demand analysis in dietary management.

[0070] In other embodiments, obtaining the second environmental data includes: Obtain the operating parameters of the equipment and environmental parameters in the kitchen, and generate kitchen cooking data based on the operating parameters and environmental parameters.

[0071] In the above solution, generating kitchen cooking data based on equipment operating parameters and environmental parameters can improve the comprehensiveness and practicality of kitchen cooking data, providing precise scenario support for dietary recommendations. By integrating equipment operating parameters (such as smart stove temperature and duration, refrigerator food inventory, and seasoning usage) and environmental parameters (temperature, humidity, and oil fume concentration), the generated kitchen cooking data can accurately reflect the actual state of the kitchen, thereby further improving the accuracy of dietary recommendations.

[0072] In other embodiments, generating a recommended nutrition list based on first and second regulatory data includes: Based on the analysis of users' real-time nutritional needs using the first regulatory indicator, and combined with the second regulatory data, food ingredients are matched and screened to build a nutritional adaptation model and generate dynamic adjustment rules.

[0073] In the above scheme, real-time nutritional needs are analyzed based on the first regulatory indicator, which can accurately capture the user's current health status and avoid the lag of traditional recommendations that rely on historical data; dietary recommendations are further optimized by combining the second regulatory data; and dynamic adjustment rules are implemented based on actual cooking conditions to further improve the accuracy of dietary recommendations.

[0074] In other specific embodiments, this application also discloses a usage monitoring unit: using the electromagnetic flowmeter of the smart salt container and the pressure sensor of the kitchen scale to collect the amount of salt and oil used and the weight of the ingredients in real time.

[0075] The nutrition calculation engine dynamically calculates the calorie, sodium, and fat intake of the current recipe based on a nutrition database.

[0076] At the same time, a nutritional balance index is established, and nutrient reference intakes are generated based on secondary environmental data.

[0077] In other specific embodiments, this application also discloses a dietary management method, which includes using blockchain technology to construct a distributed ledger to store family cooking data and members' physiological parameter change curves.

[0078] Establish a model to predict the amount of trans fatty acids produced, and calculate the content of harmful substances based on cooking temperature and time; establish a health risk assessment matrix, and combine user physiological data and the nutritional components of food to generate risk levels.

[0079] Secondly, embodiments of this application provide a storage device, which includes a controller, a memory, and a computer program stored in the memory and executable by the controller, wherein when the computer program is executed by the controller, it implements the dietary management method of any embodiment.

[0080] In one specific embodiment, a lightweight AI chip integrated into the refrigerator is also included, which has a built-in facial recognition algorithm and feature desensitization module, and erases the original facial image in real time after the physiological parameters are extracted.

[0081] The storage device proposed in this application embodiment can implement the dietary management method of any embodiment, making it easy to be deployed in smart home devices and further improving the ease of use of smart homes.

[0082] Thirdly, embodiments of this application provide a storage medium for computer-readable storage, the storage medium storing one or more programs, which can be executed by one or more processors to implement the dietary management method of any embodiment.

[0083] The storage medium proposed in this application embodiment can implement the dietary management method of any embodiment, thus improving the ease of use of smart homes.

[0084] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0085] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

[0086] The above description is merely an embodiment of this application and is not intended to limit the scope 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 the claims of this application.

[0087] Although embodiments of this application have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of this application, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A dietary management method based on health data, characterized in that, Includes the following steps: Acquire first camera data, the first camera data including facial images; Based on the first camera data, real-time body state data is predicted and a first control index is generated. The first control index includes at least some of the parameters corresponding to the body state data. Acquire second environmental data, including kitchen cooking data, generate second regulatory data based on the second environmental data, and generate a recommended nutrition list based on the first regulatory data and the second regulatory data.

2. The dietary management method according to claim 1, characterized in that, The acquisition of the first camera data includes: Multispectral images are acquired using a kitchen camera, and facial data signals are extracted from the multispectral images. The facial data signal is corrected in real time according to the motion compensation model, and the facial image is reconstructed based on the corrected facial data signal.

3. The dietary management method according to claim 2, characterized in that, The real-time correction of the facial data signal based on the motion compensation model includes: Global motion parameters are determined based on optical flow, local deformation corrections are performed based on the global motion parameters to obtain local correction parameters, and time series prediction is performed based on the global motion parameters and the local deformation correction parameters.

4. The dietary management method according to claim 3, characterized in that, The determination of global motion parameters based on optical flow includes: A stable set of facial feature points is extracted from multispectral images, and an iterative algorithm is used to calculate the inter-frame optical flow vector for the selected feature points. Rigid body motion model fitting is performed based on optical flow vectors, and global motion parameters are output.

5. The dietary management method according to claim 4, characterized in that, The step of obtaining local correction parameters by performing local deformation correction based on the global motion parameters includes: Based on the global motion parameters, local dense feature points of the face are extracted, and local feature points of adjacent frames are matched using a neighborhood matching algorithm; Identify regions with nonlinear deformation and perform local pixel remapping on these regions based on piecewise affine transformation.

6. The dietary management method according to claim 3, characterized in that, The time series prediction based on the global motion parameters and the local deformation correction parameters includes: The global motion parameters and the local deformation correction parameters are used as state vectors to construct the state transition equation; The observed values ​​are determined based on the state transition equation and the image noise level, and the motion state is predicted.

7. The dietary management method according to claim 1, characterized in that, The acquisition of the second environmental data includes: The operating parameters of the equipment and environmental parameters in the kitchen are obtained, and the kitchen cooking data is generated based on the operating parameters of the equipment and the environmental parameters.

8. The dietary management method according to claim 1, characterized in that, The step of generating a recommended nutrition list based on the first regulatory data and the second regulatory data includes: Based on the analysis of the user's real-time nutritional needs using the first regulatory index, and combined with the second regulatory data, food ingredients are matched and screened to build a nutritional adaptation model and generate dynamic adjustment rules.

9. A storage device, characterized in that, The storage device includes a controller, a memory, and a computer program stored in the memory and executable by the controller, wherein when the computer program is executed by the controller, it implements the dietary management method as described in any one of claims 1 to 8.

10. A storage medium for computer-readable storage, characterized in that, The storage medium stores one or more programs, which can be executed by one or more processors to implement the dietary management method according to any one of claims 1 to 8.

Images