Visual aid-based magnetic therapy back massager

Abstract

The application discloses a visual aid-based magnetic therapy back massage instrument and belongs to the technical field of massage instruments.The application comprises a massage instrument main body and a control module, a shoulder part and a rolling bin are connected through a chest band and a back strap, the back strap is inserted into the connecting part to assist in positioning, double fixation avoids massage displacement, the back of the massage instrument main body is provided with a supporting plate, the inside of a massage head is provided with an infrared magnetic therapy stone, the control module automatically divides back muscle zones according to a user's back image, identifies and marks key nursing areas through skin color and muscle texture changes, drives the massage head to massage based on the visual recognition of the back muscle zones, and automatically adjusts the pressure and the magnetic therapy mode of the massage head, which can automatically adapt to the real-time state of the user's back, can iteratively optimize the scheme with the use frequency, greatly improves the massage pertinence and long-term nursing effect, meets the labor strain relief needs of different users, and the user can intuitively monitor the massage area and key nursing parts through a terminal.

Description

Technical Field

[0001] This invention relates to the field of massage device technology, and in particular to a vision-assisted magnetic therapy back massager. Background Technology

[0002] Existing back massagers still have many technical limitations in practical applications, making it difficult to meet users' needs for precise and personalized care.

[0003] Because there are significant differences in the distribution of back muscles, the location of strain, and the degree of stiffness among different users, traditional massagers often use fixed massage trajectories or a single intensity mode. They cannot adjust the care plan according to the actual condition of the user's back muscles, and often cannot accurately target the strained area. Not only is it difficult to achieve the ideal relief effect, but it may also cause unnecessary compression of healthy muscle tissue due to deviation in massage position and improper intensity, thus reducing the user experience.

[0004] Most back massagers have a simple support structure and fixing method, which cannot adjust the angle according to the user's usage scenario (such as standing to collect images or sitting to massage). This not only affects the fit between the back and the massage head during the massage, but may also lead to inaccurate detection data due to angle deviation during the image acquisition stage.

[0005] Meanwhile, the human-computer interaction of existing massagers is relatively simple, and parameters are mostly adjusted through physical buttons or basic displays on the device itself. Users cannot intuitively understand the real-time position of the massage head, key care areas, and massage effects. Summary of the Invention

[0006] The purpose of this invention is to provide a vision-assisted magnetic back massager to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a vision-assisted magnetic therapy back massager, comprising a massager body and a control module. The front of the massager body is provided with a cloth cover and a cotton pad, and a massage head is provided inside the cloth cover and cotton pad. Roll-up compartments are provided on both sides of the massager body. A camera component is provided at the top center of the massager body. Shoulder supports are provided on both sides of the camera component. The shoulder supports and the roll-up compartments are connected by a chest strap and a back strap. A support plate is provided on the back of the massager body.

[0008] The massage head is equipped with an infrared magnetic therapy stone inside;

[0009] The control module is arranged in the interior of the massage instrument main body, and is used for automatically dividing the back muscle partitions according to the user back image collected by the camera assembly before massage, identifying the strain and stiffness areas and marking the key care areas through skin color and muscle texture changes, driving the massage head to massage based on the visual recognition of the back muscle partitions, and automatically adjusting the pressure and magnetic therapy mode of the massage head.

[0010] Further, the top of the winding bin is provided with a winding motor, the inside of the winding bin is provided with a winding shaft, the output shaft of the winding motor is fixedly connected with one end of the winding shaft, one end of the chest binder is fixedly connected on the winding shaft, the first plug buckle is connected on the end of the chest binder away from the winding shaft, the two chest binders are detachably connected through the first plug buckle, and the top of the first plug buckle is provided with a connecting piece.

[0011] One end of the back strap is connected on the shoulder part, the second plug buckle is connected on the end of the back strap away from the shoulder part, and the back strap is detachably connected on the top of the connecting piece through the second plug buckle.

[0012] Further, the shoulder part is in an overall arc structure, one side of one end of the shoulder part is connected with the shoulder joint through a shoulder shaft pin, the shoulder joint is fixedly connected on the massage instrument main body, the other end of the shoulder part is connected with the back strap, and the inner side of the arc of the shoulder part is provided with a shoulder pad.

[0013] Further, the back of the massage instrument main body is provided with a containing groove and a plate groove, the support plate is arranged in the interior of the plate groove, the top of the support plate is provided with a connecting shaft, the two ends of the connecting shaft are connected in the interior of the containing groove through rotary bearings, two parallel guide grooves are arranged in the interior of the plate groove, and a guide sliding block is slidably arranged on the guide groove;

[0014] The inner side of the guide groove is provided with a gas cylinder, the top of the gas cylinder is provided with a vertically upward telescopic rod, and the end of the telescopic rod away from the gas cylinder is connected with the guide sliding block.

[0015] The outer side of the guide groove is provided with a connecting arm, the two ends of the connecting arm are respectively connected with a first joint and a second joint through shaft pins, the first joint is fixedly connected with the support plate, and the second joint is fixedly connected with the guide sliding block.

[0016] Further, the interior of the massage instrument main body is provided with a moving groove, the moving groove is arranged below the cloth cover cotton pad, and the massage head is movably arranged in the interior of the moving groove.

[0017] Further, the control module comprises a back state recognition unit, the back state recognition unit divides the back muscle partitions based on the back image, identifies the strain and stiffness areas and marks the key care areas, and the specific process comprises the following steps:

[0018] Obtaining an original image of a user's back collected by a camera assembly, performing denoising, white balance correction and edge enhancement preprocessing on the original image to obtain a clear back image;

[0019] Extracting skin color feature values and muscle texture feature information in the clear back image, and establishing a back image feature library;

[0020] Calling a preset human back muscle anatomy model, and performing feature matching on the feature data in the back image feature library and the anatomy model;

[0021] According to the matching result, combining the physiological distribution rule of back muscles, the user's back image is automatically divided into trapezius muscle partition, latissimus dorsi muscle partition, erector spinae muscle partition and rhomboid muscle partition, and a back muscle partition map is generated;

[0022] In the divided back muscle partition, skin color change data and muscle texture change data of each partition are extracted respectively;

[0023] Comparing the skin color change data with a preset normal skin color threshold range to screen out skin color abnormal areas; comparing the muscle texture change data with a preset normal texture feature template to screen out texture disorder areas;

[0024] Performing intersection analysis on the skin color abnormal areas and the texture disorder areas to determine a suspected strain stiffness area;

[0025] Obtaining user historical massage data and health record information, combining the position, area and feature difference of the suspected strain stiffness area, and calculating a strain stiffness coefficient of the area;

[0026] When the strain stiffness coefficient is greater than or equal to a preset coefficient threshold, the area is determined as a strain stiffness area, and based on the importance and strain stiffness degree of the area, a key nursing area is marked, and each key nursing area is assigned a corresponding nursing priority.

[0027] Further, the control module includes a massage driving unit, which drives the massage head to perform a massage action according to a preset rule, and the specific process includes:

[0028] Based on the divided back muscle partitions, a corresponding mapping relationship between the partitions and the massage heads is established, and a target massage head group corresponding to each muscle partition is determined;

[0029] According to the marked key nursing area and the nursing priority, a massage strategy database is queried to determine a target massage strategy; the muscle partition corresponding to the key nursing area with high nursing priority is preferentially executed;

[0030] According to the target massage strategy, driving instructions are sent to the target massage head group to control the target massage head group to move in the preset track in the corresponding muscle partition to achieve partition massage coverage, and the moving speed of the massage head in the key care area is 0.6-0.8 times that of the non-key area.

[0031] Further, the control module includes a massage data storage unit, which is used to record massage parameter adjustment data and user skin feedback data in real time during the massage process, and establish a massage effect database.

[0032] When the same user uses the number of times reaches the preset number of times threshold, the control module generates a personalized massage parameter and magnetotherapy mode recommendation model of the user based on historical data in the massage effect database;

[0033] In the subsequent massage process, the control module preferentially calls the personalized recommendation model to determine the initial massage parameter and magnetotherapy mode, and then combines the real-time collected back dynamic image to fine-tune the parameter.

[0034] Further, the control module includes a communication unit, which is used to be signal-connected with a user terminal, the user terminal including a mobile computer, a tablet or a mobile phone terminal program, and the user terminal is used to display the user back image in real time, and the current position of the massage head and the key care area are marked on the user back image by different colors during the massage process.

[0035] Further, the construction method of the massage strategy database includes:

[0036] Obtain multi-source data in a massage scene, the multi-source data including user back physiological feature data, user attribute data, historical massage strategy effect data, doctor professional strategy data and historical execution data;

[0037] The user back physiological feature data is subjected to semantic segmentation to be divided into four partitions of trapezius muscle, latissimus dorsi muscle, erector spinae muscle and rhomboid muscle, and each partition is assigned a unique partition ID; the physiological features of each partition are subjected to feature extraction and labeling processing to obtain a feature label combination of each partition; the partition ID, the feature label combination, the acquisition time stamp and the blockchain notarization hash value are stored by using a columnar database to establish a partition ID-feature label combination joint index to obtain a back partition feature base library;

[0038] The massage strategy is decomposed into three types of meta-units: magnetotherapy meta-unit, massage action meta-unit and time control meta-unit; and each meta-unit is assigned a unique meta-unit ID; based on the physician professional strategy data and historical execution data, the fitness of each meta-unit to the feature label is calculated, and a meta-unit-feature label fitness matrix is established; the compatibility between different meta-units is calculated, and a meta-unit-meta-unit compatibility matrix is established; the meta-unit ID, fitness matrix, compatibility matrix and meta-unit parameters are stored in a graph database to obtain a massage strategy meta-library, wherein the nodes are meta-units and the edges are adaptation / compatibility relationships, and the edge attributes are fitness / compatibility values;

[0039] The priority coefficient of each back sub-region is determined; the feature matching degree coefficient is calculated by performing feature matching on the feature label combination and the meta-unit combination; the historical effect value coefficient of the meta-unit combination is calculated based on the historical execution data and the physician professional strategy data; the total value score of the meta-unit combination is calculated based on the priority coefficient, the feature matching degree coefficient and the historical effect value coefficient; a feature label combination-meta-unit combination-total value score correlation is established and stored in a relational database to obtain a feature-strategy value correlation library;

[0040] A feature-strategy sample set matched by artificial is obtained, samples with complete feature labels and a total value score greater than or equal to a preset threshold in the sample set are screened, local correlation features of the feature labels are extracted using a CNN layer, global correlation of multi-sub-region features is processed by introducing sub-region priority attention weight through a Transformer layer, a double-loss function is constructed by combining value loss and conflict risk loss, and a massage strategy adaptation model is obtained based on training of the sample set;

[0041] A massage strategy database is constructed based on the back sub-region feature base library, the massage strategy meta-library, the feature-strategy value correlation library and the massage strategy adaptation model.

[0042] Further, the corresponding nursing priority is assigned to each key nursing area, including:

[0043] Based on three sub-dimensions of anatomical function, daily stress characteristics and historical strain probability of the user's back muscles, a basic priority score of the key nursing area of the user's back muscles is calculated, each sub-dimension is assigned a weight and a weighted sum is obtained to obtain a basic priority score P1 of each key nursing area of the user;

[0044] Based on the strain level, real-time visual features and real-time feedback of each key nursing area of the user, the basic priority score P1 of each key nursing area of the user is corrected to obtain a target priority score P2 of each key nursing area of the user;

[0045] The target priority score of each key nursing area of the user is calculated to obtain a plurality of target priority scores P2.

[0046] Calculate the difference of any two target priority scores P2 to obtain a plurality of target differences;

[0047] Compare the plurality of target differences with the preset difference threshold respectively, and determine the two key nursing areas corresponding to the target difference as the conflict key nursing areas when the target difference is less than or equal to the preset difference threshold;

[0048] Conflict resolution is performed on the conflict key nursing area; wherein the conflict resolution includes at least one of time conflict, parameter conflict and position conflict;

[0049] The target priority score P2 of all key nursing areas is queried in the priority score-priority level table to determine the corresponding nursing priority of each key nursing area.

[0050] Compared with the prior art, the beneficial effects of the present application are:

[0051] 1. The camera assembly of the present application collects back images through a high-definition camera and an infrared thermal imaging sensor, a control module divides the partition in combination with a human back muscle dissection model, identifies the strained and stiff areas through skin color and muscle texture comparison and marks the key nursing areas, and the massage driving unit schedules the massage heads according to the partition mapping relationship and the nursing priority, the key area massage head moves at a slower speed to enhance the penetration force, and the infrared magnetic therapy stone acts synchronously, the image is collected again after massage to detect the effect, the data storage unit accumulates historical data to generate a personalized recommendation model, which upgrades from blind massage to precise nursing, can automatically adapt to the real-time state of the user's back, and can optimize the scheme with the number of uses, greatly improves the massage targeting and long-term nursing effect, and meets the needs of different users for strain relief.

[0052] 2. The support plate of the present application is driven by a pneumatic cylinder, a telescopic rod and a connecting arm, can adjust the opening and closing angle around the connecting shaft, can be adjusted to a vertical angle before massage to ensure accurate imaging, can be adjusted to a tilt angle according to the sitting and standing posture during massage to make the back fit the cloth cover and the cotton pad, the arched shoulder part can fit the shoulder pad and can adjust the angle, the chest strap is adjusted in length by a winding motor and fixed by a plug buckle, the back strap and the connecting piece are inserted and connected to assist positioning, double fixation avoids displacement during massage, can adapt to multiple scenes such as image collection, massage and storage, and can meet the wearing needs of users of different body types, ensure stable and deviation-free massage process, uniform stress and comfort, and improve the convenience and comfort of use.

[0053] 3. The back state recognition unit of the control module of the application accurately analyzes the image generation partition map and the key care area, the data storage unit records the massage parameters and skin feedback data in real time, and the back image is displayed in real time and the massage head position and key care area are marked with different colors, so that the user does not need to rely on subjective feeling to judge the massage state, and can directly monitor the massage area and key care part through the terminal, and can adjust in time if the position deviation is found, and the terminal supports manual adjustment of massage strength and magnetic therapy mode, combined with the personalized recommendation model of the control module, the initial scheme has been highly adapted to the user's demand, and the fine adjustment is more convenient, so that the user can control the whole massage process and improve the operation experience and the sense of trust in the product. BRIEF DESCRIPTION OF DRAWINGS

[0054] Figure 1 It is a schematic diagram of the whole structure of the massage instrument main body of the application;

[0055] Figure 2 It is a schematic diagram of the back structure of the massage instrument main body of the application;

[0056] Figure 3 It is a schematic diagram of the plate groove structure of the application;

[0057] Figure 4 It is a schematic diagram of the support plate structure of the application;

[0058] Figure 5 It is a schematic diagram of the internal structure of the massage instrument main body of the application;

[0059] Figure 6 It is a schematic diagram of the shoulder joint structure of the application;

[0060] Figure 7 It is a schematic diagram of the control module of the application.

[0061] In the figure: 1, massage instrument main body; 2, cloth cover cotton pad; 3, winding bin; 4, chest strap; 5, winding motor; 6, first plug buckle; 7, connecting piece; 8, camera assembly; 9, shoulder joint; 10, shoulder joint; 11, shoulder strap; 12, second plug buckle; 13, containing groove; 14, connecting shaft; 15, support plate; 16, plate groove; 17, guide groove; 18, first joint; 19, connecting arm; 20, second joint; 21, guide sliding block; 22, telescopic rod; 23, air cylinder; 24, moving groove; 25, massage head; 26, shoulder pad; 27, shoulder shaft pin. DETAILED DESCRIPTION

[0062] With reference to the drawings of the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described, obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work belong to the scope of protection of the present application.

[0063] Please refer to Figures 1-7 The present application provides the following technical solutions:

[0064] The visual aid-based magnetic therapy back massage instrument comprises a massage instrument main body 1 and a control module, the front of the massage instrument main body 1 is provided with a cloth cover cotton pad 2, the inside of the cloth cover cotton pad 2 is provided with a massage head 25, both sides of the massage instrument main body 1 are provided with a rolling bin 3, the middle of the top of the massage instrument main body 1 is provided with a camera assembly 8, the camera assembly 8 comprises a high-definition camera, an infrared thermal imaging sensor and a fill light, both sides of the camera assembly 8 are provided with a shoulder part 10, the shoulder part 10 and the rolling bin 3 are connected through a chest strap 4 and a back strap 11, and the back of the massage instrument main body 1 is provided with a support plate 15.

[0065] The inside of the massage head 25 is provided with an infrared magnetic therapy stone;

[0066] The control module is arranged in the inside of the massage instrument main body 1, the control module is used for automatically dividing the back muscle zones according to the user back image collected by the camera assembly 8 before massage, identifying the strain and stiffness areas and marking the key care areas through skin color and muscle texture changes, driving the massage head 25 to massage based on the visual recognition of the back muscle zones, and automatically adjusting the pressure and magnetic therapy mode of the massage head 25.

[0067] In the above embodiment, before massage, the massage instrument main body 1 is supported and erected through the support plate 15, the user back image is collected by the camera assembly 8 through the user terminal, the control module divides the back muscle zones according to the user back image, identifies the strain and stiffness areas and marks the key care areas through skin color and muscle texture changes, and the pressure and magnetic therapy mode of the massage head 25 are personalized adjusted according to the user's back condition, and meanwhile the user can also adjust the massage and magnetic therapy mode through the user terminal.

[0068] In the above embodiment, after the massage and magnetic therapy mode is determined, the user places the massage instrument main body 1 behind the back, fixes the massage instrument main body 1 on the shoulder through the shoulder part 10, connects through the chest strap 4 and the back strap 11, completes the positioning of the massage instrument main body 1, at this time, the massage and magnetic therapy work is started through the user terminal, after the massage is completed, the user back image can be collected again through the camera assembly 8 controlled by the user terminal, the condition after the back massage is detected, and in the subsequent massage, the first personalized adjustment data can be directly called or the back image can be collected again for customization.

[0069] In the above embodiment, through visual acquisition, intelligent analysis to personalized adaptation, breaking the traditional massage instrument one-size-fits-all limitations, before massage, with the aid of camera components to collect back image, control module automatically divides muscle partition, can also accurately identify the labor-stiffness area through skin color and muscle texture and mark the key care area, so that massage is no longer blind, while supporting user terminal bidirectional interaction, both meet the intelligent needs of automatically adjusting massage head pressure and magnetic therapy mode, and retain the autonomy of manual adjustment, both professional and flexible, after massage, secondary image acquisition detects the effect, can also store personalized data or re-customize, continuously adapt to the user's back state changes, greatly improve the massage pertinence and long-term care effect.

[0070] The top of the winding bin 3 is provided with a winding motor 5, the inside of the winding bin 3 is provided with a winding shaft, the output shaft of the winding motor 5 is fixedly connected with one end of the winding shaft, one end of the chest strap 4 is fixedly connected on the winding shaft, the first plug buckle 6 is connected on the end of the chest strap 4 away from the winding shaft, the two chest straps 4 are detachably connected through the first plug buckle 6, and the top of the first plug buckle 6 is provided with a connecting piece 7;

[0071] One end of the back strap 11 is connected on the shoulder part 10, the second plug buckle 12 is connected on the end of the back strap 11 away from the shoulder part 10, and the back strap 11 is detachably inserted on the top of the connecting piece 7 through the second plug buckle 12;

[0072] The shoulder part 10 is in an overall arc structure, one side of one end of the shoulder part 10 is connected with the shoulder joint 9 through the shoulder shaft pin 27, the shoulder joint 9 is fixedly connected on the massage instrument main body 1, the other side of the same end of the shoulder part 10 is detachably clamped on the shoulder joint 9 through the spring buckle, the other end of the shoulder part 10 is connected with the back strap 11, and the inner side of the arc of the shoulder part 10 is provided with a shoulder pad 26.

[0073] In the above embodiment, when fixing the massage instrument main body 1, the overall arc structure of the shoulder part 10 is placed on the shoulders, the shoulder pad 26 is in contact with the shoulders, the chest straps 4 are converged at the chest from the left and right sides, and the chest straps 4 are fixed through the first plug buckle 6, then the back strap 11 at the bottom of the shoulder part 10 on the shoulders is pulled down and inserted on the top of the connecting piece 7 through the second plug buckle 12, the positioning is completed, displacement during massage and magnetic therapy is avoided, the massage position is consistent with the expected position, and the massage and magnetic therapy effects are affected.

[0074] In the above embodiment, the arched structure of the shoulder part 10 fits the shoulder curve, and the soft shoulder pad 26 distributes pressure to avoid shoulder discomfort. The double fixing system of the chest strap 4 and the strap 11 can adjust the length by the winding motor 5 to meet the needs of users of different body types. The combination of the first plug buckle 6, the second plug buckle 12, and the connecting piece 7 realizes quick disassembly and stable connection, ensures that the instrument always fits the back during massage, avoids displacement of the massage position, guarantees precise force on the key care area, simplifies the wearing process, improves the convenience of use, and enables users to enjoy massage stably in different postures such as standing and sitting.

[0075] The back of the massage instrument main body 1 is provided with a containing groove 13 and a plate groove 16. The support plate 15 is arranged in the plate groove 16. The top of the support plate 15 is provided with a connecting shaft 14. The two ends of the connecting shaft 14 are connected in the containing groove 13 through rotary bearings. The inside of the plate groove 16 is provided with two parallel guide grooves 17. The guide grooves 17 are slidably provided with guide sliding blocks 21.

[0076] The inside of the guide groove 17 is provided with a gas cylinder 23. The top of the gas cylinder 23 is provided with a vertically upward telescopic rod 22. The end of the telescopic rod 22 away from the gas cylinder 23 is connected with the guide sliding block 21.

[0077] The outside of the guide groove 17 is provided with a connecting arm 19. The two ends of the connecting arm 19 are respectively connected with a first joint 18 and a second joint 20 through shaft pins. The first joint 18 is fixedly connected with the support plate 15. The second joint 20 is fixedly connected with the guide sliding block 21.

[0078] In the above embodiment, the driving gas cylinder 23 drives the telescopic rod 22 to extend and retract, and drives the guide sliding block 21 to slide along the guide groove 17, so as to push the support plate 15 to open and close around the connecting shaft 14, replacing the traditional fixed support structure. Users can flexibly adjust the angle according to the use scene. For example, when collecting images before massage, the massage instrument main body 1 is adjusted to a higher vertical angle by the support plate 15 to ensure that the instrument stands stably, and the shooting angle of the camera assembly 8 is accurate. During massage, the support plate is adjusted to a comfortable inclined angle according to different postures such as sitting and standing, so that the back is more fitted to the cloth cover cotton pad, and the uniformity of massage stress is improved.

[0079] The inside of the massage instrument main body 1 is provided with a moving groove 24. The moving groove 24 is arranged below the cloth cover cotton pad 2. The massage head 25 is movably arranged in the inside of the moving groove 24. The inside of the massage instrument main body 1 is provided with a driving part for driving the massage head 25 to move, rotate and press.

[0080] In the above embodiment, when the user massages, the back is close to the cloth cover cotton pad 2 on the front of the massage instrument main body 1, and the massage head 25 is driven to move, rotate and press during work, so that the user's back is synchronously massaged and magnetically treated, the infrared magnetic therapy stone and the massage action are synchronously acted, the muscles are relaxed by mechanical massage, the blood circulation of the back is promoted by magnetic therapy, the muscle strain is relieved, the dual effects of physical massage and magnetic therapy nursing are realized, the deep muscle layer is deeply acted, the precise relief is realized for the strained area, and the overall nursing effect is improved.

[0081] The control module comprises a back state recognition unit, which divides the back muscle partitions based on the back image and recognizes the strained and stiff areas and marks the key nursing areas, and the specific process comprises:

[0082] An original image of the user's back collected by the camera assembly is acquired, and the original image is preprocessed by noise reduction, white balance correction and edge enhancement to obtain a clear back image;

[0083] Skin color feature values and muscle texture feature information in the clear back image are extracted, and a back image feature library is established;

[0084] A preset human back muscle anatomical structure model is called, and the feature data in the back image feature library is matched with the anatomical structure model;

[0085] According to the matching result, the user's back image is automatically divided into trapezius muscle partitions, latissimus dorsi muscle partitions, erector spinae muscle partitions and rhomboid muscle partitions in combination with the physiological distribution law of back muscles, and a back muscle partition map is generated;

[0086] Skin color change data and muscle texture change data of each partition are extracted in the divided back muscle partitions;

[0087] The skin color change data are compared with a preset normal skin color threshold range, and the skin color abnormal area is screened out; the muscle texture change data are compared with a preset normal texture feature template, and the texture disorder area is screened out;

[0088] The skin color abnormal area and the texture disorder area are analyzed by intersection, and a suspected strained and stiff area is determined;

[0089] User historical massage data and health record information are acquired, and the position, area and feature difference of the suspected strained and stiff area are combined to calculate the strained and stiff coefficient of the area;

[0090] When the strained and stiff coefficient is greater than or equal to a preset coefficient threshold, the area is determined as a strained and stiff area, and a key nursing area is marked based on the importance and strained and stiff degree of the area, and each key nursing area is allocated a corresponding nursing priority.

[0091] In the above embodiment, the back state recognition unit extracts skin color and muscle texture features, matches with the human back muscle anatomy model, realizes scientific division of muscle partitions, makes the massage more suitable for human physiological structure, screens abnormal areas by comparing the normal threshold of skin color and texture, and performs intersection analysis to preliminarily lock the suspected strain area. In combination with user historical data, the strain stiffness coefficient is calculated to finally accurately determine the strain area and mark the key nursing area. The nursing priority is also allocated to ensure the accuracy of strain area recognition and the rationality of nursing priority, and to provide accurate guidance for subsequent massage driving.

[0092] The control module includes a massage driving unit that drives the massage head to perform a massage action according to a preset rule. The specific process includes:

[0093] Based on the divided back muscle partitions, a corresponding mapping relationship between the partitions and the massage heads is established to determine the target massage head group corresponding to each muscle partition.

[0094] According to the marked key nursing area and the nursing priority, the massage strategy database is queried to determine the target massage strategy. The muscle partition corresponding to the key nursing area with high nursing priority is preferentially executed.

[0095] According to the target massage strategy, a driving instruction is sent to the target massage head group to control the target massage head group to move in the corresponding muscle partition according to a preset trajectory, realize partition massage coverage, and the massage head moving speed of the key nursing area is 0.6-0.8 times that of the non-key area.

[0096] In the above embodiment, the massage driving unit realizes efficient allocation of massage resources by partition mapping combined with priority scheduling, guarantees the massage effect of the key nursing area, establishes the corresponding mapping relationship between the muscle partitions and the massage heads, avoids chaotic massage head action, ensures that each partition has a corresponding target massage head group responsible for it, realizes accurate partition coverage, queries the massage strategy database according to the priority of the key nursing area to determine the target massage strategy, improves the nursing efficiency, and adjusts the massage head moving speed for the key nursing area. The slow moving speed of 0.6-0.8 times allows the massage head to stay in the area for a longer time and have a deeper effect, enhancing the massage intensity and penetration effect, while the non-key area maintains normal speed, balancing overall coverage and key deep nursing. This design makes the massage process more logical, ensures that all muscle partitions can be nursed, highlights the key points, and improves the relief effect of the strain area.

[0097] The control module includes a massage data storage unit for recording massage parameter adjustment data and user skin feedback data in real time during the massage process, and establishing a massage effect database.

[0098] When the same user uses the number of times reaches the preset number threshold, the control module generates a personalized massage parameter and a magnetic therapy mode recommendation model of the user based on historical data in the massage effect database;

[0099] In the subsequent massage process, the control module preferentially calls the personalized recommendation model to determine initial massage parameters and a magnetic therapy mode, and then combines the real-time collected back dynamic image to fine-tune the parameters.

[0100] In the above embodiment, the massage data storage unit records massage parameter adjustment data and skin feedback data in real time, establishes a massage effect database, and provides rich data support for personalized analysis. When the number of uses reaches the standard, a personalized recommendation model is generated based on historical data to replace the initial general parameters, so that the initial scheme of subsequent massage has high adaptability. The recommendation model is preferentially called in subsequent massage, and the parameters are fine-tuned in combination with real-time dynamic images, avoiding the tediousness of reanalyzing each massage, while ensuring that the scheme always fits the current back condition of the user. With the increase of the number of uses, the recommendation model becomes more accurate, greatly improving the satisfaction and dependence of long-term use of the user.

[0101] The control module includes a communication unit for signal connection with a user terminal, and the user terminal includes a mobile computer, a tablet or a mobile phone terminal program. The user back image is displayed in real time through the user terminal, and the current position of the massage head and the key care area are marked on the user back image by different colors during the massage process.

[0102] In the above embodiment, the communication unit is connected with multiple types of user terminals, and the user can operate on common devices such as mobile phones and tablets, breaking the operation scene limit and improving the convenience. The back image is displayed in real time, and the position of the massage head and the key care area are marked by different colors, so that the user can clearly understand the current massage area and the key care part, realize the visual monitoring of the massage process, and intuitively judge whether the massage covers the target area according to the information displayed on the terminal. If there is deviation, it can be adjusted in time to avoid blind massage. The visual interface also makes parameter adjustment more intuitive, and the user can accurately adjust the massage intensity, magnetic therapy mode, etc. in combination with image feedback, improve the operation experience, and enhance the user's sense of control over the massage process.

[0103] The construction method of the massage strategy database includes:

[0104] Obtain multi-source data in a massage scene, and the multi-source data includes user back physiological feature data, user attribute data, historical massage strategy effect data, doctor professional strategy data and historical execution data;

[0105] perform semantic segmentation on the user's back physiological feature data, divide it into four partitions of trapezius muscle, latissimus dorsi muscle, erector spinae muscle and rhomboid muscle, and assign a unique partition ID to each partition; perform feature extraction and labeling on the physiological features of each partition to obtain a feature label combination for each partition; store the partition ID, feature label combination, collection timestamp and blockchain storage hash value in a columnar database, establish a joint index of partition ID-feature label combination, and obtain a back partition feature base library;

[0106] The massage strategy is decomposed into three types of meta-units: magnetic therapy meta-unit, massage action meta-unit and time control meta-unit; each meta-unit is assigned a unique meta-unit ID; based on the physician professional strategy data and historical execution data, the adaptation degree of each meta-unit to the feature label is calculated, and a meta-unit-feature label adaptation matrix is established; the compatibility between different meta-units is calculated, and a meta-unit-meta-unit compatibility matrix is established; the meta-unit ID, adaptation matrix, compatibility matrix and meta-unit parameters are stored in a graph database to obtain a massage strategy meta library, wherein the nodes are meta-units and the edges are adaptation / compatibility relationships, and the edge attributes are adaptation degree / compatibility values;

[0107] Determine the priority coefficient of each back partition; perform feature matching on the feature label combination and meta-unit combination to calculate a feature matching degree coefficient; based on the historical execution data and physician professional strategy data, calculate the historical effect value coefficient of the meta-unit combination; based on the priority coefficient, feature matching degree coefficient and historical effect value coefficient, calculate the total value score of the meta-unit combination; establish a feature label combination-meta-unit combination-value total score correlation, store it in a relational database, and obtain a feature-strategy value correlation library;

[0108] Obtain an artificial matching feature-strategy sample set, select samples in the sample set with complete feature labels and a total value score greater than or equal to a preset threshold, extract local correlation features of the feature labels using a CNN layer, introduce partition priority attention weight processing of global correlation of multi-partition features through a Transformer layer, combine value loss and conflict risk loss to construct a double-loss function, and train a massage strategy adaptation model based on the sample set;

[0109] Based on the back partition feature base library, massage strategy meta library, feature-strategy value correlation library and massage strategy adaptation model, a massage strategy database is constructed.

[0110] In this embodiment, the semantic segmentation uses the MaskR-CNN algorithm, and the physiological feature data needs to include infrared thermal imaging or RGB images as the segmentation input source.

[0111] In this embodiment, the physiological features include infrared temperature, muscle thickness, texture density and skin color redness ratio.

[0112] In this embodiment, the physiological characteristics of each partition are subjected to feature extraction and labeling, i.e., for example, infrared temperature 37.1-37.8°C is labeled as T2 (moderate strain), and texture density 15-20 times / cm2 is labeled as D2 (moderate tension).

[0113] In this embodiment, the columnar database includes but is not limited to ClickHouse.

[0114] In this embodiment, the magnetic therapy element unit includes, for example, a magnetic therapy strength of 1-5 levels, a frequency of 50-200 Hz, and safety constraints; the massage action element unit includes, for example, kneading / pressing / rolling / tapping techniques, a pressure of 20-50 N, and a frequency of 20-60 times / minute; and the time control element unit includes, for example, a single partition duration of 3-15 minutes and a multi-partition execution sequence.

[0115] In this embodiment, the adaptation degree = physician recommended adaptation score x 0.6 + historical execution effect score x 0.4; and the compatibility = hardware action compatibility x 0.7 + effect synergy compatibility x 0.3.

[0116] In this embodiment, the graph database includes but is not limited to Neo4j.

[0117] In this embodiment, the feature matching degree coefficient is calculated, i.e., feature matching degree coefficient = Σ (single feature label adaptation degree x feature weight) / feature total number, wherein the feature weight is based on physician professional strategy data statistics (e.g., high-frequency feature label weight is set to 1.0, and low-frequency feature label weight is set to 0.7), and the single feature label adaptation degree takes the maximum value of all element unit adaptation degrees for the feature label in the element unit combination.

[0118] In this embodiment, the historical effect value coefficient of the element unit combination is calculated, i.e., historical effect value coefficient = (physician score x 1.0 + user feedback average score x 0.8 + physiological improvement value x 0.6) / (1.0 + 0.8 + 0.6), wherein the physiological improvement value = (texture density after massage - texture density before massage) / texture density before massage x 100%.

[0119] In this embodiment, the total value score of the element unit combination is calculated, i.e., total value score = priority coefficient x feature matching degree coefficient x historical effect value coefficient, normalized to 0-100 points (normalization method: total value score = (original value / maximum historical value) x 100).

[0120] In this embodiment, the element unit combination generation method is as follows: based on the element unit-element unit compatibility matrix, compatible element unit combinations are generated through depth-first search, and combinations with compatibility <0.5 are filtered.

[0121] In this embodiment, the conflict risk score calculation method is: conflict risk score = (1-average compatibility of element unit combination) x 100, wherein average compatibility = Σ (compatibility between element units in combination) / number of element unit pairs in combination.

[0122] In this embodiment, a manually matched feature-strategy sample set is obtained, which includes input features (feature label combination + partition priority) and output labels (element unit combination + total value score + conflict risk score); samples with complete feature labels, total value scores ≥ 70 points and conflict risk scores ≤ 30 points in the sample set are screened as effective training samples; a CNN+Transformer+fully connected layer hybrid model structure is constructed: a CNN layer (3x3 convolution kernel, 32 channels) extracts local correlation features of the feature label combination; a Transformer layer (2 layers of encoder, 4 heads of attention) introduces partition priority attention weights (attention weight = partition priority coefficient / Σ each partition priority coefficient) to process the global correlation of multi-partition features; a fully connected layer (64→32→output) outputs the candidate element unit combination, total value score prediction value and conflict risk score prediction value; a double loss function (L = 0.7 x (total value score real value - value prediction value) 2 + 0.3 x (conflict risk score real value - conflict risk prediction value) 2) is constructed, an Adam optimizer (initial learning rate 0.001, decay 10% every 5 epochs) is used, and the model is trained based on the effective training samples; the training is stopped when the validation set loss is < 0.05 and the test set accuracy (Top3 candidate proportion containing physician recommended strategy) is ≥ 95%, and a massage strategy adaptation model is obtained.

[0123] The working principle and beneficial effects of the above technical solution are: different data is stored in multiple databases for classification, the advantages of each are brought into play, indexes and association relationships are established, and efficient retrieval and management are facilitated; the element unit adaptation degree and compatibility are quantified, and the strategy value is evaluated comprehensively based on multiple factors to provide a scientific basis for strategy formulation; a machine learning model is used, combined with a double loss function training, to improve the strategy adaptation accuracy and reduce the conflict risk.

[0124] The method comprises the following steps:

[0125] Based on three sub-dimensions of anatomical function, daily stress characteristics and historical strain probability of the user's back muscles, a basic priority score of the key care area of the user's back muscles is calculated, each sub-dimension is assigned a weight, and a weighted sum is obtained to obtain a basic priority score P1 of each key care area of the user.

[0126] The target priority score P2 of each key nursing area of the user is corrected based on the strain level of each key nursing area of the user, real-time visual features and real-time feedback of the user, to obtain the target priority score P2 of each key nursing area of the user.

[0127] The target priority score P2 of each key nursing area of the user is corrected based on the strain level of each key nursing area of the user, real-time visual features and real-time feedback of the user, to obtain the target priority score P2 of each key nursing area of the user.

[0128] The target priority score P2 of each key nursing area of the user is corrected based on the strain level of each key nursing area of the user, real-time visual features and real-time feedback of the user, to obtain the target priority score P2 of each key nursing area of the user.

[0129] The target priority score P2 of each key nursing area of the user is corrected based on the strain level of each key nursing area of the user, real-time visual features and real-time feedback of the user, to obtain the target priority score P2 of each key nursing area of the user.

[0130] The target priority score P2 of each key nursing area of the user is corrected based on the strain level of each key nursing area of the user, real-time visual features and real-time feedback of the user, to obtain the target priority score P2 of each key nursing area of the user.

[0131] The target priority score P2 of each key nursing area of the user is corrected based on the strain level of each key nursing area of the user, real-time visual features and real-time feedback of the user, to obtain the target priority score P2 of each key nursing area of the user.

[0132] In this embodiment, the anatomical function sub-dimension is used to quantify the protection of the key nursing area on the core function of the human body, the daily stress sub-dimension is used to quantify the stress frequency and intensity in the daily activities of the key nursing area, and the historical strain sub-dimension is used to quantify the probability of historical strain of the key nursing area, and the weight sum of the three sub-dimensions is 1.

[0133] In this embodiment, the strain emergency correction coefficient K1 is obtained according to the strain level and the visual features, and then the real-time feedback correction coefficient K2 is obtained according to the real-time state of the user fed back through the auxiliary sensor, and the target priority score P2 is calculated through the formula P2=P1×K1×K2.

[0134] In this embodiment, the conflict resolution only acts on the resource allocation strategy without modifying the priority score.

[0135] In this embodiment, the time conflict is resolved by proportionally allocating the massage time length according to the P2 scores of the key nursing areas, and the massage time length of the high-priority partition does not exceed 60% of the total massage time length; the parameter conflict is resolved by fixing the magnetic therapy parameters of the high-priority partition and adapting the magnetic therapy parameters of the low-priority partition; and the position conflict is resolved by calculating the position overlap degree of the key nursing areas and determining the massage order or time-segmented massage in combination with the P2 scores, and the conflict resolution process does not change the target priority score P2 of the key nursing area.

[0136] In this embodiment, the conflict resolution process performs the following resource allocation strategies:

[0137] (1) Time conflict resolution: The actual allocation duration of each intensive care area is calculated according to the formula wherein represents the actual allocation duration of the ith conflict intensive care area; represents the total massage duration; represents the target priority score P2 of the ith conflict intensive care area; n represents the total number of conflict intensive care areas;

[0138] (2) Parameter conflict resolution: The conflict areas are sorted in descending order of P2, and the magnetotherapy parameters of the highest priority partition are fixed , and the parameter adaptation formula is used for low-priority partitions; wherein represents the adjusted magnetotherapy intensity parameter; represents the reference magnetotherapy intensity parameter; represents the target priority score of the low-priority partition; represents the target priority score of the high-priority partition;

[0139] (3) Position conflict resolution: The position overlap degree of any two conflict intensive care areas is calculated When > 0.3, the massage order is determined in descending order of P2 score; when ≤ 0.3, synchronous massage is allowed; wherein all conflict resolution strategies take the target priority score P2 as input, output the resource allocation scheme, and the P2 value remains unchanged during the conflict resolution process.

[0140] The working principle and beneficial effects of the technical scheme are as follows: the basic priority score is calculated based on three sub-dimensions of anatomical function, daily stress characteristics and historical strain probability, the physiological characteristics and use of the back muscles are comprehensively considered, the calculation of the basic priority is more scientific, and the degree of attention of each key nursing area can be preliminarily reflected; the basic priority score is corrected in combination with the strain grade, real-time visual features and user real-time feedback, the current actual situation of the user is fully considered, the accuracy of the priority evaluation of the key nursing area is further improved, and the physiological priority can be fitted to the current needs of the user; by calculating the target priority score difference value and comparing it with the preset threshold value, the conflict key nursing area can be effectively identified, the nursing resource allocation is not unreasonable or the nursing strategy conflicts in the case of similar priority, and the rationality of the nursing arrangement is ensured; for the identified conflict key nursing area, the conflicts in time, parameters, position and the like are eliminated, the allocation and utilization of the nursing resource are optimized, the smoothness and coordination of the nursing process are improved, and the nursing strategy can be smoothly implemented; the nursing priority is determined by querying the priority score-priority grade table, the standardization of the priority allocation is realized, the nursing staff can quickly and accurately carry out the nursing work according to the priority, and the nursing efficiency is improved.

[0141] The above is only the preferred specific embodiment of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can make equivalent replacement or change according to the technical solution and the inventive concept of the present application within the technical range disclosed by the present application, which should be covered in the protection scope of the present application.

Claims

1. A visual aid-based magnetic back massager, comprising a massager body (1) and a control module, characterized in that, The front of the massage instrument main body (1) is provided with a cloth cover cotton pad (2), the inside of the cloth cover cotton pad (2) is provided with a massage head (25), both sides of the massage instrument main body (1) are provided with a rolling bin (3), the middle of the top of the massage instrument main body (1) is provided with a camera assembly (8), both sides of the camera assembly (8) are provided with a shoulder part (10), the shoulder part (10) and the rolling bin (3) are connected through a chest strap (4) and a back strap (11), the back of the massage instrument main body (1) is provided with a supporting plate (15); The inside of the massage head (25) is provided with an infrared magnetic therapy stone; The control module is arranged in the inside of the massage instrument main body (1), the control module is used for automatically dividing the back muscle partition according to the user back image collected by the camera assembly (8) before massage, identifying the strain and stiffness area and marking the key nursing area through skin color and muscle texture change, assigning corresponding nursing priority to each key nursing area based on the anatomical function of user back muscle, daily stress characteristics and historical strain probability, constructing a massage strategy database, driving the massage head (25) to massage based on the back muscle partition of visual recognition, and automatically adjusting the pressure and magnetic therapy mode of the massage head (25); The control module includes a back state recognition unit, the back state recognition unit divides the back muscle partition based on the back image and identifies the strain and stiffness area and marks the key nursing area, and the specific process includes: Obtaining the original image of the user back collected by the camera assembly, pre-processing the original image to obtain a clear back image through noise reduction, white balance correction and edge enhancement; Extracting the skin color feature value and muscle texture feature information in the clear back image, and establishing a back image feature library; Calling a preset human back muscle anatomical structure model, performing feature matching on the feature data in the back image feature library and the anatomical structure model; According to the matching result, combining the physiological distribution law of back muscle, automatically dividing the user back image into trapezius muscle partition, latissimus dorsi muscle partition, erector spinae muscle partition and rhomboid muscle partition, and generating a back muscle partition atlas; Respectively extracting the skin color change data and muscle texture change data in each partition in the divided back muscle partition; Comparing the skin color change data with a preset normal skin color threshold range to screen out abnormal skin color areas; comparing the muscle texture change data with a preset normal texture feature template to screen out texture disorder areas; Performing intersection analysis on the abnormal skin color areas and the texture disorder areas to determine suspected strain stiffness areas; Obtaining user historical massage data and health record information, combining the position, area and feature difference of the suspected strain stiffness area, and calculating the strain stiffness coefficient of the area; When the strain stiffness coefficient is greater than or equal to a preset coefficient threshold, the area is determined as a strain stiffness area, and a key nursing area is marked based on the importance and strain stiffness degree of the area, and a corresponding nursing priority is assigned to each key nursing area; The construction method of the massage strategy database includes: Acquire multi-source data in the massage scenario, including user back physiological characteristic data, user attribute data, historical massage strategy effect data, physician professional strategy data, and historical execution data; The user's back physiological feature data is semantically segmented into four major regions: trapezius, latissimus dorsi, erector spinae, and rhomboid muscles, and a unique region ID is assigned to each region. The physiological features of each region are extracted and labeled to obtain the feature label combination of each region. The region ID, feature label combination, collection timestamp, and blockchain evidence hash value are stored in a columnar database to establish a joint index of region ID and feature label combination, thus obtaining the back region feature base library. Massage strategies are decomposed into three types of meta-units: magnetic therapy meta-units, massage action meta-units, and time control meta-units; and a unique meta-unit ID is assigned to each meta-unit. Based on the physician's professional strategy data and historical execution data, the fit degree between each meta-unit and feature tag is calculated, and a meta-unit-feature tag fit degree matrix is ​​established; the compatibility between different meta-units is calculated, and a meta-unit-meta-unit compatibility matrix is ​​established; the meta-unit IDs, fit degree matrices, compatibility matrices, and meta-unit parameters are stored in a graph database to obtain a massage strategy meta-library, where nodes are meta-units, edges represent fit / compatibility relationships, and edge attributes are fit degree / compatibility values; Determine the priority coefficient of each back partition; perform feature matching on the feature label combination and meta-unit combination, and calculate the feature matching degree coefficient; calculate the historical effect value coefficient of the meta-unit combination based on the historical execution data and physician professional strategy data; calculate the total value score of the meta-unit combination based on the priority coefficient, feature matching degree coefficient and historical effect value coefficient; establish the correlation relationship between feature label combination - meta-unit combination - total value score, and store it in a relational database to obtain the feature-strategy value correlation library; Obtain a manually matched feature-policy sample set, filter samples in the sample set that have complete feature labels and a total value score ≥ a preset threshold, use a CNN layer to extract local correlation features of feature labels, introduce partition priority attention weights through a Transformer layer to process the global correlation of multi-partition features, construct a dual loss function by combining value loss and conflict risk loss, and train a massage strategy adaptation model based on the sample set. A massage strategy database is constructed based on the back partition feature base library, massage strategy meta-library, feature-strategy value association library, and massage strategy adaptation model.

2. The visual aid-based magnetic therapy back massager according to claim 1, wherein, The top of the winding chamber (3) is provided with a winding motor (5), and the inside of the winding chamber (3) is provided with a winding shaft. The output shaft of the winding motor (5) is fixedly connected to one end of the winding shaft. One end of the chest strap (4) is fixedly connected to the winding shaft. A first buckle (6) is connected to the end of the chest strap (4) away from the winding shaft. The two chest straps (4) are detachably connected through the first buckle (6). A connector (7) is provided on the top of the first buckle (6). One end of the shoulder strap (11) is connected to the shoulder piece (10), and the other end of the shoulder strap (11) is connected with the second plug buckle (12), and the shoulder strap (11) is detachably connected to the top of the connecting piece (7) through the second plug buckle (12); The shoulder piece (10) is in an overall arc structure, one side of one end of the shoulder piece (10) is connected with the shoulder joint (9) through a shoulder shaft pin (27), the shoulder joint (9) is fixedly connected to the massage instrument main body (1), the other end of the shoulder piece (10) is connected with the shoulder strap (11), and the inner side of the arc of the shoulder piece (10) is provided with a shoulder pad (26).

3. The visual aid-based magnetic back massager of claim 1, wherein, The back of the massage instrument main body (1) is provided with a containing groove (13) and a plate groove (16), the support plate (15) is arranged in the plate groove (16), the top of the support plate (15) is provided with a connecting shaft (14), the two ends of the connecting shaft (14) are connected in the containing groove (13) through rotating bearings, and two guide grooves (17) are arranged in parallel in the plate groove (16). The inner side of the guide groove (17) is provided with a gas cylinder (23), the top of the gas cylinder (23) is provided with a vertically upward telescopic rod (22), and one end of the telescopic rod (22) away from the gas cylinder (23) is connected with the guide sliding block (21). The outer side of the guide groove (17) is provided with a connecting arm (19), the two ends of the connecting arm (19) are respectively connected with a first joint (18) and a second joint (20) through shaft pins, the first joint (18) is fixedly connected with the support plate (15), and the second joint (20) is fixedly connected with the guide sliding block (21).

4. The visual aid-based magnetic therapy back massager of claim 1, wherein, The inside of the massage instrument main body (1) is provided with a moving groove (24), the moving groove (24) is arranged below the cloth cover cotton pad (2), and the massage head (25) is movably arranged in the moving groove (24).

5. The visual aid-based magnetic therapy back massager according to claim 1, wherein, The control module includes a massage driving unit, the massage driving unit drives the massage head to perform a massage action according to a preset rule, and the specific process includes: Based on the divided muscle zones, a mapping relationship between the zones and the massage heads is established, and a target massage head group corresponding to each muscle zone is determined; According to the marked key care area and the care priority, a massage strategy database is queried to determine a target massage strategy; wherein the muscle zones of the key care area with high care priority are preferentially executed; According to the target massage strategy, a driving instruction is sent to the target massage head group to control the target massage head group to move in the corresponding muscle zone according to a preset trajectory, so as to realize zone massage coverage, and the moving speed of the massage head in the key care area is 0.6-0.8 times that in the non-key area.

6. The visual aid-based magnetic therapy back massager according to claim 1, wherein, The control module includes a massage data storage unit, which is used for recording massage parameter adjustment data and user skin feedback data in real time during the massage process, and establishing a massage effect database; When the use frequency of the same user reaches a preset frequency threshold, the control module generates a personalized massage parameter and magnetotherapy mode recommendation model of the user based on historical data in the massage effect database. In the subsequent massage process, the control module preferentially calls the personalized massage parameter and magnetotherapy mode recommendation model to determine the initial massage parameter and magnetotherapy mode, and then combines the real-time collected back dynamic image to perform parameter fine-tuning.

7. The visual aid based magnetic therapy back massager as claimed in claim 1, wherein, The control module comprises a communication unit configured to be signal-connected with a user terminal, the user terminal comprising a mobile computer, a tablet or a mobile phone terminal program, and the user terminal is configured to display the image of the back of the user in real time, and the current position of the massage head and the key care area are marked on the image of the back of the user by different colors during the massage process.

8. The visual aid based magnetic therapy back massager as claimed in claim 1, wherein, The method comprises the following steps: Based on the three sub-dimensions of the anatomical function, the daily stress characteristics and the historical strain probability of the back muscles of the user, the basic priority score of the key care area of the back muscles of the user is calculated, each sub-dimension is assigned a weight, and the weighted sum is obtained to obtain the basic priority score P1 of each key care area of the user; Based on the strain level, the real-time visual features and the real-time feedback of each key care area of the user, the basic priority score P1 of each key care area of the user is corrected to obtain the target priority score P2 of each key care area of the user; The target priority score of all key care areas of the user is calculated to obtain a plurality of target priority scores P2; The difference between any two target priority scores P2 is calculated to obtain a plurality of target differences; The plurality of target differences are compared with the preset difference threshold, and when it is determined that the target difference is less than or equal to the preset difference threshold, the two key care areas corresponding to the target difference are taken as the conflict key care areas; The conflict key care areas are subjected to conflict resolution, and the conflict resolution comprises at least one of time conflict, parameter conflict and position conflict; The target priority score P2 of each key care area is queried from the priority score-priority level table to determine the corresponding care priority of each key care area.

Images