A meditation scene generation method and device based on sub-health emotional changes

Abstract

The present invention discloses a meditation scene generation method and device based on sub-healthy emotion changes, which involves the intersection of EEG recognition and model training, including: obtaining EEG signal bands before, during and after use corresponding to six sub-healthy emotions; adjusting the weight of the trained model of each sub-healthy emotion in all trained models according to the proportion of the EEG signal band in use of each sub-healthy emotion in all EEG signal bands in use, and combining with XR technology to display in MR glasses; when the proportion of the EEG signal band of each sub-healthy emotion before and after use in all EEG signal bands is consistent, the EEG signal band after use is used as the seventh sub-healthy emotion and Stable diffusion is trained; after combining the seven trained models with MR technology, a postpartum meditation scene is displayed in MR glasses, and the model is iterated in a loop. The present invention can generate corresponding meditation scenes in real time according to the user in different use states.

Description

Technical Field

[0001] The present invention relates to the intersection of electroencephalogram recognition and model training, and in particular to a meditation scene generation method and device based on sub-healthy emotional changes. Background Art

[0002] EEG recognition technology is a communication system that does not rely on normal output pathways composed of peripheral nerves and muscles. It extracts EEG signals from the scalp and analyzes these signals through digital processing technology to "read" the human brain's intentions. It also provides a communication and control channel between users and external facilities. This is an emerging interaction mode between humans and computers.

[0003] In the prior art, meditation headbands are mostly used to monitor the user's meditation state and provide corresponding meditation indicators to help them meditate better and relieve the negative emotions of users with sub-healthy emotions. They can also assist users in meditation through technology such as music playback.

[0004] However, the above-mentioned meditation headband technology cannot generate corresponding meditation scenes in real time according to the different sub-health emotions generated by the user during and after use. Summary of the invention

[0005] The embodiments of the present invention provide a method and device for generating a meditation scene based on sub-healthy emotional changes, which can solve the problem in the prior art that corresponding meditation scenes cannot be generated in real time according to different usage states of the user.

[0006] The embodiment of the present invention provides a meditation scene generation method based on sub-healthy emotion changes, comprising the following steps: acquiring facial image samples of multiple sub-healthy emotions, and training a deep learning model through the facial image samples to obtain trained models that can generate emotion-related images corresponding to six sub-healthy emotions; and acquiring each trained model, loading and parsing it with MR technology and displaying it as a meditation scene in MR glasses; collecting EEG data of the MR glasses user before, during and after use through a meditation headband, and parsing the EEG data to obtain EEG signal bands of the MR glasses user before, during and after use corresponding to each sub-healthy emotion; wherein, parsing the EEG data to obtain EEG signal bands of the MR glasses user before and after use corresponding to each sub-healthy emotion specifically comprises: analyzing the brain-computer interface of the meditation headband The invention collects EEG data of MR glasses users separately; transmits the EEG data to the EEG recognition terminal through the built-in Bluetooth of the meditation headband, and parses the EEG data through the EEG recognition terminal to generate corresponding series, which are the EEG signal bands corresponding to various sub-healthy emotions; calculates the proportion of the EEG signal bands of MR glasses users with each sub-healthy emotion in all the EEG signal bands of sub-healthy emotions in use, and adjusts the weight of the trained model of each sub-healthy emotion in the weight parameters of the deep learning model according to the proportion; performs three-dimensional scene modeling on the adjusted trained model through XR technology to obtain an adjusted meditation scene that can be displayed in the MR glasses; wherein, the weight of adjusting the trained model of each sub-healthy emotion in all the trained models specifically includes: using the deep learning model Stable The multi-model fusion function and model weight adjustment function of diffusion select the trained model of each sub-health emotion, and adjust the weight according to the adaptation ratio of the trained model of each sub-health emotion; the obtained adjusted meditation scene that can be displayed in the MR glasses specifically includes: obtaining a two-dimensional image according to the trained model adjusted by the sub-health emotion input; identifying features in the two-dimensional image through automatic feature recognition and modeling in XR technology; connecting the artificial intelligence ChatGpt to the visualization language Grasshopper, performing three-dimensional scene modeling on the identified features, and rendering the image into a three-dimensional view through a deep learning model, that is, To adjust the meditation scene; compare the proportion of the EEG signal bands of each sub-healthy emotion before and after use to the EEG signal bands before, during and after use; when the proportions are the same, use the EEG signal band after use as the EEG signal band corresponding to the seventh sub-healthy emotion; train the deep learning model according to the facial image samples of the seventh sub-healthy emotion to obtain the seventh trained model; load and analyze the seven trained models using MR technology and display them as improved meditation scenes in MR glasses; collect EEG data of MR glasses users through a meditation headband, and analyze the EEG data to obtain the EEG signal bands before, during and after use corresponding to the seven sub-healthy emotions.

[0007] Furthermore, the comparison of the proportion of the EEG signal bands of each sub-healthy emotion before and after use to the EEG signal bands before, during and after use specifically includes: when the proportions match, it means that adjusting the postpartum meditation scene is beneficial to the improvement of the user's sub-healthy emotions; when the proportions do not match, it means that the sub-healthy emotion iteration under the brain-computer signal recognition accuracy in the current use stage has reached saturation, and has also reached the limit of EEG recognition in the current use stage.

[0008] Furthermore, the acquiring of facial image samples of a plurality of sub-healthy emotions specifically includes: screening out model training image samples that conform to the six sub-healthy emotion impact dimensions of worry, anxiety, irritability, tension, entanglement and boredom through big data screening and meditation environment design methods.

[0009] The embodiment of the present invention provides a meditation scene generation device based on sub-health emotion changes, including:

[0010] The MR glasses display module is used to obtain facial image samples of various sub-healthy emotions, and train the deep learning model through the facial image samples to obtain trained models that can generate emotion-related images corresponding to six sub-healthy emotions; and obtain each trained model and load it with MR technology for analysis and display it as a meditation scene in the MR glasses; the meditation headband EEG signal acquisition module is used to collect the EEG data of the MR glasses user before, during and after use through the meditation headband, and analyze the EEG data to obtain the EEG signal bands of the MR glasses user before, during and after use corresponding to each sub-healthy emotion; wherein, the analysis of the EEG data to obtain the EEG signal bands of the MR glasses user before, during and after use corresponding to each sub-healthy emotion specifically includes: collecting the EEG data of the MR glasses user through the brain-computer interface of the meditation headband The MR glasses meditation scene generation module calculates the proportion of the EEG signal bands of MR glasses users with each sub-healthy emotion in the use of all EEG signal bands of sub-healthy emotions, and adjusts the weight of the trained model of each sub-healthy emotion in the weight parameters of the deep learning model according to the proportion; the adjusted trained model is used for three-dimensional scene modeling through XR technology to obtain an adjusted meditation scene that can be displayed in the MR glasses; wherein, the weight of the trained model of each sub-healthy emotion in the trained model is adjusted by the deep learning model Stable The multi-model fusion function and model weight adjustment function of diffusion select the trained model of each sub-healthy emotion, and adjust the weight according to the adaptation ratio of the trained model of each sub-healthy emotion; the said obtaining the adjusted meditation scene that can be displayed in the MR glasses specifically includes: obtaining a two-dimensional image according to the trained model adjusted by the sub-healthy emotion input; identifying features in the two-dimensional image through automatic feature recognition and modeling in XR technology; connecting the artificial intelligence ChatGpt to the visualization language Grasshopper, performing three-dimensional scene modeling on the identified features, and rendering the image into a three-dimensional view through a deep learning model, that is, adjusting the meditation scene; the MR glasses meditation scene improvement module is used to compare the proportion of the EEG signal band of each sub-healthy emotion before and after use to the EEG signal band before, during and after use; when the proportion is the same, the EEG signal band after use is used as the EEG signal band corresponding to the seventh sub-healthy emotion; according to the facial image sample of the seventh sub-healthy emotion, the deep learning model is trained to obtain the seventh trained model; the seven trained models are loaded and analyzed by MR technology and displayed as improved meditation scenes in the MR glasses;The EEG signal feedback collection module is used to collect EEG data of MR glasses users through the meditation headband, and analyze the EEG data to obtain the EEG signal bands before, during and after use corresponding to the seven sub-healthy emotions. ;

[0011] The embodiment of the present invention provides a meditation scene generation method and device based on sub-health emotion changes. Compared with the prior art, the beneficial effects thereof are as follows:

[0012] The EEG data of MR glasses users before, during and after use are collected through a meditation headband, and the EEG data are analyzed to obtain the EEG signal bands of MR glasses users before, during and after use corresponding to each sub-healthy emotion; the proportion of the EEG signal bands of MR glasses users with each sub-healthy emotion in use to all EEG signal bands of sub-healthy emotions is calculated, and the weight of the trained model of each sub-healthy emotion in the weight parameters of the deep learning model is adjusted according to the proportion to all trained models; the adjusted trained model is used for three-dimensional scene modeling through XR technology to obtain the adjusted meditation scene that can be displayed in the MR glasses; respectively The proportion of the EEG signal bands of each sub-healthy emotion before and after use to the EEG signal bands before, during and after use is compared; when the proportions are the same, the EEG signal band after use is used as the EEG signal band corresponding to the seventh sub-healthy emotion; based on the facial image samples of the seventh sub-healthy emotion, the deep learning model is trained to obtain the seventh trained model; the seven trained models are loaded and analyzed using MR technology and displayed as improved meditation scenes in MR glasses; the EEG data of MR glasses users are collected through a meditation headband, and the EEG data are analyzed to obtain the EEG signal bands before, during and after use corresponding to the seven sub-healthy emotions.

[0013] Among them, according to the proportion of the EEG signal band in use of each sub-healthy emotion in all the EEG signal bands in use, the weight of the trained model of each sub-healthy emotion in all the trained models is adjusted; the adjusted trained model is modeled in three dimensions through XR technology to obtain an adjusted meditation scene that can be displayed in the MR glasses, so that the original meditation scene can be adaptively adjusted according to different sub-healthy emotions in the use of MR glasses; by analyzing the EEG data, the EEG signal bands of the MR glasses user corresponding to each sub-healthy emotion before, during and after use can be obtained, so as to capture the sub-healthy emotional state of the MR glasses user after use in real time; compare whether the proportion of the EEG signal band of each sub-healthy emotion before and after use is the same in all the EEG signal bands, and when they are the same, use the EEG signal band after use as the seventh sub-healthy emotion, and train Stable diffusion to obtain the seventh trained model, and then combine the seven trained models with MR technology to display them as improved meditation scenes in the MR glasses, so as to capture the sub-healthy emotional state of the user after use and generate corresponding meditation scenes in real time according to the sub-healthy emotional state after use, and finally achieve the effect of generating corresponding meditation scenes in real time under different use states. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 A technical schematic diagram of a meditation scene generation method and device based on sub-healthy emotional changes provided by an embodiment of the present invention;

[0015] Figure 2 A sub-healthy mood map of a meditation scene generation method and device based on sub-healthy mood changes provided by an embodiment of the present invention;

[0016] Figure 3 A product introduction diagram of a meditation scene generation method and device based on sub-healthy emotional changes provided by an embodiment of the present invention;

[0017] Figure 4 A design diagram of the electronic components of a headband of a meditation scene generation method and device based on sub-healthy emotional changes provided by an embodiment of the present invention;

[0018] Figure 5 A meditation scene diagram of a meditation scene generation method and device based on sub-healthy emotional changes provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0019] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without violating the connotation of the present invention, so the present invention is not limited by the specific embodiments disclosed below.

[0020] See also Figures 1 to 5 , an embodiment of the present invention provides a meditation scene generation method based on sub-health emotion changes, comprising the following steps:

[0021] Step 1: Obtain six trained models for the deep learning model Stable diffusion based on the six sub-healthy emotions collected from worry, anxiety, irritability, tension, entanglement and boredom. Specifically, the initial generative diffusion model training is performed based on the common sub-healthy emotional influencing factors, namely, worry, anxiety, irritability, tension, entanglement, and boredom. Through the method of big data screening and meditation environment design, the model training image samples for the six sub-healthy emotional influencing dimensions are screened out, and the initial model training is performed by docking the stable diffusion port.

[0022] Step 2: Obtain the postpartum meditation scene displayed in the MR glasses based on the six trained models combined with MR technology. Specifically, the six initial models are combined with MR technology to generate the corresponding postpartum meditation scene, which is applied to the user test scene and displayed in the MR glasses.

[0023] In addition, the brain-computer interface of the meditation headband collects the EEG data of the MR glasses user before, during and after use, and the EEG signal bands before, during and after use corresponding to the six sub-healthy emotions are obtained through the EEG recognition terminal. Specifically, the user's EEG data is collected through the brain-computer interface, and then transmitted to the EEG recognition terminal via Bluetooth to generate the corresponding series, and the EEG signal bands corresponding to the six initial sub-healthy emotional impact dimensions are obtained.

[0024] Step three: According to the proportion of each sub-healthy emotion in the use of EEG signal bands in the total use of EEG signal bands, adjust the weight of each sub-healthy emotion trained model in the total trained models; use the adjusted trained model to perform three-dimensional scene modeling through XR technology to obtain an adjusted postpartum meditation scene that can be displayed in MR glasses. Specifically, the user's EEG signal is secondary recognized through the brain-computer interface, that is, the user's real-time emotional changes are detected. According to the proportion of the six sub-healthy emotional impact dimensional bands after decoding the user's EEG signal, the weights of the six initial models of the stable diffusion generation port are adjusted, meditation scenes under changing emotions are generated, and corresponding postpartum meditation scenes are generated in combination with XR technology, and the adjusted scenes are presented in MR eyes.

[0025] Step 4: Compare the proportion of each sub-health emotion EEG signal band in all EEG signal bands before and after use; when the proportions match, use the EEG signal band after use as the seventh sub-health emotion; train the deep learning model Stable diffusion based on the sub-health emotion image samples collected for the seventh sub-health emotion to obtain the seventh trained model; combine the seven trained models with MR technology to display the postpartum meditation scene in MR glasses. Specifically, the user's EEG signal is identified three times through the brain-computer interface, that is, the user's EEG feedback signal is collected, and the proportion of each band signal before the user uses it is compared with the degree of change of the corresponding band signal after the user uses it. If the proportion of each signal before use is basically consistent with the proportion of each signal variable after use, it means that the adjustment is conducive to the improvement of the user's negative emotions. Therefore, the model training is carried out according to the meditation scene under this emotion, and this model is used to supplement the initial model to establish the seventh sub-health emotion impact dimension and corresponding model.

[0026] Step 5: Collect EEG data of the MR glasses user through the brain-computer interface of the meditation headband, and obtain the EEG signal bands before, during and after use corresponding to the seven sub-healthy emotions through the EEG recognition terminal. Specifically, the user's EEG signals are recognized four times according to the seven sub-healthy emotional impact dimensions, and the corresponding series of the seven dimensions are recorded. The dimensions are supplemented and the model is trained according to the above steps, and it is gradually iterated until the proportion of each signal before use and the proportion of each signal variable after use cannot match. It means that the emotional dimension has reached saturation under this brain-computer signal recognition accuracy, and the limit of the EEG recognition has also been reached. The emotion at this time should be the emotion that is closest to the user's current emotional type under the existing EEG recognition conditions. Therefore, the adaptive model system training is completed.

[0027] The present invention is a product based on a meditation headband with a brain-computer interface combined with MR glasses that can generate different scenes in real time. The main technical problems it solves are the technology of crossing EEG signal bands through EEG recognition technology and generative diffusion model technology, as well as the adaptive model iteration technology of "multi-dimensional EEG signal collection-multi-dimensional meditation scene generation-EEG signal feedback collection-new dimension model training and new dimension establishment" in the cross-field of EEG recognition-model training.

[0028] The main components of the present invention are:

[0029] It mainly includes a meditation headband and MR glasses. The EEG sensor in the meditation headband captures the user's emotions through brain waves and transmits the emotions to the MR glasses. In the glasses, the original basic architectural space model will be adjusted through the AI ​​algorithm to make it a model that is more conducive to the user's improvement of their negative emotions. Finally, through the use of this model, the user can be guided into a deeper state of mindfulness and treat psychological problems. The technology in the meditation headband and MR glasses crosses the EEG recognition technology with the generative diffusion model technology through the EEG signal band. Through the method of "multi-dimensional EEG signal collection-multi-dimensional meditation scene generation-EEG signal feedback collection-new dimension model training and new dimension establishment", the user's meditation scene can be adaptively changed with linear emotional changes, and the user's personalized meditation model can be autonomously iterated. This technology generates personalized meditation scenes by collecting and decoding the user's EEG signals, and autonomously iterates the model by comparing the user's EEG series before and after generation. Finally, the high-precision adaptive personalized real-time user meditation scene generation is achieved through the method of high-frequency EEG recognition-model training docking.

[0030] Among them, the following related technologies are used:

[0031] 1. The design idea of ​​combining the meditation headband with MR glasses.

[0032] 2. EEG recognition technology and generative diffusion model technology use the technology of crossing EEG signal bands. Generative diffusion model technology is a deep learning model.

[0033] 3. Adaptive model iteration technology of "multi-dimensional EEG signal collection - multi-dimensional meditation scene generation - EEG signal feedback collection - new dimension model training and new dimension establishment" in the cross-field of EEG recognition and model training. Specifically, multi-dimensional EEG signal collection in the present invention means collecting EEG data of MR glasses users before, during and after use, and analyzing the EEG data to obtain EEG signal bands of MR glasses users before, during and after use corresponding to each sub-health emotion.

[0034] The generation of multi-dimensional meditation scenes in the present invention represents adjusting the weight of the trained model of each sub-health emotion in the weight parameters of the deep learning model according to the proportion of all trained models; the adjusted trained model is subjected to three-dimensional scene modeling through XR technology to obtain an adjusted meditation scene that can be displayed in MR glasses.

[0035] New dimension model training and new dimension establishment in the present invention represent the comparison of the proportion of the EEG signal band of each sub-health emotion before and after use to the EEG signal band before, during and after use; when the proportion is the same, the EEG signal band after use is used as the EEG signal band corresponding to the seventh sub-health emotion. According to the facial image sample of the seventh sub-health emotion, the deep learning model is trained to obtain the seventh trained model; the seven trained models are loaded and analyzed using MR technology and displayed as improved meditation scenes in MR glasses.

[0036] In the present invention, EEG signal feedback collection means collecting EEG data of MR glasses users through a meditation headband, and analyzing the EEG data to obtain EEG signal bands before, during and after use corresponding to seven sub-health emotions.

[0037] The present invention also has the following beneficial effects:

[0038] Meditation headbands with brain-computer interfaces can help people better complete mindfulness meditation training, thereby effectively alleviating various negative emotions and playing a positive role in themselves. Through the adaptive model iteration technology of "multi-dimensional EEG signal collection-multi-dimensional meditation scene generation-EEG signal feedback collection-new dimension model training and new dimension establishment" under the cross-field of EEG recognition and model training, and the adaptive model iteration technology of "multi-dimensional EEG signal collection-multi-dimensional meditation scene generation-EEG signal feedback collection-new dimension model training and new dimension establishment" under the cross-field of EEG recognition and model training, the user's inner emotions can be captured very effectively. Through the algorithm, this emotion will be close to a new sub-health emotion that is independent of the six sub-health emotion impact dimensions of worry, anxiety, irritability, tension, entanglement, and boredom. This sub-health emotion will be closer to the user's real negative emotions, and through subsequent experimental feedback and continuous iterations, the eighth and ninth sub-health emotions will be generated. These emotions will occur according to the user's real-time situation to ensure the real-time nature of treatment. At the same time, these emotions will become closer and closer to the user's real emotions, thereby ensuring that the given virtual model is more conducive to the treatment of negative emotions, and ultimately ensuring that the headband can be both real-time and accurate.

[0039] This technology provides users with a more customized and immersive treatment process by combining a meditation headband and MR glasses. By generating the most suitable scene for the user, the user can be immersed in the scene and maximize the positive effects of meditation, thus ensuring the therapeutic effect of the headband + MR glasses product.

[0040] The embodiment of the present invention provides a meditation scene generation device based on sub-health emotion changes, including:

[0041] The MR glasses display module is used to obtain facial image samples of various sub-healthy emotions, and train the deep learning model through the facial image samples to obtain the trained models that can generate emotion-related images corresponding to the six sub-healthy emotions; and obtain each trained model and load and analyze it with MR technology to display it as a meditation scene in the MR glasses. The meditation headband EEG signal acquisition module is used to collect the EEG data of the MR glasses user before, during and after use through the meditation headband, and analyze the EEG data to obtain the EEG signal bands of the MR glasses user before, during and after use corresponding to each sub-healthy emotion; wherein, analyzing the EEG data to obtain the EEG signal bands of the MR glasses user before, during and after use corresponding to each sub-healthy emotion specifically includes: collecting the EEG data of the MR glasses user through the brain-computer interface of the meditation headband; transmitting the EEG data to the EEG recognition terminal through the built-in Bluetooth of the meditation headband, and analyzing the EEG data through the EEG recognition terminal to generate corresponding series, that is, the EEG signal bands corresponding to various sub-healthy emotions. In the MR glasses meditation scene generation module, the user calculates the proportion of the EEG signal band of each sub-healthy emotion in the use of MR glasses users to all the EEG signal bands of sub-healthy emotions, and adjusts the weight of each sub-healthy emotion trained model in the weight parameters of the deep learning model according to the proportion; the adjusted trained model is subjected to three-dimensional scene modeling through XR technology to obtain an adjusted meditation scene that can be displayed in the MR glasses; wherein, the weight of each sub-healthy emotion trained model in all trained models is adjusted, specifically including: selecting each sub-healthy emotion trained model through the multi-model fusion function and model weight adjustment function of the deep learning model Stable diffusion, and adjusting the weight according to the adaptation degree ratio of each sub-healthy emotion trained model; obtaining the adjusted meditation scene that can be displayed in the MR glasses, specifically including: obtaining a two-dimensional image according to the sub-healthy emotion input adjusted trained model; identifying features in the two-dimensional image through automatic feature recognition and modeling in XR technology; connecting the artificial intelligence ChatGpt with the visualization language Grasshopper, performing three-dimensional scene modeling on the identified features, and rendering the image into a three-dimensional view through the deep learning model, that is, the adjusted meditation scene. The MR glasses meditation scene improvement module is used to compare the proportion of the EEG signal bands of each sub-health emotion before and after use to the EEG signal bands before, during and after use; when the proportions are the same, the EEG signal band after use is used as the EEG signal band corresponding to the seventh sub-health emotion; based on the facial image samples of the seventh sub-health emotion, the deep learning model is trained to obtain the seventh trained model; the seven trained models are loaded and analyzed using MR technology and displayed as improved meditation scenes in the MR glasses.The EEG signal feedback collection module is used to collect the EEG data of MR glasses users through a meditation headband, and analyze the EEG data to obtain the EEG signal bands before, during and after use corresponding to seven sub-health emotions.

[0042] A specific embodiment is as follows:

[0043] In addition to the application scenarios of the present invention, the related technologies of the present invention can also be applied to the following situations:

[0044] 1. Creative generation of interior three-dimensional space design style based on MBTI prediction: A problem that architects often face is that it is difficult to understand or capture the real design intentions of the owner Party A, and the problem of repeatedly adjusting the plan is common. However, the emerging AIGC technology can solve these problems well. We hope to form a creative artificial intelligence model that can predict the sixteen personality types of the owner Party A based on the language text of the owner Party A, and then analyze his preference for the architectural interior design style. We conducted statistical analysis and data mining on the interior design style preferred by each personality type, and realized the prediction function of interior design style and interior design color preference for different personality types through machine learning model training. At the same time, based on the analysis results, we further trained a specific LORA model for each personality type, and by carrying these LORA models in StableDiffusion, we can achieve the task of generating architectural interior renderings in a targeted manner according to the needs of the owner Party A. This method can effectively assist architects in mining the preferences of owners, thereby improving the satisfaction of owners with the design scheme.

[0045] 2. Tsinghua University Graduation Project "Thinking without Thinking" - A popular science experience design of focused meditation based on brain-computer interaction. Focused meditation is a psychological activity that focuses attention on a single object or activity and abandons inner distractions. It has the effects of relieving anxiety and improving cognitive ability. This design is based on brain-computer interaction technology and uses 6 thematic immersive natural scenes as content carriers to achieve a popular science experience of focused meditation. In the experience scene, the audience wears an EEG collector and stares at the screen. The picture is affected by the activity of brain waves and will transition or stay between the preset normal state and the meditation state. Through the real-time brain-computer interaction picture feedback, the experiencer tries to control the activity of his own brain, so as to feel and practice meditation.

[0046] The above-mentioned embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A meditation scene generation method based on sub-healthy emotional changes, characterized in that: The following steps are involved: Obtain facial image samples of various sub-healthy emotions, and train the deep learning model through the facial image samples to obtain trained models that can generate emotion-related images corresponding to the six sub-healthy emotions; and obtain each trained model and load and analyze it using MR technology to display it as a meditation scene in MR glasses; The EEG data of the MR glasses users before, during and after use are collected through the meditation headband, and the EEG data are analyzed to obtain the EEG signal bands of the MR glasses users before, during and after use corresponding to each sub-health emotion; The step of analyzing the EEG data to obtain the EEG signal bands of the MR glasses user before, during and after use corresponding to each sub-health emotion specifically includes: collecting the EEG data of the MR glasses user respectively through the brain-computer interface of the meditation headband; transmitting the EEG data to the EEG recognition terminal through the built-in Bluetooth of the meditation headband, and analyzing the EEG data through the EEG recognition terminal to generate corresponding series, namely, the EEG signal bands corresponding to various sub-health emotions; The proportion of the EEG signal bands used by MR glasses users with each sub-health emotion in all EEG signal bands used with sub-health emotions is calculated, and the weight of the trained model of each sub-health emotion in the weight parameters of the deep learning model is adjusted according to the proportion; the adjusted trained model is used for three-dimensional scene modeling through XR technology to obtain an adjusted meditation scene that can be displayed in the MR glasses; The adjusting the weight of the trained model of each sub-healthy emotion in all trained models specifically includes: selecting the trained model of each sub-healthy emotion through the multi-model fusion function and model weight adjustment function of the deep learning model Stable diffusion, and adjusting the weight according to the adaptation degree ratio of the trained model of each sub-healthy emotion; The obtaining of the adjusted meditation scene that can be displayed in the MR glasses specifically includes: obtaining a two-dimensional image according to the trained model adjusted by the sub-healthy emotional input; identifying features in the two-dimensional image through automatic feature recognition and modeling in the XR technology; connecting the artificial intelligence ChatGpt with the visualization language Grasshopper, performing three-dimensional scene modeling on the identified features, and rendering the image into a three-dimensional view through a deep learning model, that is, adjusting the meditation scene; The proportion of the EEG signal band of each sub-health emotion before and after use to the EEG signal band before, during and after use is compared respectively; when the proportion is the same, the EEG signal band after use is used as the EEG signal band corresponding to the seventh sub-health emotion; according to the facial image samples of the seventh sub-health emotion, the deep learning model is trained to obtain the seventh trained model; the seven trained models are loaded and analyzed using MR technology and displayed as improved meditation scenes in MR glasses; The EEG data of MR glasses users are collected through a meditation headband, and the EEG data are analyzed to obtain the EEG signal bands before, during and after use corresponding to seven sub-health emotions.

2. A meditation scene generation method based on sub-healthy emotional changes as claimed in claim 1, characterized in that: The comparison of the proportion of the EEG signal band of each sub-health emotion before and after use to the EEG signal band before, during and after use specifically includes: When the proportions match, it means that adjusting the postpartum meditation scene is beneficial to improving the user's sub-healthy mood; When the proportions do not match, it means that the iteration of sub-healthy emotions under the brain-computer signal recognition accuracy in the current usage stage has reached saturation and has also reached the limit of EEG recognition in the current usage stage.

3. A meditation scene generation method based on sub-healthy emotional changes as claimed in claim 1, characterized in that: The acquiring of facial image samples of various sub-healthy emotions specifically includes: Through big data screening and meditation environment design methods, model training image samples that meet the six sub-health emotion impact dimensions of worry, anxiety, irritability, tension, entanglement and boredom were screened out.

4. A meditation scene generation device based on sub-healthy emotional changes, characterized in that: include: The MR glasses display module is used to obtain facial image samples of various sub-healthy emotions, and train the deep learning model through the facial image samples to obtain a trained model that can generate emotion-related images corresponding to six sub-healthy emotions; And obtain each trained model and load and analyze it using MR technology to display it as a meditation scene in MR glasses; The meditation headband EEG signal acquisition module is used to respectively collect EEG data of the MR glasses user before, during and after use through the meditation headband, and analyze the EEG data to obtain EEG signal bands of the MR glasses user before, during and after use corresponding to each sub-healthy emotion; wherein, the analysis of the EEG data to obtain the EEG signal bands of the MR glasses user before, during and after use corresponding to each sub-healthy emotion specifically includes: respectively collecting the EEG data of the MR glasses user through the brain-computer interface of the meditation headband; transmitting the EEG data to the EEG recognition terminal through the built-in Bluetooth of the meditation headband, and analyzing the EEG data through the EEG recognition terminal to generate corresponding series, namely, the EEG signal bands corresponding to various sub-healthy emotions; In the MR glasses meditation scene generation module, the user calculates the proportion of the EEG signal band of the MR glasses users of each sub-healthy emotion in the use of all the EEG signal bands of the sub-healthy emotions, and adjusts the weight of the trained model of each sub-healthy emotion in the weight parameters of the deep learning model according to the proportion; the adjusted trained model is subjected to three-dimensional scene modeling through XR technology to obtain an adjusted meditation scene that can be displayed in the MR glasses; wherein, the weight of the trained model of each sub-healthy emotion in all the trained models is adjusted, specifically including: selecting the trained model of each sub-healthy emotion through the multi-model fusion function and model weight adjustment function of the deep learning model Stable diffusion, and adjusting the weight according to the adaptation degree ratio of the trained model of each sub-healthy emotion; the obtained adjusted meditation scene that can be displayed in the MR glasses specifically includes: obtaining a two-dimensional image according to the trained model adjusted by the sub-healthy emotion input; identifying features in the two-dimensional image through automatic feature recognition and modeling in XR technology; connecting the artificial intelligence ChatGpt with the visualization language Grasshopper, performing three-dimensional scene modeling on the identified features, and rendering the image into a three-dimensional view through the deep learning model, that is, adjusting the meditation scene; The MR glasses meditation scene improvement module is used to compare the proportion of the EEG signal band of each sub-health emotion before and after use to the EEG signal band before, during and after use; when the proportion is the same, the EEG signal band after use is used as the EEG signal band corresponding to the seventh sub-health emotion; according to the facial image samples of the seventh sub-health emotion, the deep learning model is trained to obtain the seventh trained model; the seven trained models are loaded and analyzed using MR technology and displayed as improved meditation scenes in the MR glasses; The EEG signal feedback collection module is used to collect the EEG data of MR glasses users through a meditation headband, and analyze the EEG data to obtain the EEG signal bands before, during and after use corresponding to seven sub-health emotions.

Images