Electroencephalogram signal processing method and device, equipment, storage medium and program product

Abstract

The invention discloses an electroencephalogram signal processing method, device and equipment, a storage medium and a program product, the method is applied to electroencephalogram signal processing equipment, the electroencephalogram signal processing equipment is connected with electroencephalogram signal collecting equipment, and the method comprises the steps that N stimulation objects are output in the electroencephalogram signal processing equipment; n is a positive integer; electroencephalogram signals generated by the target object in the output process of the N stimulation objects and acquired by the electroencephalogram signal acquisition equipment are received; executing an operation corresponding to the target stimulation object; wherein the target stimulation object is determined from N stimulation objects according to the electroencephalogram signal of the target object; the method can be applied to the traffic field, and the electroencephalogram signals can be conveniently recognized to execute corresponding operation.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to methods, apparatus, devices, storage media and program products for processing electroencephalogram (EEG) signals. Background Technology

[0002] Biosignal recognition technology has been widely applied in many industries, such as the medical field, security field, and computer interaction applications like gaming. Commonly used biosignal recognition methods include electroencephalography (EEG), electromyography (EMG), and eye movement signal recognition. For EEG recognition, a current trend is to combine relevant EEG data with neural network models to differentiate EEG signals and execute different control operations. However, this approach is complex and requires designing appropriate neural network models, which is time-consuming and labor-intensive. Summary of the Invention

[0003] This application provides a method, apparatus, device, storage medium, and program product for processing electroencephalogram (EEG) signals, which can conveniently identify EEG signals and perform corresponding operations.

[0004] On one hand, embodiments of this application provide a method for processing electroencephalogram (EEG) signals. The method is applied to an EEG signal processing device connected to an EEG signal acquisition device. The method includes:

[0005] The EEG signal processing device outputs N stimulation targets; N is a positive integer.

[0006] Receive the electroencephalogram (EEG) signals generated by the target object during the output process of the N stimulation objects, acquired by the EEG signal acquisition device;

[0007] Perform an operation corresponding to a target stimulus; wherein the target stimulus is determined from the N stimulus objects based on the EEG signal of the target object.

[0008] On the one hand, embodiments of this application provide another method for processing electroencephalogram (EEG) signals, the method comprising:

[0009] Output N candidate intent information and N stimulus objects corresponding to each of the N candidate intent information; wherein, the N stimulus objects are all different and N is a positive integer;

[0010] Acquire the electroencephalogram (EEG) signals generated by the target object during the output of the N stimuli;

[0011] The EEG signals of the target object are subjected to feature analysis, and the target stimulus object that triggers the generation of the EEG signals of the target object is determined from the N stimulus objects based on the feature analysis results;

[0012] An intent recognition result is generated based on the candidate intent information corresponding to the target stimulus; wherein, the intent recognition result is used to trigger the execution of an operation that matches the candidate intent information corresponding to the target stimulus.

[0013] On one hand, embodiments of this application provide an electroencephalogram (EEG) signal processing device, which is deployed within an EEG signal processing equipment connected to an EEG signal acquisition device. The device includes:

[0014] A communication unit, used for communication interaction;

[0015] The processing unit is configured to output N stimulation objects in the EEG signal processing device; N is a positive integer; receive the EEG signals generated by the target object during the output of the N stimulation objects by the EEG signal acquisition device; and perform an operation corresponding to the target stimulation object; wherein the target stimulation object is determined from the N stimulation objects based on the target object's EEG signals.

[0016] On one hand, embodiments of this application provide another EEG signal processing device, the device comprising:

[0017] A communication unit, used for communication interaction;

[0018] The processing unit is configured to output N candidate intent information and N stimulus objects corresponding to the N candidate intent information; wherein the N stimulus objects are all different and N is a positive integer; acquire the electroencephalogram (EEG) signal generated by the target object during the output of the N stimulus objects; perform feature analysis on the EEG signal of the target object, and determine the target stimulus object that triggers the generation of the target object's EEG signal from the N stimulus objects based on the feature analysis results; generate an intent recognition result based on the candidate intent information corresponding to the target stimulus object; wherein the intent recognition result is used to trigger the execution of an operation that matches the candidate intent information corresponding to the target stimulus object.

[0019] On one hand, embodiments of this application provide a computer device, which includes an input interface and an output interface, and further includes:

[0020] Processor and computer-readable storage medium;

[0021] A computer-readable storage medium for storing computer programs;

[0022] The processor is used to run computer programs to implement the above-mentioned EEG signal processing methods.

[0023] On one hand, embodiments of this application provide a computer-readable storage medium storing a computer program adapted to be loaded by a processor and executed by the above-described electroencephalogram (EEG) signal processing method.

[0024] On one hand, embodiments of this application provide a computer program product, which includes a computer program adapted to be loaded by a processor and executed by the above-described EEG signal processing method.

[0025] In this embodiment, N stimulation objects can be output in the EEG signal processing device. After receiving the EEG signals generated by the target object during the output of the N stimulation objects from the EEG signal acquisition device, an operation corresponding to the target stimulation object is executed. The target stimulation object is determined from the N stimulation objects based on its EEG signal. By identifying the target stimulation object that induces the EEG signal of the target object from the N stimulation objects, and then executing the operation corresponding to the target stimulation object, the identification of EEG signals and the execution of corresponding operations can be conveniently performed. Attached Figure Description

[0026] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0027] Figure 1a This is a schematic diagram illustrating an intent to identify an object, as provided in an embodiment of this application.

[0028] Figure 1b This is a schematic diagram of the structure of an intent recognition system provided in an embodiment of this application;

[0029] Figure 1c This is a schematic diagram of another intent recognition system provided in an embodiment of this application;

[0030] Figure 2a This is a schematic diagram of the frequency response curve of a steady-state visual evoked potential provided in an embodiment of this application;

[0031] Figure 2b This is a power spectral density map of a steady-state visual evoked potential provided in an embodiment of this application;

[0032] Figure 2c This is a schematic diagram of a complex stimulus provided in an embodiment of this application;

[0033] Figure 2d This is a schematic diagram of different stimulus objects provided in the embodiments of this application;

[0034] Figure 3a This is a flowchart illustrating an electroencephalogram (EEG) signal processing method provided in an embodiment of this application.

[0035] Figure 3b This is a flowchart illustrating another EEG signal processing method provided in an embodiment of this application;

[0036] Figure 4a This is a schematic diagram illustrating the output of candidate intent information and stimulus objects provided in an embodiment of this application;

[0037] Figure 4b This is a schematic diagram of another output candidate intent information and stimulus object provided in an embodiment of this application;

[0038] Figure 4c This is a schematic diagram of the electrode distribution of an electroencephalogram (EEG) signal acquisition device provided in an embodiment of this application;

[0039] Figure 4d This is a schematic diagram of the electrode distribution of another electroencephalogram (EEG) signal acquisition device provided in an embodiment of this application;

[0040] Figure 4e This is a schematic diagram illustrating another intent to identify an object provided in an embodiment of this application;

[0041] Figure 5a This is a schematic diagram of an intent recognition process provided in an embodiment of this application;

[0042] Figure 5b This is a schematic diagram of another intent recognition process provided in an embodiment of this application;

[0043] Figure 5c This is a schematic diagram of an interface interaction provided in an embodiment of this application;

[0044] Figure 6 This is a flowchart illustrating another electroencephalogram (EEG) signal processing method provided in an embodiment of this application.

[0045] Figure 7a This is a schematic diagram of the power spectral density corresponding to a candidate stimulus provided in an embodiment of this application;

[0046] Figure 7b This is a schematic diagram of the power spectral density corresponding to a candidate stimulus provided in an embodiment of this application;

[0047] Figure 7c This is a schematic diagram of the power spectral density corresponding to a candidate stimulus provided in an embodiment of this application;

[0048] Figure 8 This is a schematic diagram of the structure of an electroencephalogram (EEG) signal processing device provided in an embodiment of this application;

[0049] Figure 9 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0050] The technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings.

[0051] This application provides an EEG signal processing device (also known as an intent recognition device) that can be used to execute the EEG signal processing scheme proposed in this application to achieve intent recognition. The EEG signal processing device can be a terminal device or a server. The terminal device mentioned in this application can be a smartphone, tablet computer, laptop computer, desktop computer, intelligent voice interaction device, smart home appliance, smart vehicle, and smart wearable device (such as smart glasses, smart helmet), etc., and this application does not limit this. The server mentioned in this application can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN (Content Delivery Network), and big data and artificial intelligence platforms, etc., and this application does not limit this.

[0052] In one possible implementation, the EEG signal processing device can be connected to the EEG signal acquisition device to acquire EEG signals and then perform intention recognition based on the acquired EEG signals. Optionally, the EEG signal processing device can be connected to the EEG signal acquisition device via wired and / or wireless means for data transmission, which is not limited in this application embodiment. The EEG signal acquisition device includes equipment for acquiring EEG signals, such as an EEG helmet or EEG headband. Electroencephalogram (EEG) is a type of bioelectrical signal, which is a regular electrical phenomenon closely related to life activities, generated by cells or tissues of living organisms in active or resting states. Its essence is the electrical signal generated by the transmembrane flow of sodium, potassium, and other ions.

[0053] This application provides an EEG signal processing scheme that can be applied to an EEG signal processing device connected to an EEG signal acquisition device. The scheme mainly includes: outputting N stimulus objects in the EEG signal processing device (N being a positive integer); receiving the EEG signals generated by the target object during the output of the N stimulus objects, acquired by the EEG signal acquisition device; and executing an operation corresponding to the target stimulus object, wherein the target stimulus object is determined from the N stimulus objects based on the target object's EEG signal. In one possible implementation, different stimulus objects may correspond to different candidate intent information; therefore, the EEG signal processing device executing the operation corresponding to the target stimulus object may include: executing an operation that matches the candidate intent information corresponding to the target stimulus object. In one possible implementation, before executing the operation that matches the candidate intent information corresponding to the target stimulus object, the EEG signal processing device may further: perform feature analysis on the target object's EEG signal and determine the target stimulus object that triggers the generation of the target object's EEG signal from the N stimulus objects based on the feature analysis results; generate an intent recognition result based on the candidate intent information corresponding to the target stimulus object; wherein the intent recognition result is used to trigger the execution of the operation that matches the candidate intent information corresponding to the target stimulus object. In other words, after generating the intent recognition result, the EEG signal processing device can respond to the intent recognition result and perform an operation that matches the candidate intent information corresponding to the target stimulus.

[0054] In one possible implementation, the EEG signal processing device can also output candidate intent information corresponding to the corresponding stimulus object during the output of the stimulus object. Based on this, this application embodiment provides another EEG signal processing scheme, which can output N candidate intent information and N stimulus objects corresponding to the N candidate intent information, and acquire the EEG signal generated by the target object during the output of the N stimulus objects; it can perform feature analysis on the EEG signal of the target object, and determine the target stimulus object that triggers the generation of the target object's EEG signal from the N stimulus objects based on the feature analysis results; it can generate an intent recognition result based on the candidate intent information corresponding to the target stimulus object; wherein, the N stimulus objects are all different, N is a positive integer, and the intent recognition result is used to trigger the execution of an operation that matches the candidate intent information corresponding to the target stimulus object. Further optionally, the EEG signal processing device can execute an operation that matches the candidate intent information corresponding to the target stimulus object in response to the intent recognition result.

[0055] In the aforementioned EEG signal processing scheme, the target object refers to any object (i.e., the user) that needs to be identified in terms of intent, and the stimulus object refers to information that can be used to trigger the generation of EEG signals related to that stimulus object. This application proposes that the EEG signal processing device can output N candidate intent information and N stimulus objects corresponding to each of the N candidate intent information. If the intent indicated by the output N candidate intent information contains an intent that the target object wishes to express, the target object can generate EEG signals related to the perceived stimulus object by perceiving the stimulus object corresponding to the candidate intent information. Based on this, the EEG signal processing device can acquire the EEG signals generated by the target object during the output of the N stimulus objects, and based on the feature analysis of the (acquired) target object's EEG signals, determine the target stimulus object that triggers the generation of the target object's EEG signals from the N stimulus objects, and generate an intent recognition result based on the candidate intent information corresponding to the target stimulus object; thus, accurate intent recognition can be achieved. Further optionally, the EEG signal processing device can, in response to the intent recognition result, perform an operation that matches the candidate intent information corresponding to the target stimulus object.

[0056] Candidate intent information refers to information that can be used to represent intent. Candidate intent information may include one or more of the following: text information, image information, video information, audio information, etc., and can be set according to specific needs; this application embodiment does not impose any limitations. Stimulus objects may include one or more of the following: visual stimuli, auditory stimuli, tactile stimuli, etc., and can be set according to specific needs; this application embodiment does not impose any limitations. When the stimulus objects include visual stimuli (i.e., the stimulus objects are visual stimuli), the N stimulus objects output by the EEG signal processing device can be, for example, different images (e.g., images of different colors, images of different shapes, images flashing at different frequencies, images moving with different trajectories, etc.); when the stimulus objects include auditory stimuli... In the case of an auditory stimulus, the N stimulus objects output by the EEG signal processing device can be different audio frequencies (e.g., audio frequencies of different pitches, loudnesses, timbres, etc.); when the stimulus objects include tactile stimuli (i.e., the stimulus objects are tactile), the N stimulus objects output by the EEG signal processing device can be different vibrations (e.g., vibrations of different frequencies, intensities, durations, etc.); when the stimulus objects include both visual and auditory stimuli, the N stimulus objects output by the EEG signal processing device can be different videos; for ease of explanation, the following embodiments of this application use visual stimuli and image stimulation as examples.

[0057] In an exemplary application scenario, N=4, and the first to fourth candidate intent information are respectively: the text "eat" representing the intent to "eat", the text "drink" representing the intent to "drink", the text "sleep" representing the intent to "sleep", and the text "go to the toilet" representing the intent to "go to the toilet". The stimulus objects corresponding to the first to fourth candidate intent information are different images; please refer to [link to relevant documentation]. Figure 1a This is a schematic diagram illustrating an embodiment of the present application for recognizing an object's intent. The EEG signal processing device can output the aforementioned four candidate intent information and four stimulus objects. One of the candidate intent information can be the text "eating," as indicated by label 101, and the corresponding stimulus object can be as indicated by label 102. If the EEG signal processing device acquires the EEG signals generated by the target object during the output of these four stimulus objects, and determines the target stimulus object as indicated by label 102 based on the EEG signals, it can generate an intent recognition result to trigger the execution of an operation matching the candidate intent information (i.e., the text "eating") corresponding to the target stimulus object. Further optionally, the EEG signal processing device can, in response to the intent recognition result, execute an operation matching the candidate intent information corresponding to the target stimulus object, for example, outputting the candidate intent information corresponding to the target stimulus object, i.e., outputting the text "eating."

[0058] In one possible implementation, the above-described EEG signal processing scheme can be executed independently by an EEG signal processing device. Based on this, please refer to [link to relevant documentation]. Figure 1b This is a schematic diagram of an intent recognition system provided in an embodiment of this application. The intent recognition system may include an electroencephalogram (EEG) signal acquisition device 111 and an EEG signal processing device 112; the aforementioned EEG signal processing scheme is provided by... Figure 1b When the intent recognition system shown is executed, the EEG signal processing device 112 can output N candidate intent information and N stimulus objects corresponding to the N candidate intent information; the EEG signal acquisition device 111 can acquire the EEG signals generated by the target object during the output of the N stimulus objects and transmit them to the EEG signal processing device 112; the EEG signal processing device 112 can acquire the EEG signals generated by the target object during the output of the N stimulus objects and execute the relevant process of generating intent recognition results based on the EEG signals; further optionally, the EEG signal processing device 112 can perform related operations in response to the intent recognition results. The EEG signal acquisition device 111 and the EEG signal processing device 112 can achieve data interaction based on wired and / or wireless communication methods, such as data transmission via Bluetooth, which is not limited in this embodiment.

[0059] In another possible implementation, the above-described EEG signal processing scheme can also be executed collaboratively by multiple computing devices. For example, when the EEG signal processing device is a terminal device, it can be executed collaboratively by the terminal device and a server. Based on this, please refer to [link to relevant documentation]. Figure 1c This is a schematic diagram of another intent recognition system provided in an embodiment of this application. The intent recognition system may include an electroencephalogram (EEG) signal acquisition device 121, an EEG signal processing device 122, and a server 123. For example, the above-mentioned EEG signal processing scheme is... Figure 1c When the intent recognition system shown is executed, the EEG signal processing device 122 can output N candidate intent information and N stimulus objects corresponding to the N candidate intent information; the EEG signal acquisition device 121 can acquire the EEG signals generated by the target object during the output of the N stimulus objects and transmit them to the EEG signal processing device 122; the EEG signal processing device 122 can send the acquired EEG signals to the server 123; the server 123 can execute the relevant process of generating intent recognition results based on the EEG signals and return the generated intent recognition results to the EEG signal processing device 122; further optionally, the EEG signal processing device 122 can perform relevant operations in response to the intent recognition results. For example, the above-mentioned EEG signal processing scheme is... Figure 1c When the intent recognition system shown is executed, the EEG signal processing device 122 can output N candidate intent information and N stimulation objects corresponding to the N candidate intent information; the EEG signal acquisition device 121 can acquire the EEG signals generated by the target object during the output of the N stimulation objects and transmit them to the EEG signal processing device 122; the EEG signal processing device 122 can send the acquired EEG signals to the server 123; the server 123 can execute the relevant process of determining the target stimulation object based on the EEG signals and return feedback information for instructing the target stimulation object to the EEG signal processing device 122; the EEG signal processing device 122 can generate an intent recognition result according to the candidate intent information corresponding to the target stimulation object; further optionally, the EEG signal processing device 122 can perform related operations in response to the intent recognition result. The devices with communication needs can achieve data interaction based on wired and / or wireless communication methods, such as data transmission via Bluetooth, which is not limited in this embodiment. For ease of explanation, the following embodiments of this application take the EEG signal processing scheme being executed by the EEG signal processing device alone as an example.

[0060] In one possible implementation, a visual evoked potential (VEP) is an electroencephalogram (EEG) signal generated in the cerebral cortex after the human eye is stimulated by visual stimuli such as images or flashes of light. In other words, when the stimulus output by the EEG signal processing device is a visual stimulus, it can be used to trigger the generation of a visual evoked potential (i.e., it can be used to induce a visual evoked potential). A visual evoked potential is a type of event-related potential (ERP), where an ERP is an EEG signal associated with a specific event or stimulus, reflecting the brain's physiological response and cognitive processes to external stimuli. The stimulus in this embodiment can induce the generation of an ERP signal. Visual evoked potentials can be divided into transient visual evoked potentials and steady-state visual evoked potentials. A transient visual evoked potential is triggered by a single visual stimulus. Its signal reaches a peak after induction and then quickly returns to a resting state during oscillation. One transient visual evoked potential corresponds to one visual stimulus. Steady-state visual evoked potentials (SSVEPs) are generated by a visual stimulus that is repeated at a fixed frequency (called the stimulus frequency, usually greater than 4 Hz). The repeated stimulus causes the VEP signals to overlap, so SSVEPs have strong rhythmicity, which is reflected in the fact that the power changes with frequency can be more easily shown through power spectral density (SPD) analysis.

[0061] The SSVEP signal has the following characteristics:

[0062] (1) SSVEP response frequency.

[0063] Based on the stimulation frequency, SSVEP can generally be divided into three response bands: low frequency (SSVEP response evoked by stimulation frequencies less than 15 Hz), mid frequency (SSVEP response evoked by stimulation frequencies between 15 Hz and 30 Hz), and high frequency (SSVEP response evoked by stimulation frequencies greater than 30 Hz). Please refer to [link to relevant documentation]. Figure 2a This is a schematic diagram of the frequency response curve of a steady-state visual evoked potential provided in an embodiment of this application; by Figure 2aIt can be seen that the low-frequency band SSVEP response is the strongest, followed by the mid-frequency band, and the high-frequency band response is the weakest. Based on this, when it is necessary to set the stimulation frequency to induce SSVEP, the low-frequency band, mid-frequency band, etc. can be selected first. Since the stimulation frequency around 10Hz and around 20Hz induces a higher amplitude of SSVEP, the stimulation frequency around 10Hz in the low-frequency band and around 20Hz in the mid-frequency band can be selected first.

[0064] (2) Harmonic response.

[0065] The frequency harmonic response of SSVEP means that SSVEP not only generates an electrical signal at the stimulation frequency, but also responds at harmonics of the stimulation frequency. Please see [link to relevant documentation]. Figure 2b This application provides a power spectral density map of steady-state visual evoked potentials (SSVEPs). It obtains a 10-second SSVEP signal evoked by a visual stimulus (i.e., the visual stimulus object) with a stimulation frequency of 10 Hz, processes it to obtain the power spectral density, and presents it in a graph for easy observation. The horizontal axis represents frequency, and the vertical axis represents the power of the SSVEP signal at the corresponding frequency. Based on... Figure 2b It can be seen that obvious peaks appeared at the stimulation frequency (10Hz) and at the harmonics of the stimulation frequency (20Hz and 30Hz). This is the harmonic response characteristic of SSVEP, that is, SSVEP not only generates an electrical signal at the stimulation frequency, but also responds at the harmonics of the stimulation frequency.

[0066] Based on the octave response of SSVEP, when multiple stimulus frequencies need to be set to induce SSVEP, different stimulus frequencies should avoid being octaves or subdivisions of each other. For example, stimuli with frequencies of 10Hz and 20Hz should not appear in the same stimulus paradigm at the same time. Doing so can avoid the influence of the octave response characteristics of SSVEP, thereby improving the recognition accuracy when using SSVEP for intent recognition.

[0067] (3) The effect of stimulus form on SSVEP.

[0068] In principle, SSVEP will be induced as long as the target of stimulation provides visual stimulation to a certain object (i.e., a certain user) at an appropriate frequency. However, different forms of stimulation may affect the SSVEP signal induced. Generally speaking, stimuli can be divided into simple stimuli and complex stimuli.

[0069] Simple stimuli can be visual stimuli that display a single color at a specific frequency, such as an LED bulb displaying a certain color at a certain frequency, or a display screen (e.g., a CRT display or an LCD display) displaying an image of a certain color at a certain frequency, and so on. Complex stimuli can be visual stimuli represented by checkerboard patterns, grids, etc.; please refer to [link to relevant documentation]. Figure 2c This is a schematic diagram of a complex stimulus provided in an embodiment of this application, including a grid stimulus and a checkerboard stimulus. The grid image used for the grid stimulus can change continuously during the output process, for example, it can change back and forth between the grid image shown as 201 and the grid image shown as 202. The checkerboard image used for the checkerboard stimulus can change continuously during the output process, for example, it can change back and forth between the checkerboard image shown as 203 and the checkerboard image shown as 204.

[0070] Studies have shown that complex stimuli elicit stronger SSVEP signals and are more easily identifiable than simple stimuli. However, complex stimuli also have some drawbacks. On the one hand, while complex stimuli can induce stronger SSVEP, this characteristic is limited to low-to-mid-frequency visual stimuli. In other words, in high-frequency stimuli, the SSVEP induced by complex stimuli may actually be weaker than that induced by simple stimuli. On the other hand, due to their complex image design, resembling a checkerboard or grid, complex stimuli require more space to display than simple stimuli. Based on the above description, the stimulation method can be selected according to specific needs. For example, simple stimuli can display a single-color image at a specific frequency, while complex stimuli can use a checkerboard pattern. In some alternative implementations, a rotated, centrally symmetrical graphic can also be used as the stimulus source.

[0071] As described above, the N stimulus objects output by the EEG signal processing device can be set according to specific needs. For example, when the stimulus objects are visual, a visual stimulation method that displays images at a specific frequency can be selected. Therefore, the N stimulus objects output by the EEG signal processing device include visual stimulation images displayed at a specific frequency. Different stimulus objects have different stimulation frequencies or image characteristics (i.e., different stimulus objects have different stimulation frequencies or visual stimulation images). The stimulation frequency and visual stimulation images of the N stimulus objects can both be set according to specific needs. For an example, please refer to [link to example]. Figure 2dThis is a schematic diagram of different stimulation objects provided in an embodiment of this application; wherein, the EEG signal processing device outputs N stimulation objects, N=4, the first stimulation object includes a first visual stimulation image displayed at a first stimulation frequency, the second stimulation object includes a second visual stimulation image displayed at a second stimulation frequency, the third stimulation object includes a third visual stimulation image displayed at a third stimulation frequency, and the fourth stimulation object includes a fourth visual stimulation image displayed at a fourth stimulation frequency, wherein each stimulation frequency is different, and each visual stimulation image is different. Specifically, each visual stimulation image has the same shape and size, but different colors. Regarding the selection of stimulus colors, related studies have shown that red visual stimuli typically exhibit the highest classification accuracy and information transfer rate at different stimulus frequencies; white visual stimuli generally perform well in EEG systems based on steady-state visual evoked signals; 10Hz blue visual stimuli usually elicit the most significant SSVEP signal response; and purple visual stimuli typically achieve relatively high classification accuracy and information transfer rate in their SSVEP signals. Based on these findings, in one possible application scenario, the colors of the four visual stimulus images can be selected from red, white, blue, and purple. It is known that... Figure 2d The layout of N stimulus objects is only an exemplary description. The layout of N stimulus objects can also be designed according to specific needs. This application embodiment does not impose any limitations. For example, the distance between different stimulus objects can be set to be as large as possible in order to better stimulate the human visual system.

[0072] It should be particularly noted that the collection and processing of relevant data (such as electroencephalogram (EEG) signals) in this application should strictly comply with the requirements of laws and regulations, obtain the informed consent or separate consent of the personal information subject, and conduct subsequent data use and processing within the scope authorized by laws and regulations and the personal information subject. When the relevant embodiments of this application are applied to specific products or technologies, the relevant data collection, use, and processing processes should comply with the requirements of laws and regulations. Before collecting biological information (including EEG signals), the information processing rules should be communicated and the separate consent of the subject should be obtained. Biological information should be processed in strict accordance with the requirements of laws and regulations and personal information processing rules, and technical measures should be taken to ensure the security of relevant data.

[0073] Based on the above description, embodiments of this application provide a method for processing electroencephalogram (EEG) signals. This method can be applied to an EEG signal processing device (i.e., the method can be executed by the EEG signal processing device), and the EEG signal processing device can be connected to an EEG signal acquisition device; see also Figure 3a This is a flowchart illustrating an electroencephalogram (EEG) signal processing method provided in an embodiment of this application. The EEG signal processing method may include the following steps S301-S303:

[0074] S301 outputs N stimulus objects in the EEG signal processing device; N is a positive integer.

[0075] S302 receives the electroencephalogram (EEG) signals generated by the target object during the output process of N stimulus objects, obtained by the EEG signal acquisition device.

[0076] S303, perform the operation corresponding to the target stimulus; wherein, the target stimulus is determined from N stimulus objects based on the target object's EEG signal.

[0077] In one possible implementation, the stimulus objects may include visual stimulus images displayed at a stimulus frequency. Different stimulus objects have different stimulus frequencies or image features. The N visual stimulus images, each including one of the N stimulus objects, are displayed at intervals at different positions within the display area supported by the EEG signal processing device; or, the N visual stimulus images are displayed sequentially at the same position within the display area supported by the EEG signal processing device. Regarding the case where the N visual stimulus images are displayed at intervals at different positions within the display area supported by the EEG signal processing device, the layout of the N visual stimulus images can be designed according to specific needs. This application embodiment does not impose limitations. For example, if the N visual stimulus objects include a first visual stimulus object, a second visual stimulus object, a third visual stimulus object, and a fourth visual stimulus object, the layout of the N visual stimulus images can be as follows: Figure 2d As shown. In a scenario where N visual stimulus images are sequentially displayed at the same location within a display area supported by an EEG signal processing device, the EEG signal processing device can output prompting information to the target object. This prompts the target object to gaze at the visual stimulus image corresponding to a specific candidate intent if they wish to express that intent. This induces relevant EEG signals during the output of the visual stimulus image, enabling intent recognition based on the EEG signals and avoiding gazing at visual stimulus images that do not express the intended intent. It is understood that the specific content and format of the prompting information (e.g., audio format, text format, etc.) can be designed according to specific needs, and this embodiment does not impose any limitations.

[0078] In one possible implementation, different stimuli correspond to different candidate intent information. The EEG signal processing device performing an operation corresponding to a target stimulus may include performing an operation that matches the candidate intent information corresponding to the target stimulus. This matching operation may include the following processes: outputting the candidate intent information corresponding to the target stimulus; or sending an operation control command to the target device, causing the target device to perform an operation matching the candidate intent information corresponding to the target stimulus in response to the operation control command. The target device includes devices that can be controlled by the EEG signal processing device, such as the EEG signal processing device itself, a smart wheelchair, a smart home device, a smart wearable device, etc. For example, in one application scenario, if the target device is a smart wheelchair, the EEG signal processing device can control the smart wheelchair to perform operations such as stop, forward, and backward that match the candidate intent information corresponding to the target stimulus object, based on operation control commands. The corresponding candidate intent information can be, for example, information that can indicate the corresponding operation intent, such as the text information "stop" indicating the intention to stop, the text information "forward" indicating the intention to move forward, the text information "backward" indicating the intention to move backward, and so on. In another application scenario, if the target device is a smart curtain, the EEG signal processing device can control the smart curtain to perform operations such as open and close that match the candidate intent information corresponding to the target stimulus object, based on operation control commands.

[0079] In one possible implementation, before the EEG signal processing device performs the operation of matching candidate intent information corresponding to the target stimulus object, it may further: perform feature analysis on the EEG signal of the target object, and determine the target stimulus object that triggers the generation of the target object's EEG signal from N stimulus objects based on the feature analysis results; generate an intent recognition result based on the candidate intent information corresponding to the target stimulus object; wherein, the intent recognition result is used to trigger the execution of the operation that matches the candidate intent information corresponding to the target stimulus object. This process is followed by... Figure 3b The method embodiments shown are described in detail, and will not be repeated here.

[0080] In one possible implementation, the EEG signal processing device may further: display an intent recognition trigger interface within the EEG signal processing device; the intent recognition trigger interface includes a first trigger control and a second trigger control; when the first trigger control is triggered, the step of outputting N stimulus objects in the EEG signal processing device is executed; when the second trigger control is triggered, a data acquisition interface is displayed; the data acquisition interface is used to trigger the acquisition of pre-stored EEG signals of target objects from storage space. The first trigger control is used to trigger the process of receiving EEG signals acquired by the EEG signal acquisition device in real time and performing intent recognition, while the second trigger control is used to trigger the process of acquiring pre-stored EEG signals from storage space and performing intent recognition; for example, the first trigger control may be as follows: Figure 5c As indicated by the 502 error code, it displays as "Online Experiment," and the second trigger control can be as follows: Figure 5c As indicated by label 501, it is displayed as "offline experiment," and the EEG signal processing device can, as shown... Figure 5c The page includes a first trigger control and a second trigger control; the data acquisition interface (also called the data import interface) can, as shown by mark 503, import EEG signals from the EEG signals stored in the storage space by triggering the component displayed as "Select File" to perform the relevant process of intention recognition based on EEG signals. It should be understood that the page layout shown in the embodiments of this application is exemplary, and any page layout that can achieve the corresponding function is within the protection scope of the embodiments of this application.

[0081] In this embodiment, by identifying the target stimulus that induces the EEG signal of the target object from N stimulus objects, and then executing the operation corresponding to the target stimulus object, the EEG signal can be conveniently identified and the corresponding operation executed. Furthermore, the EEG signal processing device can output N stimulus objects corresponding to different candidate intent information; and after receiving the EEG signal generated by the target object during the output of the N stimulus objects from the EEG signal acquisition device, an operation matching the candidate intent information corresponding to the target stimulus object is executed; wherein, the target stimulus object is determined from the N stimulus objects based on the target object's EEG signal. By acquiring the object's EEG signal to perform intent recognition and execute the operation matching the candidate stimulus object (corresponding to the target stimulus object) expressing the corresponding intent, the intent recognition process is unaffected by environmental factors such as light, exhibiting high stability, wide applicability, and improved intent recognition accuracy, further enhancing the accuracy of operations performed based on intent recognition.

[0082] Based on the above description, this application provides another method for processing electroencephalogram (EEG) signals, see [link to relevant documentation]. Figure 3bThis is a flowchart illustrating another electroencephalogram (EEG) signal processing method provided in an embodiment of this application. The EEG signal processing method can be executed by an EEG signal processing device and may include the following steps S311-S314:

[0083] S311, output N candidate intent information and N stimulus objects corresponding to each of the N candidate intent information.

[0084] In this embodiment, the N stimulus objects are all different, and N is a positive integer; the candidate intention information refers to the information that can be used to represent the intention, and the stimulus object refers to the information that can be used to trigger the generation of EEG signals related to the stimulus object. The N candidate intention information and the N stimulus objects corresponding to the N candidate intention information can be set according to specific needs, and this embodiment does not impose any restrictions.

[0085] In one possible implementation, the N stimulus objects can be visual stimuli, including visual stimulus images displayed at a stimulus frequency. Different stimulus objects have different stimulus frequencies or image characteristics (i.e., different stimulus objects have different stimulus frequencies or visual stimulus images). The stimulus frequencies and visual stimulus images of the N stimulus objects can be set according to specific needs. For example, in one possible application scenario, the N stimulus objects output by the EEG signal processing device can be as follows: Figure 2d As shown; for ease of explanation, the embodiments of this application will subsequently use N stimulus objects with different stimulus frequencies and different visual stimulus images as examples. In one possible implementation, the N candidate intent information can be text information, based on Figure 2d Please refer to the stimulus objects shown. Figure 4a This is a schematic diagram illustrating the output of candidate intent information and stimulus objects provided in an embodiment of this application. It can be seen that... Figure 4a The layout of candidate intent information and its corresponding stimulus objects is merely an exemplary description. The layout of candidate intent information and its corresponding stimulus objects can be designed according to specific needs. This application's embodiments do not impose limitations. For example, please refer to... Figure 4b This is a schematic diagram illustrating another output candidate intent information and stimulus object provided in an embodiment of this application. Figure 4b The layout of candidate intent information and its corresponding stimulus methods and Figure 4a The layouts shown are different.

[0086] In one possible implementation, taking the EEG signal processing device outputting a stimulus object at a stimulation frequency as an example, such as displaying a visual stimulus image at a stimulation frequency, the EEG signal processing device can control the display of the visual stimulus image based on the stimulation frequency, for example, it can control the visual stimulus image to flash continuously, control the display brightness of the visual stimulus image to change continuously based on the stimulation frequency, and so on. In one optional implementation, the EEG signal processing device can use sinusoidal encoding to set the stimulation frequency. Sinusoidal encoding uses the frequency and phase information of a sinusoidal waveform to encode visual stimuli. A sine wave is a periodic waveform with distinct frequency and phase characteristics. Sinusoidally encoded visual stimuli are typically achieved by controlling the frequency of brightness or color changes on a display, which match the waveform of the sine wave. The EEG signal processing device can typically acquire the stimulation frequency and generate a sinusoidal waveform based on it, controlling the output of the stimulus object based on the generated sinusoidal waveform. When the stimulus object includes a visual stimulus image displayed at the stimulation frequency, the EEG signal processing device can control the display of the visual stimulus image based on the generated sinusoidal waveform. The visual stimulus image continuously flashes under the control of the sinusoidal waveform, and the flashing frequency and phase of the visual stimulus image match the sinusoidal waveform.

[0087] Using sinusoidal encoding to output stimuli offers the following advantages: ① High stability: The stability and periodicity of sinusoidal waveforms enable sinusoidally encoded visual stimuli to elicit stable and predictable SSVEP responses, helping to reduce noise and interference, improve the signal quality of the evoked EEG signals, and thus enhance the accuracy of intent recognition based on the evoked EEG signals. ② Good flexibility: By adjusting the frequency and phase of the sinusoidal waveform, visual stimuli of different frequencies can be easily achieved (i.e., stimuli can be conveniently output at different frequencies). ③ Improved stimulus precision: For stimuli with frequencies that are not divisible by an integer, sinusoidal encoding, through its continuous waveform changes, can effectively reduce the instability of the stimulus, thereby achieving higher visual stimulus precision, which is particularly important for applications requiring precise control of stimulus frequency. ④ Ease of implementation: The sinusoidal waveform is a simple and widely used mathematical function, making it relatively easy to set the stimulus frequency using sinusoidal encoding, thus simplifying the design and implementation process of visual stimuli.

[0088] In one possible implementation, if among the intentions indicated by the N output candidate intention information, there exists an intention that the target object wishes to express, then the target object can generate an EEG signal related to the perceived stimulus object by perceiving the stimulus object corresponding to the candidate intention information. In another possible implementation, the EEG signal processing device can also output prompting information to prompt the target object to express the intention expressed by the candidate intention information corresponding to the stimulus object by perceiving the stimulus object. The information type and content of this prompting information can be designed according to specific needs, and this application embodiment does not impose limitations. For example, the prompting information can be text information, audio information, video information, graphic information, etc.; for example, based on... Figure 4a The candidate intent information and the output method of the stimulus object are shown. The prompt information can be audio information such as "Please look at the image above the image you want to express" or "Please express the intent below the image by looking at the image".

[0089] S312, acquire the EEG signals generated by the target object during the output of N stimuli.

[0090] In one possible implementation, the EEG signal processing device can acquire the EEG signals generated by the target object during the output process of N stimulus objects through the EEG signal acquisition device. That is, the EEG signal acquisition device can acquire the EEG signals generated by the target object during the output process of N stimulus objects and transmit them to the EEG signal processing device. Optionally, after acquiring the EEG signals, the EEG signal processing device can output and display the EEG signals. In one optional implementation, the EEG signals generated by the target object during the output of N stimuli acquired by the EEG signal processing device can be EEG signals generated by the target object in a specific brain region. This specific brain region is the brain region where the EEG signals triggered by the stimuli output by the EEG signal processing device are generated. For example, when the stimuli are visual stimuli, since SSVEP signals mainly appear in the occipital lobes and parietal lobes of the brain, the specific brain region can be set to one or more of the occipital lobes and parietal lobes. By acquiring the EEG signals of a specific brain region and performing intention recognition based on the EEG signals of the specific brain region, interference from other EEG signals unrelated to the stimuli can be reduced, thereby improving the accuracy of intention recognition. Based on this, in one optional implementation, after outputting the stimulus, the EEG signal processing device can control the EEG signal acquisition device to collect the EEG signals generated by the target object from a specific brain region corresponding to the stimulus object during the output of N stimulus objects (i.e., collecting the EEG signals of the stimulus object in a specific brain region). The EEG signal acquisition device only collects and transmits the EEG signals from the specific brain region, which can save data transmission resources and improve data transmission efficiency. After receiving the EEG signals transmitted by the EEG signal acquisition device, the EEG signal processing device can filter out the EEG signals from the specific brain region corresponding to the stimulus object to obtain the EEG signals generated by the target object during the output of N stimulus objects.

[0091] Please see Figure 4c This diagram illustrates the electrode distribution of an electroencephalogram (EEG) signal acquisition device according to an embodiment of this application. It represents the electrode placement standard recommended by the International Society for Electroencephalography (ESE). Odd numbers indicate electrodes located in the left hemisphere, and even numbers indicate electrodes located in the right hemisphere. The size of the number indicates the distance of the electrode from the midline; electrodes closer to the midline have smaller numbers. The letters in the diagram represent abbreviations for the brain regions corresponding to the electrodes. For example, electrodes P and O correspond to the occipital lobe and parietal lobe, respectively. Additionally, the letter A corresponds to the auricular electrode. Electrodes A1 and A2 are typically used as reference electrodes in EEG signal acquisition devices and are therefore also used when acquiring EEG signals.

[0092] The EEG signal acquisition device can be selected according to specific needs, and the embodiments of this application do not impose any limitations. For example, an electrode-based EEG signal acquisition device can be selected; for instance, the DSI-24 EEG signal acquisition device based on dry electrodes, or the EPOC Flex Saline Sensor EEG signal acquisition device based on wet electrodes, etc. Generally speaking, EEG signal acquisition devices based on wet electrodes typically have advantages such as good electrode conductivity, high signal-to-noise ratio, and superior performance, while EEG signal acquisition devices based on dry electrodes have significant advantages in terms of ease of operation and wearing comfort, and electrode adjustment is simple. Therefore, the EEG signal acquisition device can be selected according to specific needs. Please refer to [link to relevant documentation]. Figure 4d This application provides a schematic diagram of the electrode distribution of another electroencephalogram (EEG) signal acquisition device, illustrating the electrodes of the DSI-24 EEG signal acquisition device. Figure 4c The electrode correspondence shown is as follows: the electrodes of the DSI-24 specifically correspond to the following electrodes: Fp1, Fp2, Fz, F3, F4, F7, F8, Cz, C3, C4, T7 / T3, T8 / T4, Pz, P3, P4, P7 / T5, P8 / T6, O1, O2, A1, A2. The DSI-24 EEG signal acquisition device has the following advantages: easy to wear and can be worn for extended periods, supporting low-to-medium intensity exercise; portable design for easy deployment; supports Bluetooth wireless transmission with a sampling rate of 300Hz; and features a hot-swappable battery for continuous operation.

[0093] S313, perform feature analysis on the EEG signal of the target object, and determine the target stimulus object that triggers the generation of the EEG signal of the target object from N stimulus objects based on the feature analysis results.

[0094] Since different stimuli can be used to trigger the generation of EEG signals related to the corresponding stimuli, the purpose of performing feature analysis on the EEG signals of the target object and determining the target stimulus object that triggers the generation of the target object's EEG signals from among N stimuli objects is to analyze which of the N stimuli objects evokes the EEG signals that the target object's EEG signals satisfy, so as to determine the target stimulus object from among the N stimuli objects.

[0095] In one possible implementation, the EEG signal processing device performs feature analysis on the EEG signal of the target object and determines the target stimulus object that triggers the generation of the EEG signal of the target object from N stimulus objects based on the feature analysis results. This process can involve the following steps: preprocessing the EEG signal of the target object to obtain an intermediate EEG signal; the intermediate EEG signal is the signal within the target frequency band of the target object's EEG signal; performing feature extraction processing on the intermediate EEG signal to obtain EEG signal features; the feature analysis results include the EEG signal features; and determining the target stimulus object that triggers the generation of the EEG signal of the target object from the N stimulus objects based on the EEG signal features and the reference EEG signal features associated with each of the N stimulus objects.

[0096] This refers to the preprocessing process, specifically the process of "preprocessing the EEG signals of the target object to obtain intermediate EEG signals".

[0097] Preprocessing is used to improve signal quality and usability. For example, during the acquisition of EEG signals, EEG signal acquisition devices may be subject to interference factors. These interference factors can create artifacts by superimposing their signals onto the acquired EEG signals. These artifacts can affect the accuracy of intent recognition. Therefore, preprocessing is necessary before directly using the acquired EEG signals to filter out artifacts and improve the signal quality. Alternatively, depending on specific needs, preprocessing can be omitted to directly extract features from the target object's EEG signals, obtain EEG signal features, and then determine the target stimulus object based on these features.

[0098] Preprocessing typically includes operations such as filtering and noise reduction. Filtering removes unwanted frequency bands from the signal to retain the desired target frequency band. The range of the target frequency band can be set according to specific needs, and this application embodiment does not impose any limitations. For example, SSVEP signals have high time-varying sensitivity, and their interference components can be classified into the following categories: power frequency interference from electromagnetic waves in the form of voltage to the EEG signal acquisition device, electromyographic interference caused by muscle movement, electrode interference caused by poor electrode contact or poor conductivity of electrode materials, cardiac interference with a frequency of approximately 1 Hz caused by heartbeat, and electrooculography artifacts caused by blinking and eye movement. Generally, electromyographic artifacts and electrooculography artifacts are mainly concentrated at low frequencies (less than 5 Hz). For example, a high-pass filter can be used to filter out low-frequency artifact signals. Power frequency interference is usually a high-frequency signal (e.g., 50 Hz), and a low-pass filter can be used to filter out high-frequency artifact signals. Of course, a band-pass filter can also be used to filter out low-frequency and high-frequency signals, and the appropriate method can be selected according to specific needs. This application embodiment does not impose any limitations. For example, if N=4, and the stimulation frequencies of the first to fourth stimulus objects are 8Hz, 9Hz, 10Hz, and 11Hz respectively, SSVEP has the characteristic of harmonic response, that is, SSVEP not only generates electrical signals at the stimulation frequency, but also responds at harmonics of the stimulation frequency (usually the 2nd and 3rd harmonics). Based on this, the target frequency range in online tasks can be set to 6Hz to 30Hz, and the target frequency range in offline tasks can be set to 6Hz to 40Hz. Among them, online tasks include tasks that collect EEG signals in real time and perform intention recognition, and offline tasks include tasks that use offline data (such as EEG signals collected in historical time periods) to perform intention recognition. The main reason for setting different target frequency ranges for offline and online tasks is that offline tasks usually do not have much requirement for real-time performance, so a relatively wide filtering range can be set to capture more signal features. Online tasks, on the other hand, usually focus more on real-time performance and efficiency, and need to process data and respond in a short time. Therefore, it may be necessary to limit the filtering range more precisely to ensure fast and accurate data processing.

[0099] The process related to feature extraction, namely "the process of extracting features from intermediate EEG signals to obtain EEG signal features", refers to the extraction of useful feature information from signals to facilitate subsequent classification and recognition. Feature extraction methods include extracting time-domain features, extracting frequency-domain features, and extracting time-frequency-domain features.

[0100] In one possible implementation, the EEG signal features obtained by the EEG signal processing device through feature extraction processing of the intermediate EEG signal may include: EEG feature values ​​of the intermediate EEG signal at multiple frequencies, wherein the EEG feature value of the intermediate EEG signal at any frequency includes: the power of the intermediate EEG signal at the corresponding frequency; that is, the EEG signal features obtained by the EEG signal processing device through feature extraction processing of the intermediate EEG signal are: the power spectral density obtained by processing the intermediate EEG signal, which includes: the power of the intermediate EEG signal at multiple frequencies; further optionally, the power spectral density can be represented as a graph for observation, with the horizontal axis of the power spectral density graph representing frequency and the vertical axis representing the power corresponding to the frequency, that is, representing the power of the signal at the corresponding frequency. The method for determining the power spectral density can be selected according to specific needs, and this application embodiment does not impose any limitations; for example, a determination method based on Discrete Fourier Transform (DFT) or a determination method based on Fast Fourier Transform (FFT) can be selected.

[0101] In another possible implementation, the EEG signal features obtained by the EEG signal processing device through feature extraction processing of the intermediate EEG signal may include: EEG feature values ​​of the intermediate EEG signal at multiple frequencies, wherein the EEG feature value of the intermediate EEG signal at any frequency includes: the normalized result of the intermediate EEG signal at the corresponding frequency; based on this, the process of the EEG signal processing device performing feature extraction processing on the intermediate EEG signal to obtain EEG signal features may include the following steps: performing frequency domain transformation processing on the intermediate EEG signal to obtain multiple frequency domain signals; wherein, one frequency domain signal corresponds to one frequency; performing normalization processing on the amplitude of the multiple frequency domain signals respectively to obtain the normalized result of the intermediate EEG signal at the frequency corresponding to each frequency domain signal; and constructing EEG signal features based on the frequency corresponding to each frequency domain signal and the normalized result of the intermediate EEG signal at the corresponding frequency. As a further option, the EEG signal features constructed based on the normalization results can be represented as a graph for observation. The horizontal axis of the normalization result graph represents frequency, and the vertical axis represents the normalization result corresponding to the frequency, that is, the normalization result of the signal at the corresponding frequency.

[0102] The EEG signal processing device performs frequency domain transformation on the intermediate EEG signal to obtain multiple frequency domain signals. These signals can be obtained by performing a Fourier transform on the intermediate EEG signal. Optionally, a Discrete Fourier Transform (DFT) or a Fast Fourier Transform (FFT) can be performed on the intermediate EEG signal. This application embodiment does not impose limitations, but for ease of explanation, the following embodiments of this application will use FFT as an example. According to Fourier analysis theory, any complex signal (including EEG) can be regarded as the sum of sine and cosine basis functions with specific frequencies. Therefore, the intermediate EEG signal can be decomposed into sine and cosine waves in a continuous frequency range through Fourier transform, that is, decomposed into frequency domain signals at multiple frequencies.

[0103] The process of normalizing the amplitudes of multiple frequency domain signals to obtain the normalized results of the intermediate EEG signal at the corresponding frequencies of each frequency domain signal can be illustrated by the following formula 1:

[0104]

[0105] Where P represents the normalized result of the intermediate EEG signal at the corresponding frequency of each frequency domain signal, x(n) represents the intermediate EEG signal, FFTx(n) represents the amplitude of each frequency domain signal obtained by performing a Fast Fourier Transform on the intermediate EEG signal, and ∑FFTx(n) represents the sum of the amplitudes of each frequency domain signal obtained by performing a Fast Fourier Transform on the intermediate EEG signal. This represents the ratio of the amplitude of each frequency domain signal obtained by performing a Fast Fourier Transform (FFT) on the intermediate EEG signal to the sum of the amplitudes of the signals at each frequency. The larger the normalization result at a certain frequency, the stronger the response of the EEG signal at that frequency. After converting the time domain signal to the frequency domain using FFT, normalizing the amplitude of the obtained frequency domain signal can effectively measure the intensity ratio of different frequency domain signals among all frequency domain signals. In other words, it can effectively measure the response degree of the EEG signal at different frequencies, and can more accurately extract features related to the stimulation frequency, thereby improving the accuracy of intention recognition based on these features.

[0106] The process of determining the target stimulus object based on the characteristics of EEG signals is as follows: "Based on the characteristics of EEG signals and the reference EEG signal characteristics associated with each of the N stimulus objects, the target stimulus object that triggers the generation of the target object's EEG signal is determined from the N stimulus objects."

[0107] The reference EEG signal features associated with any stimulus object can be obtained by feature extraction processing of the reference EEG signal corresponding to the corresponding stimulus object. The relevant process of feature extraction processing of the reference EEG signal corresponding to the stimulus object is similar to the relevant process of feature extraction processing of the intermediate EEG signal described above, and will not be repeated here. The feature terms of the reference EEG signal features are the same as the feature terms of the EEG signal features. That is, if the EEG signal features are composed of power, then the reference EEG signal features are also composed of power. If the EEG signal features are composed of normalization results, then the reference EEG signal features are also composed of normalization results. Optionally, the reference EEG signal corresponding to the stimulus object can be selected according to specific needs. This application embodiment does not impose any restrictions. For example, the reference EEG signal corresponding to the stimulus object can be the EEG signal generated by the sample object during the output process of the corresponding stimulus object.

[0108] In one possible implementation, the stimulus object includes a visual stimulus image displayed at a stimulus frequency. Different stimulus objects have different stimulus frequencies or image features. The EEG signal features include: EEG feature values ​​of intermediate EEG signals at multiple frequencies. The EEG feature value of intermediate EEG signals at any frequency includes: the power of intermediate EEG signals at the corresponding frequency, or the normalized result of intermediate EEG signals at the corresponding frequency. In the process of determining the target stimulus object that triggers the generation of the target object's EEG signal from the N stimulus objects based on the EEG signal features and the reference EEG signal features associated with each of the N stimulus objects, the EEG signal processing device may perform the following steps: determining the feature similarity between the EEG signal features and the reference EEG signal features associated with each of the N stimulus objects; and determining the stimulus object indicated by the maximum feature similarity as the target stimulus object that triggers the generation of the target object's EEG signal. The feature similarity between the EEG signal features and the reference EEG signal features associated with any stimulus object can be used to indicate the degree of similarity between the distribution of the EEG feature values ​​indicated by the EEG signal features at multiple frequencies and the distribution of the EEG feature values ​​indicated by the reference EEG signal features at multiple frequencies. The more similar the distribution, the greater the likelihood that the EEG signal representing the target object was triggered by that stimulus object. For example, in an optional implementation, determining the feature similarity between the EEG signal features and the reference EEG signal features associated with any stimulus object can be achieved through a similarity model. This similarity model can be selected according to specific needs, such as a machine learning model, a neural network model, etc., and this application embodiment does not impose any limitations.

[0109] In another possible implementation, the stimulus object includes a visual stimulus image displayed at a stimulus frequency. Different stimulus objects have different stimulus frequencies or image characteristics. The EEG signal characteristics include: EEG feature values ​​of intermediate EEG signals at multiple frequencies. The EEG feature value of the intermediate EEG signal at any frequency includes: the power of the intermediate EEG signal at the corresponding frequency, or the normalized result of the intermediate EEG signal at the corresponding frequency. In the process of determining the target stimulus object that triggers the generation of the target object's EEG signal from the N stimulus objects based on the EEG signal characteristics and the reference EEG signal characteristics associated with each of the N stimulus objects, the EEG signal processing device may perform the following steps: The EEG feature value at the i-th stimulation frequency is determined from the EEG signal features of the target object, and the reference EEG feature value at the i-th stimulation frequency is determined based on the reference EEG signal features associated with the i-th stimulation object; wherein, the i-th stimulation frequency is the stimulation frequency of the i-th stimulation object among N stimulation objects, i∈[1,N]; the feature difference between the EEG feature value corresponding to the i-th stimulation frequency and the reference EEG feature value is determined to obtain the feature difference corresponding to the i-th stimulation frequency, until the feature differences corresponding to each of the N stimulation frequencies are obtained; the stimulation object corresponding to the stimulation frequency with the smallest feature difference is determined as the target stimulation object that triggers the generation of the EEG signal of the target object. For ease of explanation, in this embodiment, the stimulation frequency of the i-th stimulation object among N stimulation objects is called the i-th stimulation frequency, and the candidate intent information corresponding to the i-th stimulation object is called the i-th candidate intent information.

[0110] For example, if N=4, and the stimulation frequencies of the first to fourth stimulation objects are 8Hz, 9Hz, 10Hz, and 11Hz respectively, the EEG signal processing device can determine the reference EEG feature value at the first stimulation frequency (8Hz) based on the reference EEG signal features associated with the first stimulation object, the reference EEG feature value at the second stimulation frequency (9Hz) based on the reference EEG signal features associated with the second stimulation object, the reference EEG feature value at the third stimulation frequency (10Hz) based on the reference EEG signal features associated with the third stimulation object, and the reference EEG feature value at the fourth stimulation frequency (11Hz) based on the reference EEG signal features associated with the fourth stimulation object. Optionally, the feature difference between the EEG feature value corresponding to the i-th stimulation frequency and the reference EEG feature value can be the absolute value of the difference between the EEG feature value corresponding to the i-th stimulation frequency and the reference EEG feature value, or the value after weighted processing of the absolute value of the difference. This application embodiment does not impose any limitations. The smaller the feature difference between the EEG feature value corresponding to the i-th stimulation frequency and the reference EEG feature value, the closer the response of the target object's EEG signal at the i-th stimulation frequency is to the response of the reference EEG signal induced by the i-th stimulation object at the i-th stimulation frequency. Therefore, the stimulation object corresponding to the stimulation frequency with the smallest feature difference can be identified as the target stimulation object that triggers the generation of the target object's EEG signal.

[0111] In another possible implementation, where the stimulus includes a visual stimulus image displayed at a stimulus frequency, and different stimuli have different stimulus frequencies, the SSVEP signal induced by the visual stimulus has the strongest response near the stimulus frequency. For example, see [link to relevant documentation]. Figure 2b The SSVEP signal induced by a 10Hz stimulation frequency has the highest power near 10Hz. Based on this, the EEG signal processing device can also determine the EEG feature value at each of the N stimulation frequencies and the highest EEG feature value in the EEG signal characteristics of the target object. The stimulation object corresponding to the stimulation frequency of the EEG feature value with the smallest difference from the highest EEG feature value at each stimulation frequency is determined as the target stimulation object that triggers the generation of the target object's EEG signal.

[0112] In one possible implementation, the stimulus object includes a visual stimulus image displayed at a stimulus frequency. Different stimulus objects have different stimulus frequencies or image features. The EEG signal processing device performs feature analysis on the EEG signal of the target object and determines the target stimulus object that triggers the generation of the target object's EEG signal from N stimulus objects based on the feature analysis results. The following steps can be performed: determining the reference matrix corresponding to the i-th stimulus frequency among the N stimulus frequencies; where i∈[1,N], the reference matrix corresponding to the i-th stimulus frequency is obtained based on the reference EEG signal corresponding to the i-th stimulus frequency; performing canonical correlation analysis on the target object's EEG signal and the reference matrix corresponding to the i-th stimulus frequency to obtain the correlation coefficient corresponding to the i-th stimulus frequency; where the correlation coefficient corresponding to the i-th stimulus frequency indicates the degree of matching between the target object's EEG signal and the reference matrix corresponding to the i-th stimulus frequency; using the correlation coefficients corresponding to each of the N stimulus frequencies as feature analysis results, and determining the target stimulus object that triggers the generation of the target object's EEG signal from N stimulus objects based on the feature analysis results.

[0113] For ease of explanation, in this embodiment, the stimulation frequency of the i-th stimulus object among N stimulus objects is referred to as the i-th stimulation frequency, and the candidate intent information corresponding to the i-th stimulus object is referred to as the i-th candidate intent information; wherein, the reference EEG signal corresponding to the i-th stimulation frequency is the reference EEG signal corresponding to the i-th stimulus object. The reference EEG signal corresponding to the stimulus object can be selected according to specific needs, and this embodiment does not impose any restrictions. For example, the reference EEG signal corresponding to the stimulus object can be the EEG signal generated by the sample object during the output process of the corresponding stimulus object. The correlation coefficient corresponding to the i-th stimulation frequency is used to indicate the degree of matching between the target object's EEG signal and the reference matrix corresponding to the i-th stimulation frequency. Furthermore, it can also indicate the degree of matching between the target object's EEG signal and the reference EEG signal corresponding to the i-th stimulation frequency. The greater the matching degree, the more similar the EEG signal is to the reference EEG signal. Based on this, when the EEG signal processing device determines the target stimulus object from N stimulus objects to trigger the generation of the target object's EEG signal according to the feature analysis results, it can select the stimulus object corresponding to the stimulation frequency with the largest correlation coefficient as the target stimulus object. For example, if the stimulation frequency with the largest correlation coefficient is the N-th stimulation frequency, then the N-th stimulus object corresponding to the N-th stimulation frequency can be determined as the target stimulus object. In other words, the core idea of ​​the Canonical Correlation Analysis (CCA) algorithm is to calculate the correlation coefficient between the SSVEP signal and the template signal (i.e., the reference EEG signal) corresponding to the N stimulation frequencies, and select the maximum correlation coefficient to complete the matching between the SSVEP signal and the template signal.

[0114] Please refer to Equation 2.1 below, which shows how the reference matrix corresponding to the i-th stimulus frequency is determined:

[0115]

[0116] Among them, Y i Let Y represent the reference matrix corresponding to the i-th stimulus frequency. Since the SSVEP response is mainly composed of its fundamental frequency, first harmonic, and second harmonic, the reference matrix Y is... i The structure should also include the fundamental frequency and multiple harmonic components; N h f represents the harmonic order. s The sampling frequency used by the EEG signal acquisition device to acquire EEG signals, where N is the number of sampling points.

[0117] The process of performing canonical correlation analysis on the EEG signal of the target object and the reference matrix corresponding to the i-th stimulus frequency to obtain the correlation coefficient corresponding to the i-th stimulus frequency can be shown by the following formula 2.2:

[0118]

[0119] Where X represents the EEG signal of the target object, which includes m multi-channel signals. The signals of different channels are acquired through different electrodes in the EEG signal acquisition device. The EEG signal of the target object can be represented as: X = (x1, x2, ..., x...) m ) T m is a positive integer; the CCA algorithm finds a set of linear transformations W X and W Y Make X and the reference matrix Y i The correlation coefficient is the largest; for example, the CCA algorithm can be set to use four electrodes, O1, O2, P3, and P4, and use the PZ electrode as a reference to obtain multi-channel EEG signals.

[0120] After obtaining the correlation coefficient for each of the N stimulation frequencies, the EEG signal processing device can use the stimulation frequency corresponding to the maximum correlation coefficient as the target stimulation frequency, and determine the stimulus object corresponding to the target stimulation frequency as the target stimulus object. The method for determining the target stimulation frequency can be shown in the following formula 2.3, where, f represents the target stimulus frequency. i ρ represents the frequency of the i-th stimulus. i The correlation coefficient corresponding to the i-th stimulus frequency is:

[0121]

[0122] As we know, FFT can analyze single-channel EEG signals, while CCA algorithm can compare multi-channel EEG signals with template signals in applications, reducing the impact of noise and improving information transmission rate. In addition, since the output of CCA algorithm is the correlation coefficient of EEG signal relative to each template signal, in some application scenarios, such as finding EEG signals with good evoked effects, the judgment of the maximum correlation coefficient can be used to further identify EEG signals with poor evoked effects and remove them. For example, by setting a correlation coefficient threshold, EEG signals with a maximum correlation coefficient lower than the threshold can be removed. EEG signals with a maximum correlation coefficient lower than the threshold are considered to have poor evoked effects, that is, the matching degree between the EEG signal and the template signal at each stimulation frequency is very low.

[0123] S314, Generate intent recognition results based on the candidate intent information corresponding to the target stimulus object.

[0124] The intent recognition result is used to trigger the execution of an operation that matches the candidate intent information corresponding to the target stimulus. Optionally, the EEG signal processing device can, in response to the intent recognition result, execute an operation that matches the candidate intent information corresponding to the target stimulus.

[0125] In one possible implementation, the candidate intent information corresponding to the target stimulus can be used as the intent recognition result. In another possible implementation, the candidate intent information corresponding to the target stimulus can be processed by information conversion to obtain the intent recognition result; information conversion may include, but is not limited to, information format conversion (e.g., converting audio format candidate intent information into text format as the intent recognition result), information content conversion (e.g., converting text format candidate intent information "go to the toilet" into text format intent recognition result "the recognized intent is: go to the toilet"), etc. The embodiments of this application are not limited, as long as an intent recognition result that can indicate the candidate intent information corresponding to the target stimulus is generated.

[0126] In an exemplary application scenario, N=4, and the first to fourth candidate intent information are respectively: the text "up" representing an "up" intent, the text "down" representing a "down" intent, the text "left" representing a "left" intent, and the text "right" representing a "right" intent. The stimulus objects corresponding to the first to fourth candidate intent information are different images; please refer to [link to relevant documentation]. Figure 4eThis is a schematic diagram of another object intent recognition method provided in this application embodiment. The EEG signal processing device can output the above-mentioned four candidate intent information and four stimulus objects. One of the candidate intent information can be the text "down" as shown in 401, and the corresponding stimulus object can be the text "down". If the EEG signal processing device obtains the EEG signal generated by the target object during the output of the four stimulus objects, and determines the target stimulus object as the stimulus object shown in 402 based on the EEG signal, it can generate an intent recognition result for triggering the execution of an operation that matches the candidate intent information (i.e., the text "down") corresponding to the target stimulus object. Further optionally, the EEG signal processing device can respond to the intent recognition result and execute an operation that matches the candidate intent information corresponding to the target stimulus object, such as outputting the candidate intent information corresponding to the target stimulus object, i.e., outputting the text "down". For example, if the first to fourth candidate intent information are respectively: "page up" representing the intent to "page up", "page down" representing the intent to "page down", "back to the previous page" representing the intent to "return to the previous level", and "close application" representing the intent to "close application", and if the target stimulus is determined to be the fourth stimulus based on the EEG signal, an intent recognition result can be generated to trigger the execution of an operation that matches the candidate intent information (i.e., the text "close application") corresponding to the target stimulus. Further optionally, the EEG signal processing device can respond to the intent recognition result and execute an operation that matches the candidate intent information corresponding to the target stimulus, i.e., close the application.

[0127] Please see Figure 5a This is a schematic diagram of an intent recognition process provided in an embodiment of this application. First, the target object can wear an EEG signal acquisition device. After the target object wears the EEG signal acquisition device, the EEG signal processing device can output N candidate intent information and N stimuli corresponding to each of the N candidate intent information. The EEG signal processing device can acquire the EEG signals generated by the target object during the output of the N stimuli through the EEG signal acquisition device. Then, the EEG signal processing device can obtain the intent recognition result based on the target object's EEG signals. In this process, the EEG signal processing device can preprocess the EEG signals to obtain intermediate EEG signals (i.e., the preprocessing process), extract features to obtain EEG signal features (i.e., the feature extraction process), and then perform intent recognition based on the EEG signal features to obtain the intent recognition result (i.e., the intent recognition process).

[0128] Please see Figure 5bThis is a schematic diagram of another intent recognition process provided in an embodiment of this application. When the data import method is selected, the EEG signal processing device can import and read EEG signals from stored EEG signals for subsequent intent recognition processes based on EEG signals, adapting to offline tasks. Alternatively, when the data acquisition method is selected, the EEG signal processing device can acquire and store EEG signals through an EEG signal acquisition device, and import and read EEG signals from the stored EEG signals for subsequent intent recognition processes based on EEG signals, adapting to online tasks. After acquiring EEG signals, the EEG signal processing device can perform preprocessing, feature extraction, intent recognition, and other related processes to obtain intent recognition results. Optionally, the EEG signal processing device also supports displaying the acquired EEG signals in a graphical format, displaying preprocessed data in a graphical format, displaying feature-extracted data in a graphical format, displaying intent recognition results in a graphical format, and so on.

[0129] Please see Figure 5c This is a schematic diagram of an interface interaction provided in an embodiment of this application. By triggering the offline task option as marked 501, an offline task can be executed. By triggering the online task option as marked 502, an online task can be executed. The offline task can flexibly process data, is highly operable, and is convenient for information analysis. The online task has strong real-time performance and is closer to practical applications. If the offline task option as marked 501 is triggered, it can jump to the data import interface as marked 503 to import the EEG signal from the stored EEG signal in the form of a file import and read it for subsequent intention recognition based on the EEG signal.

[0130] In this embodiment, N candidate intent information and their corresponding stimuli can be output. By acquiring the EEG signals generated by the target object during the output of the N stimuli, and based on the feature analysis of the acquired target object's EEG signals, the target stimulus that triggers the generation of the target object's EEG signals is determined from the N stimuli. Then, an intent recognition result can be generated based on the candidate intent information corresponding to the target stimulus. Using the target's EEG signals for intent recognition makes the process unaffected by environmental factors such as light, resulting in high accuracy, high stability, and wide applicability. Furthermore, by outputting different stimuli corresponding to the N candidate intent information, the target object can trigger the generation of EEG signals related to the corresponding stimuli by perceiving different stimuli. That is, the EEG signals triggered by different stimuli have strong specificity, allowing for effective differentiation when performing intent recognition based on EEG signals triggered by different stimuli, thus improving the accuracy of intent recognition.

[0131] Based on the above description, this application provides yet another method for processing electroencephalogram (EEG) signals, see [link to relevant documentation]. Figure 6 This is a flowchart illustrating another brainwave signal processing method provided in this application embodiment; the brainwave signal processing method can be executed by a brainwave signal processing device, and the brainwave signal processing method may include the following steps S601-S607:

[0132] S601 sequentially outputs M candidate stimulus objects and acquires the sample EEG signals generated by the sample object during the output process of different candidate stimulus objects.

[0133] In this embodiment, the M candidate stimuli are all different, where M is a positive integer greater than or equal to N. The M candidate stimuli can be set according to specific needs, and this embodiment does not impose any restrictions. When the stimulus is a visual stimulus, the candidate stimuli may include visual stimulus images displayed at a stimulus frequency. Different candidate stimuli have different stimulus frequencies or image features (i.e., different candidate stimuli have different stimulus frequencies or visual stimulus images). For ease of explanation, this embodiment will subsequently use the example of M candidate stimuli having different stimulus frequencies and different visual stimulus images. The sample object refers to the object (i.e., the user) whose EEG signals need to be collected to determine the stimulus object, and this embodiment does not impose any restrictions. The process of the EEG signal processing device outputting candidate stimuli and acquiring the relevant sample EEG signals generated by the sample object during the output of the corresponding candidate stimuli is similar to the process described above where the EEG signal processing device outputs N stimuli and acquires the relevant sample EEG signals generated by the target object during the output of the N stimuli, and will not be described in detail here.

[0134] S602, process the sample EEG signals corresponding to the M candidate stimuli respectively to obtain the power spectral density corresponding to each of the M candidate stimuli.

[0135] The process by which the EEG signal processing device processes the sample EEG signals corresponding to the candidate stimulus object to obtain the power spectral density of the corresponding candidate stimulus object is similar to the process by which the EEG signal processing device determines the power spectral density of the target object based on the EEG signal of the target object, and will not be described in detail here.

[0136] S603, based on the power spectral density corresponding to each of the M candidate stimuli, determine N stimuli from the M candidate stimuli.

[0137] The N identified stimulus objects refer to the candidate stimulus objects whose power spectral density meets the screening criteria among the M candidate stimulus objects.

[0138] In one possible implementation, the EEG signal processing device, in determining N stimulus objects from M candidate stimulus objects based on their respective power spectral densities, may perform the following steps: calling an evaluation model to evaluate the power spectral densities of the M candidate stimulus objects, obtaining evaluation values ​​for each of the M candidate stimulus objects; and selecting N candidate stimulus objects from the M candidate stimulus objects whose evaluation values ​​are greater than a reference evaluation value, as candidate stimulus objects whose power spectral densities meet the selection criteria, thus obtaining N stimulus objects. The evaluation model can be selected according to needs, such as machine learning models or deep learning models in the field of artificial intelligence. This application embodiment does not impose any restrictions. For example, the evaluation model can be a scoring model. After the power spectral density corresponding to the candidate stimulus is input into the scoring model, the scoring model can score the power spectral density corresponding to the candidate stimulus. The score can measure the degree to which the power spectral density corresponding to the candidate stimulus meets the characteristics that the power spectral density of the EEG signal theoretically induced by the candidate stimulus should have. In other words, the higher the score, the better the induction effect of the candidate stimulus. Based on this, the EEG signal processing device can select N candidate stimulus objects with evaluation values ​​greater than the reference evaluation value from M candidate stimulus objects as candidate stimulus objects whose power spectral density meets the screening conditions, thus obtaining N stimulus objects. The reference evaluation value can be set according to specific needs, and this application embodiment does not impose any restrictions. When an EEG signal processing device selects N candidate stimulus objects from M candidate stimulus objects whose evaluation values ​​are greater than a reference evaluation value, it can use multiple selection methods. For example, it can select and filter in descending order of evaluation values, randomly select N candidate stimulus objects from each candidate stimulus object whose evaluation value is greater than the reference evaluation value, and sequentially determine whether the evaluation value of the candidate stimulus object is greater than the reference evaluation value. When it is determined that the evaluation value is greater than the reference evaluation value, the corresponding candidate stimulus object is determined as the stimulus object, until N stimulus objects are determined, etc. The selection method can be selected according to specific needs, and the embodiments of this application do not impose any limitations.

[0139] Artificial intelligence (AI) as described above refers to the theories, methods, technologies, and application systems that utilize digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results. In other words, AI is a comprehensive technology within computer science that attempts to understand the essence of intelligence and produce a new type of intelligent machine capable of reacting in a manner similar to human intelligence. AI studies the design principles and implementation methods of various intelligent machines, enabling them to possess perception, reasoning, and decision-making capabilities. AI technology is a comprehensive discipline involving a wide range of fields, encompassing both hardware and software technologies. Fundamental AI technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, pre-trained model technology, operating / interactive systems, and mechatronics. Pre-training models, also known as foundational models or large models, refer to deep neural networks (DNNs) with large parameters. These DNNs are trained on massive amounts of unlabeled data, leveraging the function approximation capabilities of large-parameter DNNs to enable PTMs (Programmable Memory Models) to extract common features from the data. Through fine-tuning and parameter-efficient fine-tuning techniques (PEFT), they are suitable for downstream tasks. Artificial intelligence software technologies mainly include computer vision, speech processing, natural language processing, and machine learning / deep learning. Machine learning (ML) is a multidisciplinary field involving probability theory, statistics, approximation theory, convex analysis, and algorithm complexity theory. It specifically studies how computers can simulate or implement human learning behavior to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental way to endow computers with intelligence; its applications span all areas of artificial intelligence. Machine learning and deep learning typically include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and pre-trained learning. Pre-trained models are the latest development in deep learning, integrating all of these techniques.

[0140] In one possible implementation, the power spectral density corresponding to the target candidate stimulus includes: the power of the EEG signal generated by the target candidate stimulus at multiple frequencies, wherein the target candidate stimulus is any one of M candidate stimuli; the power spectral density corresponding to the target candidate stimulus satisfies the screening conditions including: the frequency corresponding to the maximum power peak in the power spectral density corresponding to the target candidate stimulus matches the stimulation frequency of the target candidate stimulus, and the minimum peak difference between the maximum power peak in the power spectral density corresponding to the target candidate stimulus and the other power peaks is greater than a target difference threshold; wherein the target difference threshold is determined based on the maximum power peak.

[0141] The frequency corresponding to the maximum power peak in the power spectral density of the target candidate stimulus is matched with the stimulus frequency of the target candidate stimulus to determine whether the target candidate stimulus has the strongest response at its stimulus frequency. For example, by... Figure 2b It is known that the SSVEP signal induced by a 10Hz stimulation frequency has the highest power near 10Hz. The EEG signal processing device can determine the frequency corresponding to the maximum power peak in the power spectral density of the target candidate stimulus object and match it with the stimulation frequency of the target candidate stimulus object if the frequency interval between the frequency corresponding to the maximum power peak in the power spectral density of the target candidate stimulus object and the stimulation frequency of the target candidate stimulus object is less than a specified frequency threshold. The frequency threshold can be set according to specific needs, and this application embodiment does not impose any restrictions. The greater the difference between the maximum power peak and the minimum peak value in the power spectral density corresponding to the target candidate stimulus, i.e., the greater the difference between the maximum power peak and the second largest power peak, the stronger the response of the EEG signal induced by the target candidate stimulus at its stimulation frequency. Based on this, the intensity of the response of the EEG signal induced by the target candidate stimulus at its stimulation frequency can be measured by determining whether the difference between the minimum peak value and the minimum peak value in the power spectral density corresponding to the target candidate stimulus is greater than the target difference threshold. This can then be used to measure the stimulation effect of the target candidate stimulus. The target difference threshold can be set according to specific needs. For example, the target difference threshold can be determined based on the maximum power peak. For example, the target difference threshold can be set to 80% of the maximum power peak, 60% of the maximum power peak, etc.

[0142] In one possible implementation, if there are fewer than N candidate stimuli whose power spectral density meets the screening criteria among the M candidate stimuli, the EEG signal processing device can acquire new candidate stimuli and determine whether to identify the new candidate stimuli as stimuli based on the power spectral density corresponding to the new candidate stimuli, until N stimuli are identified.

[0143] For example, if M = N = 4, the stimulus frequency of the first candidate stimulus is 8 Hz, and the visual stimulus image is a red patch; the stimulus frequency of the second candidate stimulus is 9 Hz, and the visual stimulus image is a white patch; the stimulus frequency of the third candidate stimulus is 10 Hz, and the visual stimulus image is a blue patch; the stimulus frequency of the fourth candidate stimulus is 11 Hz, and the visual stimulus image is a purple patch. Please refer to [link to relevant documentation]. Figure 7aThis is a schematic diagram of the power spectral density corresponding to a candidate stimulus object provided in an embodiment of this application. The power spectral density diagram corresponding to the first candidate stimulus object can be shown as marked 701, the power spectral density diagram corresponding to the second candidate stimulus object can be shown as marked 702, the power spectral density diagram corresponding to the third candidate stimulus object can be shown as marked 703, and the power spectral density diagram corresponding to the fourth candidate stimulus object can be shown as marked 704. The horizontal axis of the power spectral density diagram represents frequency, and the vertical axis represents the power corresponding to the frequency. As can be seen from the power spectral density diagram corresponding to the second candidate stimulus object shown in marked 702, the maximum power peak value in the power spectral density corresponding to the second candidate stimulus object is very close to the minimum peak value of the other power peak values. That is, the EEG signal (sample EEG signal) induced by the second candidate stimulus object shows a very strong response around 18Hz. If this candidate stimulus object is used as the stimulus object, it will affect the accuracy of intention recognition based on the EEG signal induced by the corresponding stimulus object. Therefore, the second candidate stimulus object will not be identified as the stimulus object.

[0144] Since white visual stimuli typically perform well in EEG systems based on steady-state visual evoked stimuli, EEG signal processing devices can construct multiple new candidate stimuli by adjusting the stimulation frequency of a second candidate stimulus while keeping the visual stimulus image unchanged. For example, seven new candidate stimuli are constructed with stimulation frequencies of 7Hz, 8Hz, 10Hz, 11Hz, 12Hz, 13Hz, and 14Hz. (See [link to relevant documentation]). Figure 7b This is a schematic diagram of the power spectral density corresponding to another candidate stimulus object provided in an embodiment of this application. The power spectral density diagram corresponding to the first new candidate stimulus object can be shown as labeled 711, the power spectral density diagram corresponding to the second new candidate stimulus object can be shown as labeled 712, the power spectral density diagram corresponding to the third new candidate stimulus object can be shown as labeled 713, and the power spectral density diagram corresponding to the fourth new candidate stimulus object can be shown as labeled 714. Please refer to... Figure 7c This is a schematic diagram of the power spectral density corresponding to another candidate stimulus provided in the embodiments of this application. The power spectral density diagram corresponding to the fifth new candidate stimulus can be shown as marked 715, the power spectral density diagram corresponding to the sixth new candidate stimulus can be shown as marked 716, and the power spectral density diagram corresponding to the seventh new candidate stimulus can be shown as marked 717. The horizontal axis of the power spectral density diagram represents the frequency, and the vertical axis represents the power corresponding to the frequency.

[0145] The power spectral density (PSD) plot of the EEG signal evoked by the first new candidate stimulus at 7Hz shows that the maximum power peak should theoretically be around 7Hz. However, in the PSD plot marked 711, the maximum power peak appears around 14Hz, indicating a weak SSVEP response and poor stimulation effect of the first new candidate stimulus at 7Hz. Similarly, the PSD plot of the EEG signal evoked by the second new candidate stimulus at 8Hz shows that the maximum power peak occurs at a position with a large error range from 8Hz, meaning the frequency corresponding to the maximum power peak does not match 8Hz. This also indicates a weak SSVEP response and poor stimulation effect of the second new candidate stimulus at 8Hz. Finally, the PSD plot of the EEG signal evoked by the fourth new candidate stimulus at 11Hz shows the same weak SSVEP response and poor stimulation effect. The power spectral density maps of the EEG signals induced by the new candidate stimuli at 10Hz, 12Hz, 13Hz, and 14Hz all indicate ideal feature extraction results, meaning the power spectral density corresponding to each new candidate stimuli meets the screening criteria. Therefore, a new candidate stimulus can be selected from the new candidate stimuli corresponding to 10Hz, 12Hz, 13Hz, and 14Hz to obtain N stimulus objects. For example, since the stimulation frequency of 10Hz already matches the third candidate stimulus, and the new candidate stimulus corresponding to 14Hz already shows a strong small power peak inducing SSVEP signal near 11Hz before 14Hz, to avoid errors, a new candidate stimulus can be further selected from the new candidate stimuli corresponding to 12Hz and 13Hz, for example, selecting the new candidate stimulus corresponding to 12Hz (i.e., stimulation frequency of 12Hz, visual stimulus image of white block).

[0146] As can be seen, the above shows that the selection of candidate stimuli can be based on power spectral density, or it can be based on normalization results. The relevant processes are similar, but they may be adaptively adjusted when setting various condition thresholds. This application will not elaborate on these details in the embodiments.

[0147] S604 outputs N candidate intent information and N stimuli corresponding to each candidate intent information; where the N stimuli are all different and N is a positive integer.

[0148] S605, acquire the EEG signals generated by the target object during the output of N stimuli.

[0149] S606 performs feature analysis on the EEG signal of the target object and determines the target stimulus object that triggers the generation of the target object's EEG signal from N stimulus objects based on the feature analysis results.

[0150] S607, Generate an intent recognition result based on the candidate intent information corresponding to the target stimulus object; wherein, the intent recognition result is used to trigger the execution of an operation that matches the candidate intent information corresponding to the target stimulus object.

[0151] The relevant processes of steps S604 to S607 have been described in steps S311 to S314 above, and will not be repeated here.

[0152] The EEG signal processing method and the intent recognition system implementing the EEG signal processing method proposed in this application can be applied to multiple fields, such as rehabilitation medicine, smart home control, wheelchair control, gaming, and virtual reality. It can provide convenient human-computer interaction methods for different fields, with a wide range of application scenarios. Furthermore, it boasts high real-time performance and accuracy, enabling rapid and accurate identification of the target's intent and corresponding responses, thus improving the efficiency of human-computer interaction and user experience. It also exhibits high stability, coping with various environmental interferences and being suitable for diverse application scenarios. Moreover, the visual stimulus images displayed based on stimulus frequency (i.e., periodically displayed visual stimulus images) can reduce user visual fatigue and improve user comfort to a certain extent.

[0153] In this embodiment, M candidate stimuli can be output sequentially, and sample EEG signals generated by the sample object during the output of different candidate stimuli can be acquired respectively. Then, by analyzing the power spectral density obtained from the sample EEG signals corresponding to the M candidate stimuli, N stimuli that can be used for intent recognition can be determined from the M candidate stimuli. The power spectral density of the candidate stimuli can be evaluated by calling an evaluation model, and the magnitude of the evaluation value can be used to decide whether the candidate stimuli can be identified as a stimulus. Alternatively, the frequency corresponding to the maximum power peak in the power spectral density of the candidate stimuli can be compared with the stimulation frequency of the candidate stimuli, and the difference between the maximum power peak and the minimum peak value in the power spectral density of the candidate stimuli can be used to determine whether the candidate stimuli can be identified as a stimulus. Stimuli with good stimulation effects can be selected from the M candidate stimuli, thereby making the accuracy of intent recognition based on the EEG signals induced by the determined N stimuli high.

[0154] Based on the description of the above method embodiments, this application also discloses an electroencephalogram (EEG) signal processing device; the EEG signal processing device may be a computer program running in a computer device, the computer device may be the aforementioned EEG signal processing device, and the EEG signal processing device may execute... Figure 3a , Figure 3b or Figure 6 The steps in the illustrated method flow are shown below. Please refer to [link / reference]. Figure 8 This is a schematic diagram of the structure of an electroencephalogram (EEG) signal processing device provided in an embodiment of this application. The EEG signal processing device may include a communication unit 801 and a processing unit 802.

[0155] In one possible implementation, Figure 8 The EEG signal processing device shown can be deployed within an EEG signal processing equipment, which is connected to an EEG signal acquisition device. This EEG signal processing device performs the aforementioned... Figure 3a The method flow shown is as follows:

[0156] Communication unit 801 is used for communication interaction;

[0157] The processing unit 802 is configured to output N stimulation objects in the EEG signal processing device; N is a positive integer; receive the EEG signals generated by the target object during the output of the N stimulation objects obtained by the EEG signal acquisition device; and perform an operation corresponding to the target stimulation object; wherein the target stimulation object is determined from the N stimulation objects based on the target object's EEG signal.

[0158] In one implementation, different stimuli correspond to different candidate intent information; when the processing unit 802 performs an operation corresponding to a target stimulus, it can perform an operation that matches the candidate intent information corresponding to the target stimulus, wherein performing the operation that matches the candidate intent information corresponding to the target stimulus may include:

[0159] Output the candidate intent information corresponding to the target stimulus object;

[0160] Alternatively, an operation control command may be sent to the target device to cause the target device to perform an operation that matches the candidate intent information corresponding to the target stimulus in response to the operation control command.

[0161] In one embodiment, the stimulus objects include visual stimulus images displayed at a stimulus frequency. Different stimulus objects have different stimulus frequencies or image characteristics. The N visual stimulus images included by the N stimulus objects are displayed at different positions in the display area supported by the EEG signal processing device. Alternatively, the N visual stimulus images are displayed sequentially at the same position in the display area supported by the EEG signal processing device.

[0162] In one embodiment, before performing the operation of matching candidate intent information corresponding to the target stimulus, the processing unit 802 is further configured to:

[0163] The EEG signals of the target object are subjected to feature analysis, and the target stimulus object that triggers the generation of the EEG signals of the target object is determined from the N stimulus objects based on the feature analysis results;

[0164] An intent recognition result is generated based on the candidate intent information corresponding to the target stimulus; wherein, the intent recognition result is used to trigger the execution of an operation that matches the candidate intent information corresponding to the target stimulus.

[0165] In one embodiment, the processing unit 802 is further configured to:

[0166] An intent recognition trigger interface is displayed in the EEG signal processing device; the intent recognition trigger interface includes a first trigger control and a second trigger control.

[0167] When the first trigger control is triggered, the step of outputting N stimulation objects in the EEG signal processing device is performed.

[0168] When the second trigger control is triggered, a data acquisition interface is displayed; the data acquisition interface is used to trigger the acquisition of the pre-stored EEG signals of the target object from the storage space.

[0169] In one possible implementation, Figure 8 The brain signal processing device shown performs the above-mentioned functions. Figure 3b as well as Figure 6 The method flow shown is as follows:

[0170] Communication unit 801 is used for communication interaction;

[0171] Processing unit 802 is configured to output N candidate intent information and N stimulus objects corresponding to the N candidate intent information; wherein the N stimulus objects are all different and N is a positive integer; acquire the electroencephalogram (EEG) signal generated by the target object during the output of the N stimulus objects; perform feature analysis on the EEG signal of the target object, and determine the target stimulus object that triggers the generation of the target object's EEG signal from the N stimulus objects based on the feature analysis results; generate an intent recognition result based on the candidate intent information corresponding to the target stimulus object; wherein the intent recognition result is used to trigger the execution of an operation that matches the candidate intent information corresponding to the target stimulus object.

[0172] In one embodiment, when the processing unit 802 performs feature analysis on the EEG signal of the target object and determines the target stimulus object that triggers the generation of the EEG signal of the target object from the N stimulus objects based on the feature analysis results, it may specifically be used to:

[0173] The EEG signal of the target object is preprocessed to obtain an intermediate EEG signal; the intermediate EEG signal is the signal in the target frequency band of the target object's EEG signal.

[0174] The intermediate EEG signal is subjected to feature extraction processing to obtain EEG signal features; the feature analysis results include the EEG signal features;

[0175] Based on the EEG signal characteristics and the reference EEG signal characteristics associated with each of the N stimulus objects, the target stimulus object that triggers the generation of the EEG signal of the target object is determined from the N stimulus objects.

[0176] In one embodiment, the stimulus object includes a visual stimulus image displayed at a stimulus frequency. Different stimulus objects have different stimulus frequencies or image features. The EEG signal features include: EEG feature values ​​of the intermediate EEG signal at multiple frequencies. The EEG feature value of the intermediate EEG signal at any frequency includes: the power of the intermediate EEG signal at the corresponding frequency, or the normalization result of the intermediate EEG signal at the corresponding frequency.

[0177] When processing unit 802 determines the target stimulus that triggers the generation of the target object's EEG signal from the N stimulus objects based on the EEG signal features and the reference EEG signal features associated with each of the N stimulus objects, it can specifically be used to:

[0178] Determine the feature similarity between the EEG signal features and the reference EEG signal features associated with each of the N stimulus objects;

[0179] The stimulus object indicated by the maximum feature similarity is determined as the target stimulus object that triggers the generation of the EEG signal of the target object.

[0180] In one embodiment, the stimulus object includes a visual stimulus image displayed at a stimulus frequency. Different stimulus objects have different stimulus frequencies or image features. The EEG signal features include: EEG feature values ​​of the intermediate EEG signal at multiple frequencies. The EEG feature value of the intermediate EEG signal at any frequency includes: the power of the intermediate EEG signal at the corresponding frequency, or the normalization result of the intermediate EEG signal at the corresponding frequency.

[0181] When processing unit 802 determines the target stimulus that triggers the generation of the target object's EEG signal from the N stimulus objects based on the EEG signal features and the reference EEG signal features associated with each of the N stimulus objects, it can specifically be used to:

[0182] The EEG feature value at the i-th stimulation frequency is determined from the EEG signal features of the target object, and the reference EEG feature value at the i-th stimulation frequency is determined based on the reference EEG signal features associated with the i-th stimulation object; wherein, the i-th stimulation frequency is the stimulation frequency of the i-th stimulation object among the N stimulation objects, i∈[1,N];

[0183] Determine the feature difference between the EEG feature value corresponding to the i-th stimulation frequency and the reference EEG feature value to obtain the feature difference corresponding to the i-th stimulation frequency, until the feature difference corresponding to each of the N stimulation frequencies is obtained;

[0184] The stimulus object with the stimulation frequency corresponding to the smallest feature difference is determined as the target stimulus object that triggers the generation of the EEG signal of the target object.

[0185] In one embodiment, the EEG signal features include: EEG feature values ​​of the intermediate EEG signal at multiple frequencies, wherein the EEG feature value of the intermediate EEG signal at any frequency includes: the normalized result of the intermediate EEG signal at the corresponding frequency; the processing unit 802, when performing feature extraction processing on the intermediate EEG signal to obtain EEG signal features, may specifically be used for:

[0186] The intermediate EEG signal is subjected to frequency domain transformation to obtain multiple frequency domain signals; wherein, each frequency domain signal corresponds to a frequency.

[0187] The amplitudes of the multiple frequency domain signals are normalized respectively to obtain the normalized results of the intermediate EEG signal at the corresponding frequencies of each frequency domain signal.

[0188] Based on the frequencies corresponding to each frequency domain signal and the normalization results of the intermediate EEG signal at the corresponding frequencies, EEG signal features are constructed.

[0189] In one embodiment, the stimulus object includes visual stimulus images displayed at a stimulus frequency. Different stimulus objects have different stimulus frequencies or image features. When the processing unit 802 performs feature analysis on the EEG signal of the target object and determines the target stimulus object that triggers the generation of the EEG signal of the target object from the N stimulus objects based on the feature analysis results, it can be specifically used for:

[0190] Determine the reference matrix corresponding to the i-th stimulation frequency among N stimulation frequencies; where i∈[1,N], and the reference matrix corresponding to the i-th stimulation frequency is obtained based on the reference EEG signal corresponding to the i-th stimulation frequency;

[0191] A canonical correlation analysis is performed on the EEG signal of the target object and the reference matrix corresponding to the i-th stimulation frequency to obtain the correlation coefficient corresponding to the i-th stimulation frequency; wherein, the correlation coefficient corresponding to the i-th stimulation frequency is used to indicate the degree of matching between the EEG signal of the target object and the reference matrix corresponding to the i-th stimulation frequency;

[0192] The correlation coefficients corresponding to the N stimulation frequencies are used as feature analysis results, and the target stimulation object that triggers the generation of the EEG signal of the target object is determined from the N stimulation objects based on the feature analysis results.

[0193] In one embodiment, the processing unit 802 is further configured to:

[0194] M candidate stimuli are output sequentially, and the sample EEG signals generated by the sample object during the output of different candidate stimuli are obtained respectively; wherein, the M candidate stimuli are all different, and M is a positive integer greater than or equal to N;

[0195] The sample EEG signals corresponding to the M candidate stimulus objects are processed respectively to obtain the power spectral density corresponding to each of the M candidate stimulus objects;

[0196] Based on the power spectral density corresponding to each of the M candidate stimuli, N stimuli are determined from the M candidate stimuli.

[0197] The N identified stimulus objects refer to the candidate stimulus objects whose power spectral density meets the screening criteria among the M candidate stimulus objects.

[0198] In one embodiment, when the processing unit 802 determines N stimulus objects from the M candidate stimulus objects based on the power spectral density corresponding to each of the M candidate stimulus objects, it may specifically be used to:

[0199] The evaluation model is invoked to evaluate the power spectral density corresponding to each of the M candidate stimuli, thereby obtaining the evaluation value corresponding to each of the M candidate stimuli.

[0200] From the M candidate stimuli, N candidate stimuli with evaluation values ​​greater than the reference evaluation value are selected as candidate stimuli whose power spectral density meets the selection criteria, thus obtaining N stimuli.

[0201] In one embodiment, the power spectral density corresponding to the target candidate stimulus includes: the power of the EEG signal generated by the target candidate stimulus at multiple frequencies, wherein the target candidate stimulus is any one of the M candidate stimulus objects;

[0202] The power spectral density corresponding to the target candidate stimulus meets the screening conditions as follows: the frequency corresponding to the maximum power peak in the power spectral density of the target candidate stimulus matches the stimulation frequency of the target candidate stimulus, and the minimum peak difference between the maximum power peak and other power peaks in the power spectral density of the target candidate stimulus is greater than a target difference threshold; wherein, the target difference threshold is determined based on the maximum power peak.

[0203] According to another embodiment of this application, Figure 8 The various units in the illustrated EEG signal processing device can be individually or entirely combined into one or more other units, or some of the units can be further divided into multiple functionally smaller units. This achieves the same operation without affecting the technical effects of the embodiments of this application. The above-mentioned units are based on logical function division. In practical applications, the function of one unit can also be implemented by multiple units, or the function of multiple units can be implemented by one unit. In other embodiments of this application, the EEG signal processing device may also include other units. In practical applications, these functions can also be implemented with the assistance of other units, and can be implemented collaboratively by multiple units.

[0204] According to another embodiment of this application, the following can be achieved by running on a general-purpose computing device, such as a computer, which includes processing elements and storage elements such as a central processing unit (CPU), random access memory (RAM), and read-only memory (ROM), a device capable of performing operations such as... Figure 3a , Figure 3b or Figure 6The computer program for each step involved in the corresponding method shown is used to construct, as... Figure 8 The diagram illustrates an electroencephalogram (EEG) signal processing apparatus and an EEG signal processing method for implementing embodiments of this application. A computer program may be recorded on, for example, a computer-readable storage medium, loaded onto the aforementioned computing device via the computer-readable storage medium, and executed therein.

[0205] In this application embodiment, the terms "module" or "unit" refer to a computer program or part of a computer program that has a predetermined function and works with other related parts to achieve a predetermined goal, and can be implemented wholly or partially using software, hardware (such as processing circuitry or memory), or a combination thereof. Similarly, a processor (or multiple processors or memory) can be used to implement one or more modules or units. Furthermore, each module or unit can be part of an overall module or unit that includes the functionality of that module or unit.

[0206] The specific implementation of each unit of the device described in this embodiment and the beneficial effects that can be achieved can be referred to the description of the relevant content in the foregoing embodiments, and will not be repeated here.

[0207] Based on the descriptions of the above method and device embodiments, this application also provides a computer device, which can be the aforementioned electroencephalogram (EEG) signal processing device. Please refer to... Figure 9 The computer device includes at least a processor 901, an input interface 902, an output interface 903, and a computer-readable storage medium 904. The processor 901, input interface 902, output interface 903, and computer-readable storage medium 904 within the computer device can be connected via a bus or other means. The computer-readable storage medium 904 can be stored in the computer device's memory and is used to store computer programs. The processor 901 is used to execute the computer programs stored in the computer-readable storage medium 904. The processor 901 (or CPU (Central Processing Unit)) is the computing and control core of the computer device, suitable for running computer programs to implement corresponding methods or functions.

[0208] In one embodiment, the processor 901 proposed in this application can be used to perform a process related to intent recognition, specifically including: outputting N stimulus objects in the EEG signal processing device; N being a positive integer; receiving the EEG signals generated by the target object during the output of the N stimulus objects, acquired by the EEG signal acquisition device; performing an operation corresponding to the target stimulus object; wherein the target stimulus object is determined from the N stimulus objects based on the target object's EEG signals, etc. Alternatively, outputting N candidate intent information and N stimulus objects corresponding to the N candidate intent information; wherein the N stimulus objects are all different, and N is a positive integer; acquiring the EEG signals generated by the target object during the output of the N stimulus objects; performing feature analysis on the target object's EEG signals, and determining the target stimulus object that triggers the generation of the target object's EEG signals from the N stimulus objects based on the feature analysis results; generating an intent recognition result based on the candidate intent information corresponding to the target stimulus object; wherein the intent recognition result is used to trigger the execution of an operation matching the candidate intent information corresponding to the target stimulus object, etc.

[0209] This application also provides a computer-readable storage medium (Memory), which is a memory device in a computer device used to store computer programs and data. It is understood that the computer-readable storage medium here can include both the built-in storage medium in the computer device and extended storage media supported by the computer device. The computer-readable storage medium provides storage space that stores the operating system of the computer device. Furthermore, the storage space also stores a computer program adapted to be loaded and executed by the processor 901 to implement the corresponding method flow provided in this application embodiment. It should be noted that the computer-readable storage medium here can be a high-speed RAM memory or a non-volatile memory, such as at least one disk storage device; optionally, it can also be at least one computer-readable storage medium located remotely from the aforementioned processor.

[0210] In one embodiment, a processor may load and execute a computer program stored in a computer-readable storage medium to perform the aforementioned tasks. Figure 3a , Figure 3b or Figure 6 The corresponding steps in the method embodiments shown.

[0211] In practice, the EEG signal processing device is connected to the EEG signal acquisition device, and the computer program in the computer-readable storage medium is loaded and executed by the processor. Figure 3a In the corresponding steps of the method embodiment shown, the following can be performed:

[0212] The EEG signal processing device outputs N stimulation targets; N is a positive integer.

[0213] Receive the electroencephalogram (EEG) signals generated by the target object during the output process of the N stimulation objects, acquired by the EEG signal acquisition device;

[0214] Perform an operation corresponding to a target stimulus; wherein the target stimulus is determined from the N stimulus objects based on the EEG signal of the target object.

[0215] In one implementation, different stimuli correspond to different candidate intent information; when the processor 901 performs an operation corresponding to a target stimulus, it can perform an operation that matches the candidate intent information corresponding to the target stimulus, wherein performing the operation that matches the candidate intent information corresponding to the target stimulus may include:

[0216] Output the candidate intent information corresponding to the target stimulus object;

[0217] Alternatively, an operation control command may be sent to the target device to cause the target device to perform an operation that matches the candidate intent information corresponding to the target stimulus in response to the operation control command.

[0218] In one embodiment, the stimulus objects include visual stimulus images displayed at a stimulus frequency. Different stimulus objects have different stimulus frequencies or image characteristics. The N visual stimulus images included by the N stimulus objects are displayed at different positions in the display area supported by the EEG signal processing device. Alternatively, the N visual stimulus images are displayed sequentially at the same position in the display area supported by the EEG signal processing device.

[0219] In one implementation, before performing the operation of matching candidate intent information corresponding to the target stimulus, the processor 901 is further configured to:

[0220] The EEG signals of the target object are subjected to feature analysis, and the target stimulus object that triggers the generation of the EEG signals of the target object is determined from the N stimulus objects based on the feature analysis results;

[0221] An intent recognition result is generated based on the candidate intent information corresponding to the target stimulus; wherein, the intent recognition result is used to trigger the execution of an operation that matches the candidate intent information corresponding to the target stimulus.

[0222] In one implementation, the processor 901 is further configured to:

[0223] An intent recognition trigger interface is displayed in the EEG signal processing device; the intent recognition trigger interface includes a first trigger control and a second trigger control.

[0224] When the first trigger control is triggered, the step of outputting N stimulation objects in the EEG signal processing device is performed.

[0225] When the second trigger control is triggered, a data acquisition interface is displayed; the data acquisition interface is used to trigger the acquisition of the pre-stored EEG signals of the target object from the storage space.

[0226] In practice, the computer program in the computer-readable storage medium is loaded and executed by the processor. Figure 3b or Figure 6 In the corresponding steps of the method embodiment shown, the following can be performed:

[0227] Output N candidate intent information and N stimulus objects corresponding to each of the N candidate intent information; wherein, the N stimulus objects are all different and N is a positive integer;

[0228] Acquire the electroencephalogram (EEG) signals generated by the target object during the output of the N stimuli;

[0229] The EEG signals of the target object are subjected to feature analysis, and the target stimulus object that triggers the generation of the EEG signals of the target object is determined from the N stimulus objects based on the feature analysis results;

[0230] An intent recognition result is generated based on the candidate intent information corresponding to the target stimulus; wherein, the intent recognition result is used to trigger the execution of an operation that matches the candidate intent information corresponding to the target stimulus.

[0231] In one embodiment, when the processor 901 performs feature analysis on the EEG signal of the target object and determines the target stimulus object that triggers the generation of the EEG signal of the target object from the N stimulus objects based on the feature analysis results, it may specifically be used to:

[0232] The EEG signal of the target object is preprocessed to obtain an intermediate EEG signal; the intermediate EEG signal is the signal in the target frequency band of the target object's EEG signal.

[0233] The intermediate EEG signal is subjected to feature extraction processing to obtain EEG signal features; the feature analysis results include the EEG signal features;

[0234] Based on the EEG signal characteristics and the reference EEG signal characteristics associated with each of the N stimulus objects, the target stimulus object that triggers the generation of the EEG signal of the target object is determined from the N stimulus objects.

[0235] In one embodiment, the stimulus object includes a visual stimulus image displayed at a stimulus frequency. Different stimulus objects have different stimulus frequencies or image features. The EEG signal features include: EEG feature values ​​of the intermediate EEG signal at multiple frequencies. The EEG feature value of the intermediate EEG signal at any frequency includes: the power of the intermediate EEG signal at the corresponding frequency, or the normalization result of the intermediate EEG signal at the corresponding frequency.

[0236] When processor 901 determines the target stimulus that triggers the generation of the target object's EEG signal from the N stimulus objects based on the EEG signal features and the reference EEG signal features associated with each of the N stimulus objects, it can specifically be used to:

[0237] Determine the feature similarity between the EEG signal features and the reference EEG signal features associated with each of the N stimulus objects;

[0238] The stimulus object indicated by the maximum feature similarity is determined as the target stimulus object that triggers the generation of the EEG signal of the target object.

[0239] In one embodiment, the stimulus object includes a visual stimulus image displayed at a stimulus frequency. Different stimulus objects have different stimulus frequencies or image features. The EEG signal features include: EEG feature values ​​of the intermediate EEG signal at multiple frequencies. The EEG feature value of the intermediate EEG signal at any frequency includes: the power of the intermediate EEG signal at the corresponding frequency, or the normalization result of the intermediate EEG signal at the corresponding frequency.

[0240] When processor 901 determines the target stimulus that triggers the generation of the target object's EEG signal from the N stimulus objects based on the EEG signal features and the reference EEG signal features associated with each of the N stimulus objects, it can specifically be used to:

[0241] The EEG feature value at the i-th stimulation frequency is determined from the EEG signal features of the target object, and the reference EEG feature value at the i-th stimulation frequency is determined based on the reference EEG signal features associated with the i-th stimulation object; wherein, the i-th stimulation frequency is the stimulation frequency of the i-th stimulation object among the N stimulation objects, i∈[1,N];

[0242] Determine the feature difference between the EEG feature value corresponding to the i-th stimulation frequency and the reference EEG feature value to obtain the feature difference corresponding to the i-th stimulation frequency, until the feature difference corresponding to each of the N stimulation frequencies is obtained;

[0243] The stimulus object with the stimulation frequency corresponding to the smallest feature difference is determined as the target stimulus object that triggers the generation of the EEG signal of the target object.

[0244] In one embodiment, the EEG signal features include: EEG feature values ​​of the intermediate EEG signal at multiple frequencies, wherein the EEG feature value of the intermediate EEG signal at any frequency includes: the normalized result of the intermediate EEG signal at the corresponding frequency; when the processor 901 performs feature extraction processing on the intermediate EEG signal to obtain EEG signal features, it can specifically be used for:

[0245] The intermediate EEG signal is subjected to frequency domain transformation to obtain multiple frequency domain signals; wherein, each frequency domain signal corresponds to a frequency.

[0246] The amplitudes of the multiple frequency domain signals are normalized respectively to obtain the normalized results of the intermediate EEG signal at the corresponding frequencies of each frequency domain signal.

[0247] Based on the frequencies corresponding to each frequency domain signal and the normalization results of the intermediate EEG signal at the corresponding frequencies, EEG signal features are constructed.

[0248] In one embodiment, the stimulus object includes visual stimulus images displayed at a stimulus frequency. Different stimulus objects have different stimulus frequencies or image features. When the processor 901 performs feature analysis on the EEG signal of the target object and determines the target stimulus object that triggers the generation of the EEG signal of the target object from the N stimulus objects based on the feature analysis results, it may specifically be used to:

[0249] Determine the reference matrix corresponding to the i-th stimulation frequency among N stimulation frequencies; where i∈[1,N], and the reference matrix corresponding to the i-th stimulation frequency is obtained based on the reference EEG signal corresponding to the i-th stimulation frequency;

[0250] A canonical correlation analysis is performed on the EEG signal of the target object and the reference matrix corresponding to the i-th stimulation frequency to obtain the correlation coefficient corresponding to the i-th stimulation frequency; wherein, the correlation coefficient corresponding to the i-th stimulation frequency is used to indicate the degree of matching between the EEG signal of the target object and the reference matrix corresponding to the i-th stimulation frequency;

[0251] The correlation coefficients corresponding to the N stimulation frequencies are used as feature analysis results, and the target stimulation object that triggers the generation of the EEG signal of the target object is determined from the N stimulation objects based on the feature analysis results.

[0252] In one implementation, the processor 901 is further configured to:

[0253] M candidate stimuli are output sequentially, and the sample EEG signals generated by the sample object during the output of different candidate stimuli are obtained respectively; wherein, the M candidate stimuli are all different, and M is a positive integer greater than or equal to N;

[0254] The sample EEG signals corresponding to the M candidate stimulus objects are processed respectively to obtain the power spectral density corresponding to each of the M candidate stimulus objects;

[0255] Based on the power spectral density corresponding to each of the M candidate stimuli, N stimuli are determined from the M candidate stimuli.

[0256] The N identified stimulus objects refer to the candidate stimulus objects whose power spectral density meets the screening criteria among the M candidate stimulus objects.

[0257] In one embodiment, when the processor 901 determines N stimulus objects from the M candidate stimulus objects based on the power spectral density corresponding to each of the M candidate stimulus objects, it may specifically be used to:

[0258] The evaluation model is invoked to evaluate the power spectral density corresponding to each of the M candidate stimuli, thereby obtaining the evaluation value corresponding to each of the M candidate stimuli.

[0259] From the M candidate stimuli, N candidate stimuli with evaluation values ​​greater than the reference evaluation value are selected as candidate stimuli whose power spectral density meets the selection criteria, thus obtaining N stimuli.

[0260] In one embodiment, the power spectral density corresponding to the target candidate stimulus includes: the power of the EEG signal generated by the target candidate stimulus at multiple frequencies, wherein the target candidate stimulus is any one of the M candidate stimulus objects;

[0261] The power spectral density corresponding to the target candidate stimulus meets the screening conditions as follows: the frequency corresponding to the maximum power peak in the power spectral density of the target candidate stimulus matches the stimulation frequency of the target candidate stimulus, and the minimum peak difference between the maximum power peak and other power peaks in the power spectral density of the target candidate stimulus is greater than a target difference threshold; wherein, the target difference threshold is determined based on the maximum power peak.

[0262] In the computer device of this embodiment, the specific implementation of each operation and step performed by the processor and the beneficial effects that can be achieved can be referred to the description of the relevant content in the foregoing embodiments, and will not be repeated here.

[0263] This application provides a computer program product, which includes a computer program stored in a computer-readable storage medium. A processor of a computer device reads the computer program from the computer-readable storage medium and executes the computer program, causing the computer device to perform the aforementioned actions. Figure 3a , Figure 3b or Figure 6 The method embodiment shown.

[0264] It should be understood that the above-disclosed embodiments are merely preferred embodiments of this application and should not be construed as limiting the scope of this application. Therefore, any equivalent variations made in accordance with the claims of this application are still within the scope of this application.

Claims

1. A method for processing electroencephalogram (EEG) signals, characterized in that, The method is applied to an electroencephalogram (EEG) signal processing device, which is connected to an EEG signal acquisition device, and the method includes: The EEG signal processing device outputs N stimulation targets; N is a positive integer. Receive the electroencephalogram (EEG) signals generated by the target object during the output process of the N stimulation objects, acquired by the EEG signal acquisition device; Perform an operation corresponding to a target stimulus; wherein the target stimulus is determined from the N stimulus objects based on the target object's electroencephalogram (EEG) signal.

2. The method as described in claim 1, characterized in that, Different stimuli correspond to different candidate intent information; the execution of the operation corresponding to the target stimulus includes executing the operation that matches the candidate intent information corresponding to the target stimulus, wherein the execution of the operation that matches the candidate intent information corresponding to the target stimulus includes: Output the candidate intent information corresponding to the target stimulus object; Alternatively, an operation control command may be sent to the target device to cause the target device to perform an operation that matches the candidate intent information corresponding to the target stimulus in response to the operation control command.

3. The method as described in claim 1, characterized in that, The stimulation objects include visual stimulation images displayed at stimulation frequencies. Different stimulation objects have different stimulation frequencies or image characteristics. The N visual stimulation images included by the N stimulation objects are displayed at different positions in the display area supported by the EEG signal processing device; or, the N visual stimulation images are displayed sequentially at the same position in the display area supported by the EEG signal processing device.

4. The method as described in claim 2, characterized in that, Before performing the operation of matching candidate intent information corresponding to the target stimulus, the method further includes: The EEG signals of the target object are subjected to feature analysis, and the target stimulus object that triggers the generation of the EEG signals of the target object is determined from the N stimulus objects based on the feature analysis results; An intent recognition result is generated based on the candidate intent information corresponding to the target stimulus; wherein, the intent recognition result is used to trigger the execution of an operation that matches the candidate intent information corresponding to the target stimulus.

5. The method as described in claim 1, characterized in that, The method further includes: An intent recognition trigger interface is displayed in the EEG signal processing device; the intent recognition trigger interface includes a first trigger control and a second trigger control. When the first trigger control is triggered, the step of outputting N stimulation objects in the EEG signal processing device is performed. When the second trigger control is triggered, a data acquisition interface is displayed; the data acquisition interface is used to trigger the acquisition of the pre-stored EEG signals of the target object from the storage space.

6. A method for processing electroencephalogram (EEG) signals, characterized in that, include: Output N candidate intent information and N stimulus objects corresponding to each of the N candidate intent information; wherein, the N stimulus objects are all different and N is a positive integer; Acquire the electroencephalogram (EEG) signals generated by the target object during the output of the N stimuli; The EEG signals of the target object are subjected to feature analysis, and the target stimulus object that triggers the generation of the EEG signals of the target object is determined from the N stimulus objects based on the feature analysis results; An intent recognition result is generated based on the candidate intent information corresponding to the target stimulus; wherein, the intent recognition result is used to trigger the execution of an operation that matches the candidate intent information corresponding to the target stimulus.

7. The method as described in claim 6, characterized in that, The step of performing feature analysis on the EEG signal of the target object, and determining the target stimulus object that triggers the generation of the EEG signal of the target object from the N stimulus objects based on the feature analysis results, includes: The EEG signal of the target object is preprocessed to obtain an intermediate EEG signal; the intermediate EEG signal is the signal in the target frequency band of the target object's EEG signal. The intermediate EEG signal is subjected to feature extraction processing to obtain EEG signal features; the feature analysis results include the EEG signal features; Based on the EEG signal characteristics and the reference EEG signal characteristics associated with each of the N stimulus objects, the target stimulus object that triggers the generation of the EEG signal of the target object is determined from the N stimulus objects.

8. The method as described in claim 7, characterized in that, The stimulus objects include visual stimulus images displayed at stimulus frequencies. Different stimulus objects have different stimulus frequencies or image characteristics. The EEG signal characteristics include: EEG feature values ​​of the intermediate EEG signal at multiple frequencies. The EEG feature value of the intermediate EEG signal at any frequency includes: the power of the intermediate EEG signal at the corresponding frequency, or the normalization result of the intermediate EEG signal at the corresponding frequency. The step of determining the target stimulus object that triggers the generation of the target object's EEG signal from the N stimulus objects based on the EEG signal features and the reference EEG signal features associated with each of the N stimulus objects includes: Determine the feature similarity between the EEG signal features and the reference EEG signal features associated with each of the N stimulus objects; The stimulus object indicated by the maximum feature similarity is determined as the target stimulus object that triggers the generation of the EEG signal of the target object.

9. The method as described in claim 7, characterized in that, The stimulus objects include visual stimulus images displayed at stimulus frequencies. Different stimulus objects have different stimulus frequencies or image characteristics. The EEG signal characteristics include: EEG feature values ​​of the intermediate EEG signal at multiple frequencies. The EEG feature value of the intermediate EEG signal at any frequency includes: the power of the intermediate EEG signal at the corresponding frequency, or the normalization result of the intermediate EEG signal at the corresponding frequency. The step of determining the target stimulus object that triggers the generation of the target object's EEG signal from the N stimulus objects based on the EEG signal features and the reference EEG signal features associated with each of the N stimulus objects includes: The EEG feature value at the i-th stimulation frequency is determined from the EEG signal features of the target object, and the reference EEG feature value at the i-th stimulation frequency is determined based on the reference EEG signal features associated with the i-th stimulation object; wherein, the i-th stimulation frequency is the stimulation frequency of the i-th stimulation object among the N stimulation objects, i∈[1,N]; Determine the feature difference between the EEG feature value corresponding to the i-th stimulation frequency and the reference EEG feature value to obtain the feature difference corresponding to the i-th stimulation frequency, until the feature difference corresponding to each of the N stimulation frequencies is obtained; The stimulus object with the stimulation frequency corresponding to the smallest feature difference is determined as the target stimulus object that triggers the generation of the EEG signal of the target object.

10. The method as described in claim 7, characterized in that, The EEG signal features include: EEG feature values ​​of the intermediate EEG signal at multiple frequencies, wherein the EEG feature value of the intermediate EEG signal at any frequency includes: the normalized result of the intermediate EEG signal at the corresponding frequency; the feature extraction processing of the intermediate EEG signal to obtain EEG signal features includes: The intermediate EEG signal is subjected to frequency domain transformation to obtain multiple frequency domain signals; wherein, each frequency domain signal corresponds to a frequency. The amplitudes of the multiple frequency domain signals are normalized respectively to obtain the normalized results of the intermediate EEG signal at the corresponding frequencies of each frequency domain signal. Based on the frequencies corresponding to each frequency domain signal and the normalization results of the intermediate EEG signal at the corresponding frequencies, EEG signal features are constructed.

11. The method as described in claim 6, characterized in that, The stimulus objects include visual stimulus images displayed at stimulus frequencies. Different stimulus objects have different stimulus frequencies or image features. The step of performing feature analysis on the EEG signals of the target object and determining the target stimulus object that triggers the generation of the target object's EEG signals from the N stimulus objects based on the feature analysis results includes: Determine the reference matrix corresponding to the i-th stimulation frequency among N stimulation frequencies; where i∈[1,N], and the reference matrix corresponding to the i-th stimulation frequency is obtained based on the reference EEG signal corresponding to the i-th stimulation frequency; A canonical correlation analysis is performed on the EEG signal of the target object and the reference matrix corresponding to the i-th stimulation frequency to obtain the correlation coefficient corresponding to the i-th stimulation frequency; wherein, the correlation coefficient corresponding to the i-th stimulation frequency is used to indicate the degree of matching between the EEG signal of the target object and the reference matrix corresponding to the i-th stimulation frequency; The correlation coefficients corresponding to the N stimulation frequencies are used as feature analysis results, and the target stimulation object that triggers the generation of the EEG signal of the target object is determined from the N stimulation objects based on the feature analysis results.

12. The method as described in claim 6, characterized in that, The method further includes: M candidate stimuli are output sequentially, and the sample EEG signals generated by the sample object during the output of different candidate stimuli are obtained respectively; wherein, the M candidate stimuli are all different, and M is a positive integer greater than or equal to N; The sample EEG signals corresponding to the M candidate stimulus objects are processed respectively to obtain the power spectral density corresponding to each of the M candidate stimulus objects; Based on the power spectral density corresponding to each of the M candidate stimuli, N stimuli are determined from the M candidate stimuli. The N identified stimulus objects refer to the candidate stimulus objects whose power spectral density meets the screening criteria among the M candidate stimulus objects.

13. The method as described in claim 12, characterized in that, The step of determining N stimuli from the M candidate stimuli based on their respective power spectral densities includes: The evaluation model is invoked to evaluate the power spectral density corresponding to each of the M candidate stimuli, thereby obtaining the evaluation value corresponding to each of the M candidate stimuli. From the M candidate stimuli, N candidate stimuli with evaluation values ​​greater than the reference evaluation value are selected as candidate stimuli whose power spectral density meets the selection criteria, thus obtaining N stimuli.

14. The method as described in claim 12, characterized in that, The power spectral density corresponding to the target candidate stimulus includes: the power of the EEG signal generated by the target candidate stimulus at multiple frequencies, wherein the target candidate stimulus is any one of the M candidate stimulus objects; The power spectral density corresponding to the target candidate stimulus meets the screening conditions as follows: the frequency corresponding to the maximum power peak in the power spectral density of the target candidate stimulus matches the stimulation frequency of the target candidate stimulus, and the minimum peak difference between the maximum power peak and other power peaks in the power spectral density of the target candidate stimulus is greater than a target difference threshold; wherein, the target difference threshold is determined based on the maximum power peak.

15. A brainwave signal processing device, characterized in that, The device is deployed in an electroencephalogram (EEG) signal processing device, which is connected to an EEG signal acquisition device. The device includes: A communication unit, used for communication interaction; The processing unit is configured to output N stimulation objects in the EEG signal processing device; N is a positive integer; receive the EEG signals generated by the target object during the output of the N stimulation objects by the EEG signal acquisition device; and perform an operation corresponding to the target stimulation object; wherein the target stimulation object is determined from the N stimulation objects based on the target object's EEG signals.

16. A brainwave signal processing device, characterized in that, include: A communication unit, used for communication interaction; The processing unit is configured to output N candidate intent information and N stimulus objects corresponding to the N candidate intent information; wherein the N stimulus objects are all different and N is a positive integer; acquire the electroencephalogram (EEG) signal generated by the target object during the output of the N stimulus objects; perform feature analysis on the EEG signal of the target object, and determine the target stimulus object that triggers the generation of the target object's EEG signal from the N stimulus objects based on the feature analysis results; generate an intent recognition result based on the candidate intent information corresponding to the target stimulus object; wherein the intent recognition result is used to trigger the execution of an operation that matches the candidate intent information corresponding to the target stimulus object.

17. A computer device, comprising an input interface and an output interface, characterized in that, Also includes: Processor and computer-readable storage medium; The computer-readable storage medium is used to store computer programs; The processor is configured to run the computer program to implement the EEG signal processing method as described in any one of claims 1-5, or to implement the EEG signal processing method as described in any one of claims 6-14.

18. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program adapted to be loaded by a processor and executed by the electroencephalogram (EEG) signal processing method as described in any one of claims 1-5, or the computer program adapted to be loaded by a processor and executed by the EEG signal processing method as described in any one of claims 6-14.

19. A computer program product, characterized in that, The computer program product includes a computer program adapted to be loaded by a processor and executed by the electroencephalogram (EEG) signal processing method as described in any one of claims 1-5, or the computer program adapted to be loaded by a processor and executed by the EEG signal processing method as described in any one of claims 6-14.

Images