Signal processing method and device, equipment, storage medium and program product

Abstract

The invention discloses a signal processing method and device, equipment, a storage medium and a program product, and the method comprises the steps: obtaining N groups of electroencephalogram signals collected by a detection object in a brain region corresponding to a target electroencephalogram obtaining rule according to the target electroencephalogram obtaining rule; n is a positive integer; preprocessing each group of electroencephalogram signals to obtain a plurality of source signals of each group of electroencephalogram signals in the N groups of electroencephalogram signals; based on the signal features of the plurality of source signals, respectively determining a target source signal associated with the attention detection task from the plurality of source signals of each group of electroencephalogram signals; according to the target source signal corresponding to each group of electroencephalogram signals, executing an attention detection operation corresponding to the attention detection task to obtain a task processing result of the attention detection task; the task processing result comprises an attention concentration level; the method can be applied to the traffic field and can improve the accuracy of attention detection.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of computer, in particular to a signal processing method, a signal processing device, a computer device, a computer readable storage medium and a computer program product. BACKGROUND

[0002] The attention detection task aims to detect the attention concentration level of the object, and the attention concentration level can be used to indicate the attention concentration degree. In the implementation manner of the existing attention detection task, the attention concentration level of the object is usually determined by analyzing the eye movement, facial expression or behavior of the object. This manner usually needs to collect the facial image or body image of the object, and then the eye movement, facial expression or behavior of the object can be analyzed based on the facial image or body image to further determine the attention concentration level of the object. The collection of the image content such as facial image and body image is greatly affected by environmental light and other factors. For example, the quality of the image collected in a dimly lit environment is poor, which reduces the prediction accuracy of the attention concentration level, that is, the accuracy of the attention detection is reduced. SUMMARY

[0003] The embodiments of the present application provide a signal processing method, device, equipment, storage medium and program product, which can improve the accuracy of attention detection.

[0004] In one aspect, the embodiments of the present application provide a signal processing method, which comprises:

[0005] According to the target electroencephalogram acquisition rule, N groups of electroencephalogram signals collected by the detection object in the brain region corresponding to the target electroencephalogram acquisition rule are obtained; N is a positive integer;

[0006] Each group of electroencephalogram signals is preprocessed respectively to obtain a plurality of source signals of each group of electroencephalogram signals in the N groups of electroencephalogram signals;

[0007] Based on the signal features of the plurality of source signals, target source signals associated with the attention detection task are determined from the plurality of source signals of each group of electroencephalogram signals respectively;

[0008] According to the target source signals corresponding to each group of electroencephalogram signals, an attention detection operation corresponding to the attention detection task is performed to obtain a task processing result of the attention detection task; the task processing result includes an attention concentration level.

[0009] In one aspect, the embodiments of the present application provide a signal processing device, which comprises:

[0010] The communication unit is configured to perform communication interaction.

[0011] The processing unit is configured to acquire N groups of electroencephalogram signals collected by the detection object in a brain region corresponding to a target electroencephalogram acquisition rule according to the target electroencephalogram acquisition rule; N is a positive integer; each group of electroencephalogram signals is preprocessed to obtain a plurality of source signals of each group of electroencephalogram signals in the N groups of electroencephalogram signals; target source signals associated with an attention detection task are determined from the plurality of source signals of each group of electroencephalogram signals based on signal characteristics of the plurality of source signals; and an attention detection operation corresponding to the attention detection task is performed according to the target source signals corresponding to each group of electroencephalogram signals to obtain a task processing result of the attention detection task; the task processing result includes an attention concentration level.

[0012] In an aspect, an embodiment of the present application provides a computer device, the computer device comprising an input interface and an output interface, and further comprising:

[0013] a processor and a computer readable storage medium;

[0014] a computer readable storage medium configured to store a computer program;

[0015] a processor configured to run the computer program to implement the above-mentioned signal processing method.

[0016] In an aspect, an embodiment of the present application provides a computer readable storage medium, the computer readable storage medium storing a computer program, the computer program being adapted to be loaded by a processor and to execute the above-mentioned signal processing method.

[0017] In an aspect, an embodiment of the present application provides a computer program product, the computer program product comprising a computer program, the computer program being adapted to be loaded by a processor and to execute the above-mentioned signal processing method.

[0018] In the embodiments of the present application, N groups of brain electrical signals collected by the detection object in the corresponding brain region of the target object can be obtained according to the target brain electrical acquisition rule corresponding to the attention detection task, and then the task processing result containing the attention concentration level can be determined based on the brain electrical signals, and the attention concentration level can indicate the degree of attention concentration of the target object. The attention detection process is not affected by environmental factors such as light, and the accuracy of the attention concentration level obtained by attention detection is high, that is, the accuracy of attention detection is high, the stability is high, and the application range is wide. Moreover, in the process of attention detection based on brain electrical signals, the target source signal associated with the attention detection task can be extracted from the obtained brain electrical signals, and then the attention detection operation corresponding to the attention detection task can be performed according to the extracted target source signal associated with the attention detection task, and the task processing result containing the attention concentration level is obtained. Compared with the task processing result obtained by directly performing the attention detection operation corresponding to the attention detection task according to the obtained brain electrical signals, the accuracy of the task processing result can be improved, that is, the accuracy of the obtained attention concentration level can be improved, and the accuracy of the attention detection can be improved. BRIEF DESCRIPTION OF DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0020] Figure 1a is a structural schematic diagram of a signal processing system provided by an embodiment of the present application;

[0021] Figure 1b is a structural schematic diagram of another signal processing system provided by an embodiment of the present application;

[0022] Figure 2 is a flowchart of a signal processing method provided by an embodiment of the present application;

[0023] Figure 3a is a schematic diagram of collecting brain electrical signals provided by an embodiment of the present application;

[0024] Figure 3b is a schematic diagram of detecting attention information of a detection period provided by an embodiment of the present application;

[0025] Figure 3c is a schematic diagram of detecting attention evaluation value of a detection period provided by an embodiment of the present application;

[0026] Figure 3d is another detection period attention evaluation value schematic diagram provided by an embodiment of the application;

[0027] Figure 3e is an interactive interface schematic diagram provided by an embodiment of the application;

[0028] Figure 3f is a viewing feature data schematic diagram provided by an embodiment of the application;

[0029] Figure 4 is a flowchart of another signal processing method provided by an embodiment of the application;

[0030] Figure 5a is a determination target number schematic diagram provided by an embodiment of the application;

[0031] Figure 5b is a determination attention detection model schematic diagram provided by an embodiment of the application;

[0032] Figure 6 is a signal processing device structure schematic diagram provided by an embodiment of the application;

[0033] Figure 7 is a computer device structure schematic diagram provided by an embodiment of the application. DETAILED DESCRIPTION

[0034] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application.

[0035] The signal processing scheme provided by the embodiments of the application can acquire N groups of brain electrical signals collected by a detection object in a target brain electrical acquisition rule corresponding brain region according to the target brain electrical acquisition rule, and respectively pre-process each group of brain electrical signals to obtain a plurality of source signals of each group of brain electrical signals in the N groups of brain electrical signals, and respectively determine a target source signal associated with an attention detection task from the plurality of source signals of each group of brain electrical signals based on signal features of the plurality of source signals. Wherein, N is a positive integer. The attention detection operation corresponding to the attention detection task can be performed according to the target source signal corresponding to each group of brain electrical signals, and a task processing result of the attention detection task is obtained. Wherein, the task processing result includes an attention concentration level.

[0036] In a possible implementation, the signal processing scheme can be executed by a signal processing device, which can be a terminal device or a server. The terminal device mentioned in the embodiments of the present application can be a smart phone, a tablet computer, a notebook computer, a desktop computer, a smart voice interaction device, a smart home appliance, a smart vehicle, and a smart wearable device (for example, smart glasses, a smart helmet), and the like, which are not limited in the embodiments of the present application. The server mentioned in the embodiments of the present application can be a standalone physical server, a server cluster or a distributed system composed of multiple physical servers, a cloud server providing 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 basic cloud computing services such as big data and artificial intelligence platforms, and the like, which are not limited in the embodiments of the present application.

[0037] In the signal processing scheme, the attention detection task aims to detect an attention concentration level indicating an attention concentration degree. If the attention detection task is for a target object, the signal processing device can acquire, according to a target electroencephalogram acquisition rule, N groups of electroencephalogram signals collected by the detection object at corresponding brain regions of the target object, and then detect, based on the acquired N groups of electroencephalogram signals, an attention concentration level indicating the attention concentration degree of the target object to obtain a task processing result. The target object can be any object to be detected for attention, which is not limited in the embodiments of the present application. For ease of description, the object (i.e., the user) to be detected for attention will be referred to as the target object in subsequent embodiments of the present application.

[0038] Among them, the electroencephalogram (EEG) is one of the bioelectric signals, the bioelectric signal is the electric phenomenon closely related to the life activity and regular generated by the cells or tissues of the organism in the active or static state, and the essence is the electric signal generated by the transmembrane flow of sodium, potassium and other ions. Specifically, the electroencephalogram is the electric signal generated by the activity of the neurons in the brain, wherein the neurons, blood vessels and glial cells constitute the human brain, the number of neurons in the human brain is about 100 billion, the brain occupies the largest volume of the human brain, and is composed of two brain hemispheres, the brain is above all brain tissues and is covered by the complex cerebral cortex, and the electroencephalogram is the sum of the post-synaptic potentials when the neural cells in the cerebral cortex perform electrical activity, and the electroencephalogram can be collected by the electrode collection mode and the change of the electric potential can be shown in the form of a waveform graph. In a possible implementation manner, the electroencephalogram can be collected by the electroencephalogram signal collection device such as the electroencephalogram helmet and the electroencephalogram head ring supporting the electrode collection mode, and based on this, it can be known that the detection object in the above signal processing scheme is the electroencephalogram signal collection device for collecting the electroencephalogram, and the electroencephalogram signal collection device can be selected according to the specific requirement, which is not limited in the embodiment of the application; for example, the wireless dry EEG system DSI-24 of the scientific research level can be selected, which has 24 leads and can easily penetrate the hair, thereby solving the signal difference caused by the thickness of the hair and ensuring the reliability of the signal quality; the flexibility and convenience of the dry EEG system can adapt to various scenes, and can effectively work even in a complex environment, thereby avoiding the step of pretreating the scalp and greatly shortening the preparation time; and the electrode adopts a double-spring design, which can ensure stable pressure, improve wearing comfort, effectively reduce the motion artifact, and the electrode also adopts active or passive shielding technology to reduce the influence of electromagnetic on the signal. Among them, the detection object and the signal processing device with communication requirements can realize data interaction based on wired and / or wireless communication mode, for example, data can be transmitted through Bluetooth, and the embodiment of the application is not limited.

[0039] According to different frequency bands, the electroencephalogram signal can be generally divided into five types of signals, namely, δ wave, θ wave, α wave, β wave and γ wave. The following introduces each type of signal: ① The frequency band of δ wave is 0.5 Hz to 4 Hz, and the amplitude is about 0 μV to 200 μV, which belongs to slow wave; the electroencephalogram signal of this wave band usually appears in the case that a person is in a deep sleep state, brain lesion, hypoxia or anesthesia state, and is more obvious in the period of newborn or immature intelligence development. ② The frequency band of θ wave is 4 Hz to 8 Hz, and the amplitude is between 100 μV and 150 μV, which also belongs to slow wave; this wave band usually appears in the initial stage of adult sleep (i.e. light sleep), is more obvious in meditation, and is the main component wave band of the electroencephalogram signal of teenagers; θ wave band is closely related to the mental state of a person, and usually appears and lasts for a short period of time when depression or adverse stimulation occurs; when the spirit is in a state of joy, the wave band usually disappears, based on which it can be known that the wave band can directly reflect the mental state of a person. ③ The frequency band of α wave is between 8 Hz and 14 Hz, the amplitude is 30 μV to 100 μV, and the amplitude characteristic is gradually changing, which is shuttle-shaped; α wave generally appears when the brain of a person is in a relaxed or focused state, and is particularly obvious when awake or quiet; the position of the wave band usually appears in the occipital region and the posterior part of the parietal region of the brain. ④ The frequency band of β wave is about 14 Hz to 31 Hz, and the amplitude is 5 μV to 20 μV; the position of β wave generally appears in the frontal region or other regions of the cerebral cortex; the wave band is the electroencephalogram of a person in a daily awake state, and is closely related to the attention level, tense mental state and excited emotion, and is more obvious in the case of inner tension, high spirits and active thinking, attention concentration and the like. ⑤ The frequency band of γ wave is about 31 Hz to 50 Hz, and the amplitude range is indefinite; usually, the position of the wave band appears in the frontal region of the brain.

[0040] Based on the above, the waves related to attention in the electroencephalogram signal include the alpha wave and the beta wave. In the signal processing scheme, the signal processing device, after obtaining N groups of electroencephalogram signals collected by the detection object in the corresponding brain region, respectively pre-processes each group of electroencephalogram signals to obtain a plurality of source signals of each group of electroencephalogram signals in the N groups of electroencephalogram signals, and respectively determines a target source signal associated with the attention detection task from the plurality of source signals of each group of electroencephalogram signals based on the signal characteristics of the plurality of source signals. The process aims to extract the waves related to attention from the obtained electroencephalogram signal, that is, to extract one or more of the alpha wave and the beta wave from the obtained electroencephalogram signal. In a specific application scenario, the type of target source signal associated with the attention detection task that needs to be extracted can be set according to specific needs. For example, in one application scenario, the type of target source signal associated with the attention detection task that needs to be extracted is set to the beta wave, and then the signal processing device can only extract the beta wave as the target source signal associated with the attention detection task. For another example, in another application scenario, the type of target source signal associated with the attention detection task that needs to be extracted is set to the alpha wave and the beta wave, and then the signal processing device can extract the alpha wave and the beta wave together as the target source signal associated with the attention detection task. The present application embodiment does not limit this. In another optional implementation, although the alpha wave and the beta wave are associated with different states of consciousness and attention levels, since attention is a complex psychological process involving multiple cognitive functions and the synergistic effect of multiple brain regions, when setting the type of target source signal associated with the attention detection task that needs to be extracted, other electroencephalogram wave bands may also need to be considered, for example, the theta wave is also reflected in some forms of cognitive load and memory process, the gamma wave appears in high-level cognitive functions, and so on.

[0041] In summary, the electroencephalogram signal can be considered as a mixed signal composed of multiple source signals (for example, the above-mentioned five types of signals), and the signal processing scheme proposed in the embodiments of the present application aims to extract the target source signal associated with the attention detection task from the obtained electroencephalogram signal, and then the attention detection operation corresponding to the attention detection task can be performed according to the extracted target source signal associated with the attention detection task, to obtain the task processing result of the attention detection task. Compared with the task processing result obtained by directly performing the attention detection operation corresponding to the attention detection task according to the obtained electroencephalogram signal, the accuracy of the task processing result can be improved. Moreover, generally speaking, the amplitude of the electroencephalogram signal is very weak, with a maximum of about 100 microvolts, which is very small electrical activity and is easily disturbed by other factors. The electroencephalogram signal collected by the detection object acquired by the signal processing device usually mixes with interference signals. Based on this, the way of extracting the target source signal associated with the attention detection task from the obtained electroencephalogram signal in the above-mentioned signal processing scheme can also reduce the influence of the interference signal on the accuracy of the task processing result, so as to improve the accuracy of the task processing result.

[0042] In a possible implementation, the above-mentioned signal processing scheme can be executed by the signal processing device alone. Based on this, please refer to Figure 1a The structure schematic diagram of a signal processing system provided by the embodiments of the present application can include a detection object (i.e. an electroencephalogram signal collection device) 101 and a signal processing device 102. The above-mentioned signal processing scheme is executed by Figure 1a When the signal processing system shown in the above-mentioned signal processing scheme is executed, the signal processing device 102 can acquire N groups of electroencephalogram signals collected by the detection object 101 in the brain region corresponding to the target electroencephalogram acquisition rule according to the target electroencephalogram acquisition rule; pre-process each group of electroencephalogram signals respectively to obtain multiple source signals of each group of electroencephalogram signals in the N groups of electroencephalogram signals; determine the target source signal associated with the attention detection task from the multiple source signals of each group of electroencephalogram signals based on the signal features of the multiple source signals; and perform the attention detection operation corresponding to the attention detection task according to the target source signal corresponding to each group of electroencephalogram signals to obtain the task processing result of the attention detection task.

[0043] In another possible implementation, the above-mentioned signal processing scheme can also be executed by multiple computer devices with computing power in cooperation, for example, when the signal processing device is a terminal device, the terminal device and the server can execute the signal processing scheme in cooperation. Based on this, please refer to Figure 1bAnother structural schematic diagram of a signal processing system provided by an embodiment of the present application is shown in FIG. 11. The signal processing system can include a detection object (i.e., a brain electrical signal acquisition device) 111, a signal processing device 112, and a server 113. The devices with communication requirements can realize data interaction based on wired and / or wireless communication modes, which is not limited in the embodiments of the present application. For example, when the above signal processing scheme is executed by the signal processing system shown in FIG. 11, the signal processing device 112 can acquire N groups of brain electrical signals collected by the detection object 111 in the brain region corresponding to the target brain electrical acquisition rule according to the target brain electrical acquisition rule, and send the acquired N groups of brain electrical signals to the server 113. The server 113 can execute related processes such as preprocessing of each group of brain electrical signals, obtain the task processing result of the attention detection task, and return the task processing result to the signal processing device 112. For another example, when the above signal processing scheme is executed by the signal processing system shown in FIG. 11, the signal processing device 112 can acquire N groups of brain electrical signals collected by the detection object 111 in the brain region corresponding to the target brain electrical acquisition rule according to the target brain electrical acquisition rule, and process target source signals associated with the attention detection task for each group of brain electrical signals, and send the target source signals corresponding to each group of brain electrical signals to the server 113. The server 113 can execute the attention detection operation corresponding to the attention detection task according to the target source signals corresponding to each group of brain electrical signals, obtain the task processing result of the attention detection task, and return the task processing result to the signal processing device 112. For convenience of description, the signal processing scheme is taken as an example executed by the signal processing device in the subsequent embodiments of the present application. Figure 1b Figure 1b For another example, when the above signal processing scheme is executed by the signal processing system shown in FIG. 11, the signal processing device 112 can acquire N groups of brain electrical signals collected by the detection object 111 in the brain region corresponding to the target brain electrical acquisition rule according to the target brain electrical acquisition rule, and process target source signals associated with the attention detection task for each group of brain electrical signals, and send the target source signals corresponding to each group of brain electrical signals to the server 113. The server 113 can execute the attention detection operation corresponding to the attention detection task according to the target source signals corresponding to each group of brain electrical signals, obtain the task processing result of the attention detection task, and return the task processing result to the signal processing device 112. For convenience of description, the signal processing scheme is taken as an example executed by the signal processing device in the subsequent embodiments of the present application.

[0044] It should be particularly pointed out that the collection and processing of related data (such as brain electrical signals) in the present application should strictly comply with the requirements of laws and regulations, obtain the informed consent or separate consent of the personal information subject, and carry out subsequent data use and processing behavior within the scope of authorization of laws and regulations and the personal information subject. When the related embodiments of the present application are applied to specific products or technologies, the related data collection, use and processing process should comply with the requirements of national laws and regulations, the information processing rules should be informed before collecting biological information (including brain electrical signals) and the separate consent of the subject should be sought, and the biological information should be processed in strict accordance with the requirements of laws and regulations and the rules of personal information processing, and technical measures should be taken to ensure the safety of related data.

[0045] Based on the above description, an embodiment of the present application provides a signal processing method, which is shown in FIG. 12. Figure 2 A flowchart of a signal processing method provided by an embodiment of the present application is shown in FIG. 12. The signal processing method can be executed by a signal processing device, and can include the following steps S201-S205.

[0046] ​S201, in response to a task request about the attention detection task, determining a target electroencephalogram acquisition rule corresponding to the attention detection task according to a correspondence between the task and the electroencephalogram acquisition rule.

[0047] The electroencephalogram acquisition rule is used to indicate a brain region where the electroencephalogram signal needs to be acquired, and different tasks correspond to different electroencephalogram acquisition rules.

[0048] In a possible implementation, the correspondence between the task and the electroencephalogram acquisition rule can be pre-configured according to specific requirements. For example, as described above, the alpha wave and the beta wave are related to attention, and the delta wave and the theta wave are related to sleep. Illustratively, the brain region where the electroencephalogram signal needs to be acquired, which is indicated by the pre-configured electroencephalogram acquisition rule corresponding to the attention detection task, can be the brain region where the alpha wave and the beta wave appear. The brain region where the electroencephalogram signal needs to be acquired, which is indicated by the pre-configured electroencephalogram acquisition rule corresponding to the fatigue detection task, can be the brain region where the delta wave and the theta wave appear. The fatigue detection task aims to detect the degree of fatigue based on the electroencephalogram signal. Since the alpha wave related to attention usually appears in the occipital region and the posterior part of the parietal region of the brain, and the beta wave related to attention usually appears in the frontal region and other regions of the cerebral cortex, for example, in the requirement of an application scenario, only the beta wave is extracted as the target source signal associated with the attention detection task, and then the brain region where the electroencephalogram signal needs to be acquired, which is indicated by the pre-configured electroencephalogram acquisition rule corresponding to the attention detection task, can be the frontal region, or the frontal region and other regions. For another example, in the requirement of another application scenario, only the alpha wave is extracted as the target source signal associated with the attention detection task, and then the brain region where the electroencephalogram signal needs to be acquired, which is indicated by the pre-configured electroencephalogram acquisition rule corresponding to the attention detection task, can be the occipital region and the parietal region.

[0049] In a possible implementation, the S201 is an optional step, that is, the S202 described below can be started without initiating the task request. For example, when the detection period corresponding to the attention detection task arrives, or when a preset time period for performing the attention detection task arrives, the S202 described below is directly started.

[0050] S202, acquiring N groups of electroencephalogram signals collected by the detection object in the brain region corresponding to the target electroencephalogram acquisition rule according to the target electroencephalogram acquisition rule.

[0051] Wherein, N is a positive integer. Generally, the target object can wear an electroencephalogram signal acquisition device (i.e., a detection object) supporting electrode acquisition mode. The electroencephalogram signal acquisition device can acquire electroencephalogram signals of corresponding brain regions through electrodes arranged in different brain regions, and one electrode corresponds to the acquisition of one set of electroencephalogram signals. Different electroencephalogram signal acquisition devices can have different numbers of electrodes and can have specific wearing modes. Therefore, when the electroencephalogram signal acquisition device is selected, the wearing mode required by the selected electroencephalogram signal acquisition device needs to be worn to enable the acquisition of electroencephalogram signals. The embodiments of the present application do not limit this.

[0052] In a possible implementation, after the signal processing device determines the target electroencephalogram acquisition rule corresponding to the attention detection task, the signal processing device can output the brain region indicated by the target electroencephalogram acquisition rule to prompt that when the electroencephalogram signal acquisition device is worn, only the electrodes of the corresponding brain region can be worn, so that the electroencephalogram signal acquisition device only acquires the electroencephalogram signals of the corresponding brain region when acquiring the electroencephalogram signals and transmits them to the signal processing device. Only the electroencephalogram signals of the brain region indicated by the target electroencephalogram acquisition rule can be transmitted, which can reduce resource consumption in the process of transmitting electroencephalogram signals. In a possible implementation, after the signal processing device receives the electroencephalogram signals of at least one brain region transmitted by the electroencephalogram signal acquisition device, the signal processing device can screen the electroencephalogram signals of the brain region indicated by the target electroencephalogram acquisition rule from the received electroencephalogram signals of each brain region according to the brain region indicated by the target electroencephalogram acquisition rule, to obtain N sets of electroencephalogram signals collected by the detection object in the corresponding brain region. In a possible implementation, the signal processing device can output and display the obtained electroencephalogram signals in the form of a waveform diagram. For example, please refer to Figure 3a An example of collecting electroencephalogram signals provided by the embodiments of the present application is shown in FIG. 301. The distribution of the electrodes required to collect electroencephalogram signals can be shown as indicated by 301.

[0053] S203, respectively, pre-process each set of electroencephalogram signals to obtain a plurality of source signals of each set of electroencephalogram signals in the N sets of electroencephalogram signals.

[0054] In a possible implementation, the brain electrical signals collected by the detection object can be mixed with interference signals, such as eye electrical signals (electrical signals generated due to eye movement), heart electrical signals (electrical signals generated due to heart movement), and the like, based on which the signal processing device performs preprocessing on any one group of brain electrical signals to obtain a plurality of source signals of the group of brain electrical signals, so as to separate various source signals from the observation signals (i.e., mixed signals, here, the brain electrical signals obtained by the signal processing device) mixed with the plurality of source signals, and then determine a target source signal associated with the attention detection task from the separated various source signals for subsequent related processing processes of the attention detection.

[0055] In a possible implementation, the signal processing device respectively performs preprocessing on each group of brain electrical signals, and in the process of obtaining a plurality of source signals of each group of brain electrical signals in the N groups of brain electrical signals, the following steps can be performed: performing filtering processing on the i-th group of brain electrical signals in the N groups of brain electrical signals to obtain an intermediate signal corresponding to the i-th group of brain electrical signals; wherein i ∈ [1, N]; performing blind source separation processing on the intermediate signal corresponding to the i-th group of brain electrical signals to obtain a plurality of source signals of the i-th group of brain electrical signals.

[0056] For the filtering related process, the filtering process aims to filter out the frequency bands that are not needed in the signal, so that the useful and required signals pass through, which is an effective means of extracting original signal information from the signals affected and contaminated by noise, and the result of filtering is to obtain a specific frequency band or eliminate a specific frequency band. The filtering process can select one or more of low-pass filtering, high-pass filtering, band-pass filtering, and the like according to specific needs, which is not limited in the embodiments of the present application; the filtering process of the embodiments of the present application aims to retain the frequency band associated with the attention detection task in the obtained brain electrical signals, and the frequency band associated with the attention detection task that needs to be retained can be set according to specific needs, and at least should include the frequency band of the target source signal associated with the attention detection task that needs to be extracted, for example, when the target source signal associated with the attention detection task that needs to be extracted is set to be a beta wave, the filtering process should at least include the frequency band corresponding to the beta wave.

[0057] The process is related to blind source separation. Blind source separation (BSS), also known as blind signal separation, refers to a process of separating source signals from mixed signals (i.e., observed signals) in a case where a theoretical model of the signals and the source signals cannot be accurately obtained. The blind source separation method can be selected according to specific requirements, and embodiments of the present application do not make any limitation. For example, the independent component analysis (ICA), principal component analysis (PCA), or the like can be selected. Taking the independent component analysis as an example of the blind source separation method, the independent component analysis is a signal separation and feature extraction method, which can separate mixed signals into independent components and can be used to remove noise and artifacts. The independent component analysis can be divided into two categories: an algebraic method based on statistics and an iterative estimation method based on information theory criteria. From the implementation principle, both of them use the non-Gaussian distribution characteristics and the independence of the source signals. The assumptions of the ICA are as follows: 1. the source signals and each component are statistically independent; 2. the independent components have non-Gaussian characteristics, and at most one component is a Gaussian random variable; and 3. the mixing matrix is a square matrix. The principle of the algorithm is that the observed signal X is obtained by linear combination of the independent components S (S represents a matrix of independent components) through the mixing matrix A, that is, X = AS. Generally, the ICA algorithm can include the following steps: 1. centering X so that the mean of X is 0; 2. whitening the centered observed signal, that is, performing linear transformation on the centered X to remove the correlation; and 3. constructing an orthogonal system to obtain the independent components S. For example, in a possible implementation, based on the independent component analysis, 8 source signals can be separated from a group of electroencephalogram signals in a period of time.

[0058] In S204, based on the signal characteristics of the plurality of source signals, a target source signal associated with the attention detection task is determined from the plurality of source signals of each group of electroencephalogram signals.

[0059] In a possible implementation, the type of the target source signal associated with the attention detection task to be extracted can be set according to specific requirements. For example, in an application scenario, the type of the target source signal associated with the attention detection task to be extracted is set to be beta waves, and then the signal processing device can extract only the beta waves as the target source signal associated with the attention detection task. For another example, in another application scenario, the type of the target source signal associated with the attention detection task to be extracted is set to be alpha waves and beta waves, and then the signal processing device can extract the alpha waves and the beta waves together as the target source signal associated with the attention detection task. Embodiments of the present application do not make any limitation. For ease of description, the beta waves are taken as an example.

[0060] In a possible implementation, in the process of determining the target source signal associated with the attention detection task from the plurality of source signals of each set of electroencephalogram signals based on the signal features of the plurality of source signals, the signal processing device can perform the following steps: performing feature conversion processing on the plurality of source signals of the i-th set of electroencephalogram signals respectively to obtain a plurality of signal features of the i-th set of electroencephalogram signals; and determining the target source signal associated with the attention detection task from the plurality of source signals of the i-th set of electroencephalogram signals according to the plurality of signal features of the i-th set of electroencephalogram signals; wherein i ∈ [1, N], and the signal feature obtained by performing feature conversion processing on any one source signal of the i-th set of electroencephalogram signals, i.e., the feature extracted from the source signal, can be used to represent the characteristics of the source signal. In this embodiment of the present application, the feature extracted from the signal can be in the form of a feature vector or a feature matrix.

[0061] The above process of determining the target source signal associated with the attention detection task from the plurality of source signals of the i-th set of electroencephalogram signals according to the plurality of signal features of the i-th set of electroencephalogram signals is to highlight the differences between the plurality of source signals obtained by preprocessing the i-th set of electroencephalogram signals from the perspective of signal features. For example, the plurality of source signals obtained by preprocessing the i-th set of electroencephalogram signals are similar in some characteristics and have large differences in other characteristics. In this embodiment of the present application, the signal features are extracted to extract the characteristics with large differences, so as to better determine the target source signal associated with the attention detection task from the plurality of source signals obtained by preprocessing. That is, the target source signal can be extracted according to the characteristics of the target source signal associated with the attention detection task in the signal feature. Based on this, the type of signal feature to be extracted can also be selected according to specific requirements. In the process of performing corresponding feature conversion processing, a feature conversion processing mode suitable for the corresponding type of signal feature can be adopted, and details are not described herein. For example, the feature conversion processing includes differential entropy linear dynamic system smoothing processing, and the corresponding signal feature is a differential entropy linear dynamic system smoothing feature. Or the feature conversion processing includes differential entropy moving average processing, and the corresponding signal feature is a differential entropy moving average feature. Or the feature conversion processing includes power spectral density linear dynamic system smoothing processing, and the corresponding signal feature is a power spectral density linear dynamic system smoothing feature. Or the feature conversion processing includes power spectral density moving average processing, and the corresponding signal feature is a power spectral density moving average feature.

[0062] S205, performing an attention detection operation corresponding to the attention detection task according to the target source signal corresponding to each set of electroencephalogram signals to obtain a task processing result of the attention detection task.

[0063] The task processing result can include an attention concentration level, which can be used to indicate the degree of attention concentration. In a possible implementation, the attention concentration level in the task processing result can be one of a plurality of preset attention concentration levels, which can be set according to specific requirements. For example, in the requirement of one application scenario, the plurality of preset attention concentration levels are the first level and the second level in the order of low to high degree of attention concentration, where the first level specifically represents inattention (i.e., distraction), and the second level specifically represents attention. For another example, in the requirement of another application scenario, the plurality of preset attention concentration levels are the first level, the second level, and the third level in the order of low to high degree of attention concentration, where the first level specifically represents severe inattention, the second level specifically represents mild inattention, and the third level specifically represents complete attention, and so on. In a possible implementation, after obtaining the task processing result, the signal processing device can output the task processing result, and the specific format and content of the task processing result can be set according to specific requirements, which are not limited in the embodiments of the present application. For example, if the plurality of preset attention concentration levels are the first level and the second level in the order of low to high degree of attention concentration, and the signal processing device detects the first level of attention concentration level, the task processing result can be output in a text format, and the specific content can be: the detected attention concentration level is the first level (inattention), for example, the task processing result can also be output in a voice format, and so on, which are not limited in the embodiments of the present application.

[0064] In a possible implementation, in the process of obtaining the task processing result of the attention detection task according to the target source signal corresponding to each group of electroencephalogram signals, the signal processing device can perform the following steps: performing feature extraction processing on the target source signal corresponding to each group of electroencephalogram signals to obtain the electroencephalogram signal features of each group of electroencephalogram signals; calling the attention detection model to perform attention analysis on the electroencephalogram signal features of each group of electroencephalogram signals to obtain an attention evaluation value at a current acquisition time point; the current acquisition time point is the acquisition time point of the N groups of electroencephalogram signals, and the attention evaluation value is used to indicate the probability of attention concentration; and generating the task processing result of the attention detection task based on the attention evaluation value at the current acquisition time point and a historical attention evaluation value determined at a target period before the current acquisition time point.

[0065] The feature extraction processing related process is described.

[0066] The EEG signal feature of any one of the N groups of EEG signals is an extracted feature of the target source signal corresponding to the corresponding EEG signal, that is, a feature obtained by performing feature extraction processing on the target source signal corresponding to the corresponding EEG signal. The feature extraction process aims to extract useful feature information from the signal to facilitate subsequent application. The type of EEG signal feature to be extracted can be selected according to specific requirements. For example, the type of EEG signal feature to be extracted can be selected according to specific requirements to select one or more of time domain features, frequency domain features, time-frequency domain features, and the like. The corresponding feature extraction processing mode is selected to adapt to the type of EEG signal feature to be extracted. The embodiments of the present application do not repeat the description. For example, in one possible implementation, when performing feature extraction processing on the target source signal corresponding to each group of EEG signals, frequency domain features can be extracted, for example, one or more of power spectral density, energy, and the like. For example, in another possible implementation, when performing feature extraction processing on the target source signal corresponding to each group of EEG signals, time domain features can be extracted, for example, one or more of signal mean, variance, kurtosis, skewness, and the like. The embodiments of the present application subsequently take the extraction of power spectral density as an example. The power spectral density (PSD) of a signal refers to the power carried by each unit frequency wave after the frequency spectral density of the corresponding signal is multiplied by a suitable coefficient. Generally, the unit is watt per hertz (W / Hz) or watt per nanometer (W / nm).

[0067] The process of obtaining the attention evaluation value of the current acquisition time point by calling the attention detection model.

[0068] In a possible implementation, the attention detection model can be used to predict the attention evaluation value at the current acquisition time point according to the electroencephalogram features of each of the N groups of electroencephalogram signals. The attention detection model can be selected according to specific requirements, and the embodiments of the present application do not make any limitation. For example, the support vector machine (SVM), K-Nearest Neighbors (KNN) model, decision tree model, random forest model, neural network model, deep learning model (such as convolutional neural network model CNN, recurrent neural network model RNN), etc. in the field of artificial intelligence can be selected. In other words, the attention detection model can be implemented based on artificial intelligence technology, and specifically can be implemented based on machine learning / deep learning technology in artificial intelligence technology. In other words, artificial intelligence is a theory, method, technology and application system for using a digital computer or a machine controlled by a digital computer to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use the knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technology of computer science, which tries to understand the essence of intelligence and produce a new intelligent machine that can react in a similar way to human intelligence. Artificial intelligence is the design principle and implementation method of various intelligent machines, so that the machine has the functions of perception, reasoning and decision-making. Artificial intelligence technology is a comprehensive discipline, involving a wide range of fields, including both hardware and software technologies. Artificial intelligence basic technologies generally include sensors, special artificial intelligence chips, cloud computing, distributed storage, big data processing technology, pre-training model technology, operation / interaction system, mechatronics, etc. Among them, the pre-training model (Pre-training model) is also called cornerstone model, large model, which refers to a deep neural network (Deep neural network, DNN) with large parameters. It is trained on a large amount of unlabeled data, and uses the function approximation capability of the large parameter DNN to extract common features from the data. Through fine tuning, parameter efficient fine tuning (PEFT) and other technologies, it is suitable for downstream tasks. Artificial intelligence software technology mainly includes computer vision technology, speech processing technology, natural language processing technology and machine learning / deep learning, etc. Machine learning (Machine Learning, ML) is a multi-disciplinary subject, involving probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory, etc. It is a subject that studies how computers simulate or implement human learning behavior to acquire new knowledge or skills, and reorganize existing knowledge structure to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental approach to making computers intelligent, and its applications are widespread in various fields of artificial intelligence.Machine learning and deep learning generally include artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and teach-in learning technologies, and the pre-trained model is the latest development of deep learning, which integrates the above technologies.

[0069] Taking the attention detection model as a neural network model, in a possible implementation, the signal processing device can input the EEG signal features of each group of EEG signals to the attention detection model, and the attention detection model can predict the attention evaluation value according to the input EEG signal features of the EEG signals. The signal processing device can take the attention evaluation value predicted by the attention detection model as the attention evaluation value at the current acquisition time point. The ability of the attention detection model to predict the attention evaluation value according to the input EEG signal features of the EEG signals can be learned in the training process of the attention detection model. The training process of the attention detection model is introduced in subsequent embodiments.

[0070] Taking the attention detection model as a KNN model, the KNN model is a model implemented based on the KNN algorithm, which is a supervised learning algorithm that can be used for classification and regression tasks. The basic idea is to predict the class or value of a sample according to the class of its neighboring samples (for classification tasks) or value (for regression tasks) in the feature space. In a possible implementation, the signal processing device calls the attention detection model to perform attention analysis on the EEG signal features of each group of EEG signals to obtain the attention evaluation value at the current acquisition time point. The following steps can be performed: calling the attention detection model, determining the feature similarity between each reference feature data in the target feature data and the reference feature data set, wherein the target feature data includes the EEG signal features of the N groups of EEG signals, and the reference feature data set is determined in the training process of the attention detection model and includes multiple EEG signal features of N groups of EEG signals for training; according to the feature similarity, screening a target number of reference feature data from the reference feature data set; wherein the feature similarity between any reference feature data screened out and the target feature data is greater than or equal to the feature similarity between other reference feature data not screened out and the target feature data; obtaining the attention labels corresponding to each reference feature data screened out, and statistically obtaining the statistical number of reference feature data corresponding to each type of attention label; wherein the types of attention labels include labels representing attention concentration and labels representing attention dispersion; and determining the attention evaluation value at the current acquisition time point according to the statistical number of each type of attention label.

[0071] Any reference feature data has the same feature dimension as the target feature data, for example, the target feature data has features A, B, and C, each corresponding to feature values A1, B1, and C1, and one reference feature data has the same features A, B, and C, each corresponding to feature values A2, B2, and C2. The target feature data specifically includes the EEG signal features of the N groups of EEG signals of the target object at the current collection time point, and one reference feature data in the reference feature data set includes the EEG signal features of the N groups of EEG signals for training. For example, the EEG signal features of the N groups of EEG signals for training can be the EEG signal features of the N groups of EEG signals of a sample object at a collection time point, and the corresponding attention label can be labeled. The process of determining the EEG signal features of the N groups of EEG signals of a sample object at a collection time point is similar to that of determining the EEG signal features of the N groups of EEG signals of the target object at the current collection time point, and will not be repeated here. For example, the N groups of EEG signals of a sample object in the brain region indicated by the target EEG acquisition rule can be collected in a period of time when the sample object is in an attention concentrated state. The EEG signal features determined based on the N groups of EEG signals collected at any collection time point in the period of time when the sample object is in the attention concentrated state can be used as the EEG signal features of the N groups of EEG signals for training, and the corresponding attention label can be labeled as a label category representing attention concentration. For example, the N groups of EEG signals of a sample object in the brain region indicated by the target EEG acquisition rule can be collected in a period of time when the sample object is in an attention dispersed state. The EEG signal features determined based on the N groups of EEG signals collected at any collection time point in the period of time when the sample object is in the attention dispersed state can be used as the EEG signal features of the N groups of EEG signals for training, and the corresponding attention label can be labeled as a label category representing attention dispersion. Based on this, the EEG signal features of the sample object can be obtained by collecting the EEG signals of one or more sample objects, and the reference feature data set can be constructed.

[0072] The feature similarity can be selected according to specific requirements, for example, the cosine similarity, the feature distance, etc. can be selected as the feature similarity, and the feature distance can be selected according to specific requirements, for example, the Euclidean distance (i.e., Euler distance), Mahalanobis distance, Manhattan distance, Minkowski distance, Chebyshev distance, etc. The Euclidean distance between any two feature data can be represented by the following formula: where x i represents the feature value of feature x at the i-th feature dimension, and y iy i represents a feature value of a feature y in an i-th feature dimension; the target number (K) can be set according to specific requirements, for example, can be determined according to the number of reference feature data in the reference feature data set, and the square root of the number of reference feature data in the reference feature data set can be determined as K, if the square root of the number of reference feature data in the reference feature data set is even, it can be converted into an odd number (for example, plus one or minus one), and the converted number is taken as K, in this way, the deviation of the KNN model prediction can be reduced.

[0073] In a possible implementation, when the signal processing device determines the attention evaluation value of the current collection time point according to the statistical quantity corresponding to each type of attention label, the ratio of the statistical quantity corresponding to the attention label of the label category used to represent attention concentration to the target number can be taken as the attention evaluation value of the current collection time point, which is used to indicate the probability of attention concentration of the target object at the current collection time point.

[0074] For the process of generating the task processing result of the attention detection task, that is, for generating the task processing result of the attention detection task based on the attention evaluation value of the current collection time point and the historical attention evaluation value determined in the target period before the current collection time point; wherein the target period can be set according to specific requirements, for example, the target period can be set as 30s before the current collection time point, 60s before the current collection time point, etc., based on which it can be known that the attention concentration level in the task processing result generated at the current collection time point can indicate the degree of attention concentration of the target object in the detection period (which contains the current collection time point and the target period), when the length of the target period is set to 0, the attention concentration level in the task processing result generated at the current collection time point can indicate the degree of attention concentration of the target object at the current collection time point.

[0075] In a possible implementation, in the process of generating the task processing result of the attention detection task based on the attention evaluation value of the current collection time point and the historical attention evaluation value determined in the target period before the current collection time point, the following steps can be performed: obtaining the historical attention evaluation values of the target period at a plurality of historical collection time points; determining the time length proportion of the attention evaluation value greater than the attention threshold value according to the attention evaluation value of the current collection time point and the historical attention evaluation values of the plurality of historical collection time points; determining the attention concentration level according to the time length proportion, and generating the task processing result of the attention detection task based on the determined attention concentration level; wherein the degree of attention concentration indicated by the determined attention concentration level is positively correlated with the time length proportion.

[0076] The historical attention evaluation value of any historical collection time point in the target period, i.e., the attention evaluation value of the corresponding historical collection time point, is determined in a similar manner as the determination of the attention evaluation value of the current collection time point, and the embodiments of the present application will not be described herein. The signal processing device can store the historical attention evaluation value of each collection time point after determining the historical attention evaluation value of each collection time point. Based on the stored attention evaluation values of each collection time point, the signal processing device can obtain the historical attention evaluation values of the multiple historical collection time points in the target period. If the attention evaluation value of a collection time point is greater than the attention threshold, the category of the predicted attention label corresponding to the collection time point is determined to be a label category representing attention concentration, otherwise the category of the predicted attention label corresponding to the collection time point is determined to be a label category representing attention dispersion. The attention threshold can be set according to specific requirements, for example, the attention threshold can be set to 0.5, 0.6, etc. The duration ratio of the attention evaluation value greater than the attention threshold is determined based on the attention evaluation value of the current collection time point and the historical attention evaluation values of the multiple historical collection time points, which aims to determine the duration ratio of the detection period when the predicted attention label indicates attention concentration, i.e., to determine the duration ratio of the detection period when the target object is predicted to be in the attention concentration state.

[0077] In a possible implementation, in the process of determining the attention concentration level according to the duration ratio by the signal processing device, one attention concentration level can be determined from the preset multiple attention concentration levels according to the duration ratio, wherein the attention concentration degree indicated by the determined attention concentration level is positively correlated with the duration ratio. In a specific implementation, the signal processing device can obtain the duration ratio range corresponding to each preset attention concentration level, and determine the corresponding attention concentration level according to the duration ratio range to which the determined duration ratio belongs. The duration ratio range corresponding to each preset attention concentration level can be set according to specific requirements, and the embodiments of the present application will not be limited. For example, in one application scenario, the multiple preset attention concentration levels are the first level and the second level in the order of low to high attention concentration degree, wherein the first level specifically represents attention dispersion (i.e., attention dispersion), and the corresponding duration ratio range is set to [0, 50%); the second level specifically represents attention concentration, and the corresponding duration ratio range is set to (50%, 100%]. For another example, in another application scenario, the duration ratio range corresponding to the first level is set to [0, 70%); the duration ratio range corresponding to the second level is set to (70%, 100%].

[0078] In a possible embodiment, when the signal processing device generates a task processing result of an attention detection task based on the determined attention concentration level, the generated task processing result may include the attention concentration level; optionally, the task processing result may also include attention information of the detection period, and the attention information of the detection period may be used to indicate the proportion of the time period in which the target object is predicted to be in a state of concentration during the detection period, and may be used to indicate the proportion of the time period in which the target object is predicted to be in a state of distraction during the detection period; the specific format of the attention information of the detection period is not limited in the embodiment of the present application. For example, if the proportion of the time period in which the target object is predicted to be in a state of concentration during the detection period is 91%, and the proportion of the time period in which the target object is predicted to be in a state of distraction during the detection period is 9%, exemplarily, a detection period of attention information output in a text format may be "the proportion of the time period in which the detection period is in a state of concentration is: 91%, and the proportion of the time period in which the detection period is in a state of distraction is: 9%"; exemplarily, a detection period of attention information output in a graphical format may be as follows Figure 3b As shown, it is specifically a pie chart. Of course, the attention information of the detection period can also be represented based on other chart formats such as bar charts and line charts, and this embodiment of the application does not limit it. Optionally, the task processing result can also include the attention evaluation value of the detection period, wherein the attention evaluation value of the detection period can be represented based on various formats such as text and charts, and this embodiment of the application does not limit it. Please refer to Figure 3c , which is a schematic diagram of an attention evaluation value for a detection period provided in an embodiment of the present application, wherein the attention evaluation value is used to indicate the probability of attention concentration, and the closer the attention evaluation value is to 1, the higher the degree of attention concentration; it can be known that the signal processing device also supports outputting the attention evaluation value of any historical period, and information such as the predicted proportion of time that the target object is in a concentrated or distracted state for the corresponding historical period. Of course, the task processing result can also carry feedback information or suggestion information applicable to the predicted attention concentration level to help the target object improve its attention state and improve work or study efficiency.

[0079] The above shows the related processing process that when the attention evaluation value is used to indicate the probability of attention concentration, the time length ratio of the detection period corresponding to the prediction that the target object is in the attention concentration state is determined, and then the attention concentration level is determined according to the time length ratio, and the task processing result of the attention detection task is generated based on the determined attention concentration level; in another possible implementation, the attention evaluation value can also be set to represent the probability of attention dispersion. Based on this, the data processing device can take the ratio of the statistical number corresponding to the attention label of the label category used to represent attention dispersion to the target number as the attention evaluation value of the current collection time point, which is used to indicate the probability of attention dispersion of the target object at the current collection time point. In the subsequent process of determining the time length ratio of the detection period corresponding to the prediction that the target object is in the attention concentration state, the time length ratio of the attention evaluation value used to indicate the probability of attention dispersion less than a specific attention threshold value is determined as the time length ratio of the detection period corresponding to the prediction that the target object is in the attention concentration state. The specific attention threshold value can be set according to specific needs, and the embodiments of the present application are not limited. Please refer to Figure 3d Another schematic diagram of the attention evaluation value of the detection period provided by the embodiments of the present application is shown, wherein the attention evaluation value is used to indicate the probability of attention dispersion, and the closer the attention evaluation value is to 1, the higher the degree of attention dispersion, that is, the lower the degree of attention concentration.

[0080] In a possible implementation, the signal processing device supports real-time acquisition of the brain electrical signals collected by the detection object in the brain region of the target object, and processing and real-time output of the attention evaluation values of the target object at the acquisition time points. For example, the signal processing device can output the attention evaluation values of the detection object at the acquisition time points in a time period in a chart format and display the attention evaluation values on the interactive interface of the signal processing device. In another possible implementation, the signal processing device supports acquisition of the brain electrical signals collected by the detection object in a time period in the brain region of the target object, and processing and output of the attention evaluation values of the target object at the acquisition time points in the time period. For example, the detection object can first collect the brain electrical signals of the target object in a time period, and store the collected brain electrical signals in the time period. After the signal processing device acquires the brain electrical signals in the time period, the signal processing device processes and outputs the attention evaluation values of the target object at the acquisition time points in the time period. Optionally, the signal processing device also supports viewing of the brain electrical signals in the time period (which can be output in a waveform chart format for viewing), and supports viewing of intermediate data generated in the process of processing the attention evaluation values of the target object at the acquisition time points in the time period. For example, the signal processing device supports viewing of preprocessed data, which can include preprocessed N sets of brain electrical signals in the time period, the signal features of the N sets of brain electrical signals in the time period obtained based on a feature conversion process, and the brain electrical signal features of the N sets of brain electrical signals in the time period obtained based on a feature extraction process (which can be output in a waveform chart format for viewing). The signal processing device also supports viewing of the task processing result in the time period, which can include the attention evaluation values of the target object at the acquisition time points in the time period (which can be output in a chart format for viewing), the proportion of the time length in the time period during which the target object is predicted to be in the attention concentrated state, the attention concentration level for indicating the attention concentration degree in the time period determined based on the proportion of the time length in the time period during which the target object is predicted to be in the attention concentrated state, and other information.

[0081] See Figure 3eA schematic diagram of an interactive interface provided by an embodiment of the present application can include a data import component (shown as "file import"), a preprocessed data viewing component (shown as "view preprocessed data"), a feature data viewing component (shown as "view feature data"), and a task processing result viewing component (shown as "view task processing result"). Triggering the data import component can import a period of stored electroencephalogram signals into a signal processing device, i.e., the signal processing device can obtain a period of electroencephalogram signals and perform the related processes of attention detection proposed by the embodiment of the present application after obtaining the electroencephalogram signals of the period to obtain the task processing result. Triggering the preprocessed data viewing component, the signal processing device can output preprocessed data for viewing. Triggering the feature data viewing component, the signal processing device can output feature data for viewing. Triggering the processing result viewing component, the signal processing device can output the task processing result for viewing. Optionally, if the types of data that can be viewed include multiple types, the signal processing device can output the multiple types of data that can be viewed together, or can display data options of the types of data that can be viewed after triggering the related components for data viewing, and output the data of the type corresponding to the data option only when the data option is triggered. For example, please refer to Figure 3f A schematic diagram of viewing feature data provided by an embodiment of the present application can display data options of the types of data that can be viewed after the feature data viewing component is triggered. If the data option corresponding to the power spectral density moving average feature is triggered, the processed power spectral density moving average feature can be displayed (which can be displayed in the form of a chart).

[0082] In the embodiments of the present application, the N groups of brain electrical signals collected by the detection object in the corresponding brain region of the target object can be obtained according to the target brain electrical acquisition rule corresponding to the attention detection task, and then the task processing result containing the attention concentration level can be determined based on the brain electrical signals, and the attention concentration level can indicate the degree of attention concentration of the target object. The attention detection process is not affected by environmental factors such as light, and the accuracy of the attention concentration level obtained by attention detection is high, that is, the accuracy of attention detection is high, the stability is high, and the application range is wide. Moreover, in the process of attention detection based on brain electrical signals, the target source signal associated with the attention detection task can be extracted from the obtained brain electrical signals, and then the attention detection operation corresponding to the attention detection task can be performed according to the extracted target source signal associated with the attention detection task, and the task processing result containing the attention concentration level is obtained. Compared with the task processing result obtained by directly performing the attention detection operation corresponding to the attention detection task according to the obtained brain electrical signals, the accuracy of the task processing result can be improved, that is, the accuracy of the obtained attention concentration level can be improved, and the accuracy of attention detection can be improved.

[0083] Based on the above description, another signal processing method is provided in the embodiments of the present application, see Figure 4 The flowchart of another signal processing method provided in the embodiments of the present application is shown in the figure; the signal processing method can be executed by a signal processing device, and the signal processing method can include the following steps S401-S410:

[0084] S401, obtaining a sample feature data set.

[0085] The sample feature data set includes a plurality of sample feature data, and the sample feature data set is determined in the training process of the attention detection model and includes brain electrical signal features of the N groups of brain electrical signals for training.

[0086] The sample feature data set is determined in the training process of the attention detection model, in other words, the sample feature data set is a feature data set used in the training process of the attention detection model, which includes a plurality of sample feature data, and one sample feature data includes brain electrical signal features of N groups of brain electrical signals for training. For example, the brain electrical signal features of N groups of brain electrical signals of one sample object at one collection time point can be: the brain electrical signal features of N groups of brain electrical signals of one sample object at one collection time point, and the corresponding attention label can be annotated; wherein the related process of determining the brain electrical signal features of N groups of brain electrical signals of one sample object at one collection time point is similar to the related process of determining the brain electrical signal features of N groups of brain electrical signals of the target object at the current collection time point, and will not be described here.

[0087] Exemplarily, N sets of electroencephalogram signals of the sample object in the brain regions indicated by the target electroencephalogram acquisition rule can be collected in a time period in which the sample object is in the state of attention concentration, and the brain electroencephalogram signal features determined based on the N sets of electroencephalogram signals collected at any one of the collection time points in the time period in which the sample object is in the state of attention concentration can be taken as the brain electroencephalogram signal features of the N sets of electroencephalogram signals for training, and the category of the attention label corresponding thereto can be labeled as the label category for indicating attention concentration. Exemplarily, N sets of electroencephalogram signals of the sample object in the brain regions indicated by the target electroencephalogram acquisition rule can be collected in a time period in which the sample object is in the state of attention dispersion, and the brain electroencephalogram signal features determined based on the N sets of electroencephalogram signals collected at any one of the collection time points in the time period in which the sample object is in the state of attention dispersion can be taken as the brain electroencephalogram signal features of the N sets of electroencephalogram signals for training, and the category of the attention label corresponding thereto is the label category for indicating attention dispersion. Based on this, the brain electroencephalogram signals of one or more sample objects can be collected and processed to obtain the brain electroencephalogram signal features, and then the sample feature dataset can be constructed. Exemplarily, when N sets of electroencephalogram signals of the sample object in the brain regions indicated by the target electroencephalogram acquisition rule are collected in a time period in which the sample object is in the state of attention concentration, the sample object can be made to carefully listen to a piece of speech content in a quiet environment, and the electroencephalogram signals of the sample object in the process of listening to the speech content are collected. When N sets of electroencephalogram signals of the sample object in the brain regions indicated by the target electroencephalogram acquisition rule are collected in a time period in which the sample object is in the state of attention dispersion, the sample object can be made to listen to multiple pieces of speech content in a quiet environment at the same time, and perform other operations (for example, playing games) at the same time, and the electroencephalogram signals of the sample object in the process are collected.

[0088] S402, perform H times of data division processing on the sample feature dataset to obtain H data division results.

[0089] One data division result includes a training dataset and a validation dataset divided from the sample feature dataset, the sample feature data in the training dataset is taken as training data, and the sample feature data in the validation dataset is taken as validation data. H is a positive integer, which can be set according to specific requirements. For example, H can be set to 5, 10, etc. The embodiments of the present application are not limited.

[0090] The H data partition results obtained by the signal processing device performing H times of data partition processing on the sample feature data set should be different from each other. In a possible implementation, in the process of obtaining the H data partition results by performing H times of data partition processing on the sample feature data set, the signal processing device can perform random partitioning. In another possible implementation, in the process of obtaining the H data partition results by performing H times of data partition processing on the sample feature data set, the signal processing device can perform equal division processing on the sample feature data set to obtain H sample feature data subsets, and the signal processing device can take one of the H sample feature data subsets as a validation data set and take the remaining H-1 sample feature data subsets together as a training data set to obtain one data partition result, until the H sample feature data subsets are taken as validation data sets respectively to obtain the H data partition results.

[0091] S403, determining, according to the M reference quantities and the feature similarity between the validation data in the validation data set and the training data in the training data set in the H data partition results, a prediction effect evaluation value of predicting the attention label based on different data partition results under each reference quantity.

[0092] M is a positive integer, and the M reference quantities can be set according to specific requirements, which are not limited by the embodiments of the application.

[0093] In a possible implementation, in the process of determining, according to the M reference quantities, the prediction effect evaluation value of predicting the attention label based on different data partition results under each reference quantity, for the mth reference quantity in the M reference quantities and the hth data partition result in the H data partition results, where m ∈ [1, M] and h ∈ [1, H], for the sake of description, the training data set in the hth data partition result is referred to as the hth training data set, and the validation data set in the hth data partition result is referred to as the hth validation data set; the signal processing device can train the attention detection model to be trained by using the mth reference quantity and the hth training data set, where for the KNN model, determining the data set (here, the hth training data set) and the K value (here, the mth reference quantity) used by the KNN model can be regarded as the training process of the KNN model, in other words, the signal processing device can construct a trained attention detection model by using the mth reference quantity, the hth training data set, and the attention detection model to be trained; the trained attention detection model can predict the attention label of each validation data in the hth validation data set based on the mth reference quantity and the hth training data set; the related process of predicting the attention label by using the trained attention detection model is similar to the related process of predicting the attention evaluation value by using the attention detection model.

[0094] For any verification data, the signal processing device can call the trained attention detection model to determine the feature similarity between the verification data and each training data in the hth training data set; according to the feature similarity, filter out the mth reference number of training data from the hth training data set; wherein the feature similarity between any filtered training data and the verification data is greater than or equal to the feature similarity between other training data not filtered and the verification data; obtain the attention label corresponding to each filtered training data, and statistically obtain the statistical quantity of the training data corresponding to each type of attention label; according to the statistical quantity corresponding to each type of attention label, determine the attention evaluation value corresponding to the verification data, and the attention evaluation value is used to indicate the probability of attention concentration; if the attention evaluation value corresponding to the verification data is greater than the attention threshold, it can be determined that the category of the predicted attention label corresponding to the verification data is: a label category used to represent attention concentration, otherwise, it is determined that the category of the predicted attention label corresponding to the verification data is: a label category used to represent attention dispersion. Based on this, the trained attention detection model can predict the attention label of each verification data in the hth verification data set based on the mth reference number and the hth training data set, and then can determine the prediction effect evaluation value when predicting the attention label based on the hth data division result under the mth reference number based on the predicted attention label corresponding to each verification data in the hth verification data set and the corresponding labeled attention label. Optionally, the prediction effect evaluation value can be selected according to specific needs, for example, accuracy, AUC (a model evaluation index), F1 score (a model evaluation index), etc. The subsequent embodiments of the present application take accuracy as an example. Based on the above description, the signal processing device can determine the prediction effect evaluation value when predicting the attention label based on different data division results under each reference number according to the M reference numbers and the feature similarity between the verification data in the verification data set and the training data in the training data set in the H data division results.

[0095] S404, average processing the prediction effect evaluation value when predicting the attention label based on different data division results under each reference number to obtain the average prediction effect evaluation value under each reference number.

[0096] In a possible implementation, when the signal processing device averages the prediction effect evaluation values of predicting the attention label based on different data division results under each reference quantity, taking the mth reference quantity as an example, the signal processing device can average the prediction effect evaluation values of predicting the attention label based on the 1st data division result to the prediction effect evaluation values of predicting the attention label based on the Hth data division result under the mth reference quantity, to obtain the average prediction effect evaluation value under the mth reference quantity.

[0097] S405, taking the reference quantity corresponding to the target average prediction effect evaluation value as the target quantity.

[0098] The target average prediction effect evaluation value is the optimal value in the average prediction effect evaluation values, that is, the target average prediction effect evaluation value is the value with the optimal prediction effect indicated in the average prediction effect evaluation values. The average prediction effect evaluation value under any reference quantity can measure the overall prediction effect of the reference quantity in different verification data sets, and the target quantity determined based on this is more accurate. For example, when the prediction effect evaluation value is the accuracy rate, and the average prediction effect evaluation value is the average accuracy rate, the higher the accuracy rate, the better the prediction effect, and therefore the target average prediction effect evaluation value is the maximum value in the average prediction effect evaluation values, that is, the maximum average accuracy rate. When the prediction effect evaluation value is the error rate, and the average prediction effect evaluation value is the average error rate, the lower the error rate, the better the prediction effect, and therefore the target average prediction effect evaluation value is the minimum value in the average prediction effect evaluation values, that is, the minimum average error rate. For example, please refer to Figure 5a FIG. 1 is a schematic diagram of determining a target quantity provided by an embodiment of the present application. If the M reference quantities are set as integers in the interval [1, 50], as shown in the curve diagram 501, the average prediction effect evaluation value under each reference quantity, that is, the average error rate under each reference quantity, is shown. It can be known that the average error rate corresponding to the reference quantity 10 is the minimum, and therefore the target quantity can be determined as 10.

[0099] In a possible implementation, after determining the target quantity, the signal processing device can divide the sample feature data set to obtain a reference feature data set, wherein the sample feature data in the reference feature data set is used as reference feature data. The target quantity, the reference feature data set, and the attention detection model to be trained are used to construct a trained attention detection model. The trained attention detection model can predict the attention evaluation value based on the target quantity and each reference feature data in the reference feature data set. The related process has been described in the above embodiment, and will not be described here.

[0100] For the process of dividing the sample feature dataset to obtain the reference feature dataset, in a possible implementation, the signal processing device can take the sample feature dataset as the reference feature dataset; in another possible implementation, the signal processing device can select part of the sample feature data from the sample feature dataset as the reference feature data to construct the reference feature dataset. For example, the signal processing device can randomly select any data partition result corresponding training dataset as the reference feature dataset, and for another example, the signal processing device can select the training dataset corresponding to the optimal prediction effect evaluation value in the prediction effect evaluation values of different data partition results based on the prediction of the attention label under the target number.

[0101] The above-mentioned process of determining the target number and constructing the attention detection model, i.e., the cross-validation process, determines the prediction effect of different reference numbers by using the training dataset and the validation dataset after multiple divisions of the sample feature dataset. The sample feature dataset can be fully utilized, so that the target number with excellent prediction effect can be determined in the case of insufficient data amount of the sample feature dataset, for predicting the subsequent attention evaluation value. In another possible implementation, in the case of sufficient data amount of the sample feature dataset, the sample feature dataset can be divided once to obtain a data partition result, which includes a training dataset, a validation dataset and a test dataset, wherein the sample feature data in the training dataset is taken as training data, the sample feature data in the validation dataset is taken as validation data, and the sample feature data in the test dataset is taken as test data; the signal processing device can determine the prediction effect evaluation value of predicting the attention label based on the data partition result under each reference number according to the M reference numbers and the feature similarity between the validation data of the validation dataset and the training data of the training dataset, and take the reference number corresponding to the prediction effect evaluation value indicating the optimal prediction effect in the prediction effect evaluation values as the candidate number.

[0102] The signal processing device can adopt the candidate number, the training data set and the attention detection model to be trained to construct a trained attention detection model, which can predict the attention label based on the candidate number and the training data set; the signal processing device can adopt the trained attention detection model to predict the attention label of each test data in the test data set, and determine the model evaluation value based on the predicted attention label and the annotated attention label corresponding to each test data, and in the case that the model evaluation value reaches a preset condition, the candidate number is determined as the target number and the training data set is determined as the reference feature data set, and the trained attention detection model can be used for subsequent prediction of the attention evaluation value; in the case that the model evaluation value does not reach the preset condition, a reference number can be selected again as a new candidate number in the order of decreasing prediction effect indicated by each prediction effect evaluation value, and a new model evaluation value is determined in the test data set based on the new candidate number, to judge whether the new model evaluation value reaches the preset condition, until the candidate number when the new model evaluation value reaches the preset condition is determined as the target number; optionally, the model evaluation value can be selected according to specific needs, for example, one or more of accuracy, AUC and F1 score can be selected, and the present application embodiment does not limit the model evaluation value, and the preset condition can be set according to specific needs, and the present application embodiment does not limit the model evaluation value, for example, when the model evaluation value selects the accuracy, the model evaluation value reaching the preset condition can mean that the model evaluation value is greater than or equal to a preset accuracy threshold (the accuracy threshold can be set according to specific needs).

[0103] In yet another possible implementation, see Figure 5bA schematic diagram for determining an attention detection model is provided for an embodiment of the present application. The signal processing device can obtain an electroencephalogram signal, and process the electroencephalogram signal to obtain an electroencephalogram signal feature corresponding to the electroencephalogram signal through pre-processing, feature extraction and other operations. Then, a sample feature dataset can be constructed based on the processed electroencephalogram signal feature. In this process, the signal processing device obtains N groups of electroencephalogram signals of one or more sample objects at different collection time points. For one obtained N group of electroencephalogram signals, the corresponding processed electroencephalogram signal feature is used as a sample feature data. The attention label corresponding to the sample feature data can be obtained by labeling. After obtaining the sample feature dataset, the signal processing device can divide the sample feature dataset to obtain a data division result, which includes a training dataset and a test dataset. The signal processing device can select model parameters, and train the attention detection model based on the selected model parameters and the training dataset. When the attention detection model uses a KNN model, the model parameters are selected from M reference quantities. The trained attention detection model is trained based on the selected model parameters and the training dataset, that is, the selected reference quantity, the training dataset and the trained attention detection model are used to construct the trained attention detection model.

[0104] The signal processing device can test the trained attention detection model based on the test dataset. If the test result meets the expectation, the trained attention detection model can be saved for subsequent prediction of attention evaluation value. Otherwise, the model parameters are reselected for model training until the test result of the newly trained attention detection model based on the test dataset meets the expectation, and the corresponding trained attention detection model is saved. The process of testing the trained attention detection model based on the test dataset by the signal processing device is similar to the process of determining the model evaluation value based on the test dataset. The process of determining whether the test result meets the expectation, that is, the process of determining whether the model evaluation value meets the preset condition, is not described in detail.

[0105] The above describes the determination process of the final model for predicting the attention evaluation value when the attention detection model uses a KNN model. When the attention detection model uses other model structures (such as a neural network model), similar processes can be performed, such as dividing the sample feature dataset into a training dataset for model training, dividing the sample feature dataset into a validation dataset for model validation to determine appropriate model parameters (such as determining appropriate model hyperparameters), dividing the sample feature dataset into a test dataset for model testing to ensure the generalization performance of the trained model, and the like.

[0106] S406, in response to the task request about the attention detection task, determining a target electroencephalogram acquisition rule corresponding to the attention detection task according to a correspondence between the task and the electroencephalogram acquisition rule.

[0107] The electroencephalogram acquisition rule is used to indicate a brain region where the electroencephalogram signal needs to be acquired, and different tasks correspond to different electroencephalogram acquisition rules.

[0108] S407, acquiring N groups of electroencephalogram signals collected by the detection object in a brain region corresponding to the target electroencephalogram acquisition rule according to the target electroencephalogram acquisition rule; N is a positive integer.

[0109] S408, pre-processing each group of electroencephalogram signals respectively to obtain a plurality of source signals of each group of electroencephalogram signals in the N groups of electroencephalogram signals.

[0110] S409, determining a target source signal associated with the attention detection task from the plurality of source signals of each group of electroencephalogram signals based on signal characteristics of the plurality of source signals.

[0111] S410, performing an attention detection operation corresponding to the attention detection task according to the target source signal corresponding to each group of electroencephalogram signals to obtain a task processing result of the attention detection task.

[0112] The task processing result includes an attention concentration level. The related processes of steps S406 to S410 have been described in the above Figure 2 corresponding embodiments, and will not be repeated here. In one possible implementation, the S406 is an optional step, that is, the S407 can be started without initiating the task request, for example, when a detection period corresponding to the attention detection task arrives, or when a preset time period for performing the attention detection task arrives, the S407 is directly started.

[0113] In a possible implementation, in a case where the attention detection task is for a target object, i.e., the N groups of brain electrical signals are collected from the brain region of the target object, the signal processing device can also jointly predict the attention concentration level of the target object based on other object features of the target object to generate the task processing result of the attention detection task. For example, in an optional implementation, the signal processing device can further perform the following process: upon receiving the task request, obtaining a first object feature and a second object feature of the target object; wherein the first object feature and the second object feature include any one or more of a facial feature of the target object, a heart rate feature of the target object, and a blood oxygen feature of the target object, and the first object feature and the second object feature are different; predicting the attention concentration level based on the first object feature of the target object and the second object feature of the target object respectively; if the attention concentration levels predicted respectively are different, triggering the step of determining the target brain electrical acquisition rule corresponding to the attention detection task according to the correspondence between the task and the brain electrical acquisition rule.

[0114] The facial feature of the target object refers to information that can reflect the facial characteristics of the target object, for example, can include but is not limited to the facial image of the target object, the heart rate feature of the target object refers to information that can reflect the heart rate condition of the target object, for example, can include but is not limited to one or more of the heart rate value of the target object, heart rate variability (HRV) data and other information of the target object, wherein the HRV refers to the change of the heart rate interval, which is considered to be related to the activity of the autonomic nervous system, and can reflect the attention state of the individual, so the HRV data can be used to predict the attention concentration level; the blood oxygen feature of the target object refers to information that can reflect the blood oxygen condition of the target object, for example, can include but is not limited to the blood oxygen value of the target object, optionally, the blood oxygen feature of the target object can be measured by functional near-infrared spectroscopy (fNIRS) technology, fNIRS is a non-invasive brain imaging technology, which can reflect neural activity by measuring the blood oxygen level change of the cerebral cortex, and can provide similar time resolution to electroencephalogram, but has better spatial resolution. When the signal processing device predicts the attention concentration level based on the first object feature of the target object, a first model can be called for implementation, the first model refers to a model that supports predicting the attention concentration level by using the first object feature, and the model structure thereof can be selected according to specific needs, for example, a neural network model, a deep learning model and the like can be selected, and the embodiments of the present application do not make any limitation, the first model can be trained by using the first object feature of one or more sample objects; when the signal processing device predicts the attention concentration level based on the second object feature of the target object, a second model can be called for implementation, the second model refers to a model that supports predicting the attention concentration level by using the second object feature, and the model structure thereof can be selected according to specific needs, for example, a neural network model, a deep learning model and the like can be selected, and the embodiments of the present application do not make any limitation, the second model can be trained by using the second object feature of one or more sample objects.

[0115] For example, in another optional implementation, in the process of performing the attention detection operation corresponding to the attention detection task according to the target source signal corresponding to each group of electroencephalogram signals to obtain the task processing result of the attention detection task, after the electroencephalogram signal features of each group of electroencephalogram signals are extracted based on the target source signal corresponding to each group of electroencephalogram signals, the signal processing device can predict the attention concentration level based on the electroencephalogram signal features of the target object (i.e., the extracted electroencephalogram signal features of each group of electroencephalogram signals) and the specific object features of the target object, and generate the task processing result based on the predicted attention concentration level; wherein the specific object features include one or more of the first object features and the second object features, and when the signal processing device predicts the attention concentration level based on the electroencephalogram signal features of the target object and the specific object features of the target object, a third model can be called for implementation, the third model refers to a model that supports predicting the attention concentration level using electroencephalogram signal features and specific object features and the like multi-modal features, and the model structure thereof can be selected according to specific needs, for example, a neural network model, a deep learning model, and the like can be selected, and the present application embodiment is not limited, and the third model can be trained together using the electroencephalogram signal features and the specific object features of one or more sample objects.

[0116] The signal processing method proposed in the embodiments of the present application can be applied to predict the attention concentration level of a target object in various application scenarios, for example, it can be applied to education, transportation, medical treatment and the like, to help improve learning efficiency, work efficiency and safety, and it can also be applied to game, entertainment, health management and the like to provide better user experience. In a possible implementation, in the case that the attention detection task is for a target object, that is, the N groups of electroencephalogram signals are collected from the brain region of the target object, the signal processing device can further perform the following process: determining the working scenario of the target object; if the working scenario is a mobile control scenario under the first control strategy, determining whether the attention concentration degree indicated by the attention concentration level in the task processing result meets the switching condition; if the switching condition is met, switching the control strategy of the mobile object from the first control strategy to the second control strategy, and controlling the mobile object using the second control strategy; wherein the attention concentration degree indicated by the attention concentration level in the task processing result meets the switching condition includes being lower than the attention concentration degree indicated by the attention concentration level threshold.

[0117] The mobile object can be a movable device driven by the target object, and the movable device can include, but is not limited to, a vehicle, a ship, a manned aircraft, and the like. The signal processing device can also support control of the mobile object. For example, when the mobile object is a vehicle, the signal processing device can be an intelligent vehicle installed in the vehicle. The first control strategy includes a control strategy that requires manual driving, for example, a control strategy that indicates complete manual driving, a control strategy that indicates assisted manual driving, and the like. The second control strategy includes a control strategy that does not require manual driving, for example, a control strategy that indicates automatic driving. That is, if the mobile object is controlled by the target object (i.e., driven by the target object) under the first control strategy, the signal processing device can switch the control strategy of the mobile object from the first control strategy to the second control strategy and control the mobile object using the second control strategy if the attention concentration level indicated by the attention concentration degree in the task processing result is lower than the attention concentration level threshold. That is, the automatic driving of the mobile object can be implemented, and the automatic switching of the control strategy of the mobile object can be implemented when the target object has a low attention concentration degree, which can reduce the possibility of traffic accidents caused by the low attention concentration degree of the target object and improve driving safety. Optionally, the attention concentration level threshold can be set according to specific requirements. For example, if the preset multiple attention concentration levels are the first level and the second level in order of attention concentration degree from low to high, the first level specifically represents inattention (i.e., distraction), and the second level specifically represents attention, the attention concentration level threshold can be set to the second level.

[0118] In the embodiments of the present application, the prediction effects of different reference quantities can be determined by multiple divisions of the sample feature data set, and the training data set and the verification data set after the multiple divisions are used to determine the prediction effects of different reference quantities, and then the reference quantity with the best prediction effect is selected as the target quantity, and the attention detection model for predicting the attention evaluation value in the attention detection process is constructed based on the target quantity, so that the prediction effect of the attention detection model based on the target quantity is good, and the sample feature data set is fully utilized by the multiple divisions of the sample feature data set, so that the target quantity with excellent prediction effect can be determined even if the data amount of the sample feature data set is insufficient.

[0119] Based on the description of the method embodiments, the present application also discloses a signal processing device. The signal processing device can be a computer program running in a computer device, which can be the signal processing device described above. The signal processing device can perform each step in the method flow shown in Figure 2 or Figure 4 Please refer to Figure 6A structural schematic diagram of a signal processing device provided by an embodiment of the present application can include a communication unit 601 and a processing unit 602, wherein:

[0120] The communication unit 601 is configured to perform communication interaction.

[0121] The processing unit 602 is configured to acquire N groups of electroencephalogram signals collected by a detection object in a target brain electroencephalogram acquisition rule corresponding to a brain region according to the target brain electroencephalogram acquisition rule; N is a positive integer; each group of electroencephalogram signals is preprocessed to obtain a plurality of source signals of each group of electroencephalogram signals in the N groups of electroencephalogram signals; a target source signal associated with an attention detection task is determined from the plurality of source signals of each group of electroencephalogram signals based on signal characteristics of the plurality of source signals; an attention detection operation corresponding to the attention detection task is performed according to the target source signal corresponding to each group of electroencephalogram signals to obtain a task processing result of the attention detection task; the task processing result includes an attention concentration level.

[0122] In an embodiment, when the processing unit 602 is configured to preprocess each group of electroencephalogram signals to obtain a plurality of source signals of each group of electroencephalogram signals in the N groups of electroencephalogram signals, the processing unit 602 can be specifically configured to:

[0123] Filter processing is performed on the i th group of electroencephalogram signals in the N groups of electroencephalogram signals to obtain an intermediate signal corresponding to the i th group of electroencephalogram signals; wherein i ∈ [1, N];

[0124] Blind source separation processing is performed on the intermediate signal corresponding to the i th group of electroencephalogram signals to obtain a plurality of source signals of the i th group of electroencephalogram signals;

[0125] When the processing unit 602 is configured to determine a target source signal associated with an attention detection task from a plurality of source signals of each group of electroencephalogram signals based on signal characteristics of the plurality of source signals, the processing unit 602 can be specifically configured to:

[0126] Feature conversion processing is performed on the plurality of source signals of the i th group of electroencephalogram signals to obtain a plurality of signal characteristics of the i th group of electroencephalogram signals;

[0127] The target source signal associated with the attention detection task is determined from the plurality of source signals of the i th group of electroencephalogram signals according to the plurality of signal characteristics of the i th group of electroencephalogram signals;

[0128] The feature conversion processing includes differential entropy linear dynamic system smoothing processing, and the corresponding signal characteristic is a differential entropy linear dynamic system smoothing characteristic;

[0129] Or the feature conversion processing includes differential entropy moving average processing, and the corresponding signal characteristic is a differential entropy moving average characteristic;

[0130] Or the feature conversion processing includes power spectrum density linear dynamic system smoothing processing, and the corresponding signal feature is a power spectrum density linear dynamic system smoothing feature.

[0131] Or the feature conversion processing includes power spectrum density moving average processing, and the corresponding signal feature is a power spectrum density moving average feature.

[0132] In an implementation, the processing unit 602, when used to perform attention detection operations corresponding to the attention detection task according to the target source signal corresponding to each group of electroencephalogram signals, can be specifically used to:

[0133] Respectively performing feature extraction processing on the target source signal corresponding to each group of electroencephalogram signals to obtain the electroencephalogram signal feature of each group of electroencephalogram signals;

[0134] Calling the attention detection model to perform attention analysis on the electroencephalogram signal feature of each group of electroencephalogram signals to obtain an attention evaluation value at a current acquisition time point; wherein the current acquisition time point is the acquisition time point of the N groups of electroencephalogram signals, and the attention evaluation value is used to indicate the probability of attention concentration;

[0135] Generating the task processing result of the attention detection task based on the attention evaluation value at the current acquisition time point and a historical attention evaluation value determined at a target period before the current acquisition time point.

[0136] In an implementation, the processing unit 602, when used to call the attention detection model to perform attention analysis on the electroencephalogram signal feature of each group of electroencephalogram signals to obtain an attention evaluation value at a current acquisition time point, can be specifically used to:

[0137] Calling the attention detection model to determine the feature similarity between the target feature data and each reference feature data in the reference feature data set; wherein the target feature data includes the electroencephalogram signal feature of the N groups of electroencephalogram signals, and the reference feature data set is determined in the training process of the attention detection model and includes the electroencephalogram signal feature of the N groups of electroencephalogram signals for training;

[0138] According to the feature similarity, screening out a target number of reference feature data from the reference feature data set; wherein the feature similarity between any reference feature data screened out and the target feature data is greater than or equal to the feature similarity between other reference feature data not screened out and the target feature data;

[0139] Obtaining the attention label corresponding to each reference feature data screened out and statistically obtaining the statistical number of reference feature data corresponding to each type of attention label; wherein the types of attention labels include: a label type used to represent attention concentration and a label type used to represent attention dispersion.

[0140] According to the statistical quantity corresponding to each type of attention label, an attention evaluation value at the current collection time point is determined.

[0141] In an implementation, the processing unit 602 can be specifically configured to:

[0142] obtain historical attention evaluation values of a plurality of historical collection time points in a target period;

[0143] determine, according to the attention evaluation value at the current collection time point and the historical attention evaluation values of the plurality of historical collection time points, a time length proportion in which the attention evaluation value is greater than an attention threshold value;

[0144] determine an attention concentration level according to the time length proportion, and generate a task processing result of the attention detection task based on the determined attention concentration level; wherein the attention concentration degree indicated by the determined attention concentration level is positively correlated with the time length proportion.

[0145] In an implementation, the processing unit 602 is further configured to:

[0146] obtain a sample feature data set; wherein the sample feature data set includes a plurality of sample feature data, and the sample feature data set is determined in a training process of the attention detection model and includes brain electrical signal features of N groups of brain electrical signals for training;

[0147] perform H times of data division processing on the sample feature data set to obtain H data division results; wherein one data division result includes one training data set and one validation data set divided from the sample feature data set, the sample feature data in the training data set is used as training data, the sample feature data in the validation data set is used as validation data, and H is a positive integer;

[0148] determine, according to M reference quantities and feature similarities between validation data in the validation data set and training data in the training data set in the H data division results, a prediction effect evaluation value of predicting the attention label based on different data division results under each reference quantity; M is a positive integer;

[0149] perform average processing on the prediction effect evaluation values of predicting the attention label based on different data division results under each reference quantity to obtain an average prediction effect evaluation value under each reference quantity;

[0150] The reference quantity corresponding to the target average prediction effect evaluation value is taken as a target quantity; wherein the target average prediction effect evaluation value is the optimal value among the average prediction effect evaluation values.

[0151] In an embodiment, the processing unit 602 is further configured to, before acquiring the N groups of electroencephalogram signals collected by the target object in the brain region corresponding to the target electroencephalogram acquisition rule according to the target electroencephalogram acquisition rule:

[0152] In response to a task request about the attention detection task, the target electroencephalogram acquisition rule corresponding to the attention detection task is determined according to the correspondence between the task and the electroencephalogram acquisition rule.

[0153] The electroencephalogram acquisition rule is used to indicate the brain region from which the electroencephalogram signal needs to be acquired, and different tasks correspond to different electroencephalogram acquisition rules.

[0154] In an embodiment, the N groups of electroencephalogram signals are collected from the brain region of the target object; the processing unit 602 is further configured to:

[0155] Upon receiving the task request, the first object feature and the second object feature of the target object are acquired; wherein the first object feature and the second object feature include any one or more of the following: facial features of the target object, heart rate features of the target object, and blood oxygen features of the target object, and the first object feature and the second object feature are different.

[0156] The attention concentration level is predicted based on the first object feature of the target object and the second object feature of the target object, respectively.

[0157] If the attention concentration levels predicted respectively are different, the step of determining the target electroencephalogram acquisition rule corresponding to the attention detection task according to the correspondence between the task and the electroencephalogram acquisition rule is triggered.

[0158] In an embodiment, the N groups of electroencephalogram signals are collected from the brain region of the target object; the processing unit 602 is further configured to:

[0159] The working scenario of the target object is determined.

[0160] If the working scenario is a mobile control scenario under the first control strategy, it is determined whether the attention concentration degree indicated by the attention concentration level in the task processing result satisfies a switching condition.

[0161] If the switching condition is satisfied, the control strategy of the mobile object is switched from the first control strategy to the second control strategy, and the mobile object is controlled by using the second control strategy.

[0162] The attention concentration degree indicated by the attention concentration level in the task processing result satisfies the switching condition, including: being lower than the attention concentration degree indicated by the attention concentration level threshold.

[0163] According to another embodiment of the present application, Figure 6 Each unit in the signal processing apparatus shown can be combined into one or several other units respectively or all, or some of the units can be further split into a plurality of units with smaller functions to constitute, which can achieve the same operation without affecting the implementation of the technical effects of the embodiments of the present application. The above units are divided based on logical functions, and in actual application, the functions of one unit can also be implemented by multiple units, or the functions of multiple units are implemented by one unit. In other embodiments of the present application, the signal processing apparatus can also include other units, and in actual application, these functions can also be assisted by other units, and can be implemented by multiple units.

[0164] According to another embodiment of the present application, the signal processing apparatus as shown in Figure 2 or Figure 4 The corresponding method shown in the steps involved in the computer program can be constructed as the signal processing apparatus shown in Figure 6 and to implement the signal processing method of the embodiments of the present application. The computer program can be recorded on, for example, a computer readable storage medium, and loaded into the above-mentioned computing device through the computer readable storage medium, and run therein.

[0165] In the embodiments of the present application, the term "module" or "unit" refers to a computer program or a part of a computer program with a predetermined function, and works with other related parts to achieve a predetermined target, and can be implemented entirely or partially by using software, hardware (such as processing circuit or memory) or combination thereof. Similarly, one processor (or multiple processors or memory) can be used to implement one or more modules or units. In addition, each module or unit can be a part of the overall module or unit that contains the function of the module or unit.

[0166] In the embodiments of the present application, N groups of brain electrical signals collected by the detection object in the corresponding brain region of the target object can be obtained according to the target brain electrical acquisition rule corresponding to the attention detection task, and then the task processing result containing the attention concentration level can be determined based on the brain electrical signals, and the attention concentration level can indicate the degree of attention concentration of the target object. The attention detection process is not affected by environmental factors such as light, and the accuracy of the attention concentration level obtained by attention detection is high, that is, the accuracy of attention detection is high, the stability is high, and the application range is wide. Moreover, in the process of attention detection based on brain electrical signals, the target source signal associated with the attention detection task can be extracted from the obtained brain electrical signals, and then the attention detection operation corresponding to the attention detection task can be performed according to the extracted target source signal associated with the attention detection task, and the task processing result containing the attention concentration level is obtained. Compared with the task processing result obtained by directly performing the attention detection operation corresponding to the attention detection task according to the obtained brain electrical signals, the accuracy of the task processing result can be improved, that is, the accuracy of the obtained attention concentration level can be improved, and the accuracy of the attention detection can be improved.

[0167] Based on the description of the above method embodiments and device embodiments, the embodiments of the present application also provide a computer device, which can be the above signal processing device. Please refer to Figure 7 The computer device at least includes a processor 701, an input interface 702, an output interface 703, and a computer readable storage medium 704. The processor 701, the input interface 702, the output interface 703, and the computer readable storage medium 704 in the computer device can be connected by bus or other means. The computer readable storage medium 704 can be stored in the memory of the computer device, and the computer readable storage medium 704 is used to store the computer program, and the processor 701 is used to execute the computer program stored in the computer readable storage medium 704. The processor 701 (or CPU (Central Processing Unit, Central Processing Unit)) is the computing core and control core of the computer device, which is suitable for running the computer program to realize the corresponding method process or corresponding function.

[0168] In one embodiment, the processor 701 proposed in the embodiments of the present application can be used to implement the processing process related to attention detection, specifically including: obtaining N groups of electroencephalogram signals collected by a detection object in a brain region corresponding to a target electroencephalogram acquisition rule according to the target electroencephalogram acquisition rule; N is a positive integer; pre-processing each group of electroencephalogram signals respectively to obtain a plurality of source signals of each group of electroencephalogram signals in the N groups of electroencephalogram signals; determining a target source signal associated with an attention detection task from the plurality of source signals of each group of electroencephalogram signals respectively based on signal characteristics of the plurality of source signals; and performing an attention detection operation corresponding to the attention detection task according to the target source signal corresponding to each group of electroencephalogram signals to obtain a task processing result of the attention detection task; the task processing result includes an attention concentration level, and the like.

[0169] The embodiments of the present application also provide a computer readable storage medium (Memory) which is a memory device in a computer device and is used to store computer programs and data. It can be understood that the computer readable storage medium herein can include an internal storage medium in the computer device, and of course can also include an extended storage medium supported by the computer device. The computer readable storage medium provides a storage space which stores an operating system of the computer device. Moreover, the storage space also stores a computer program which is suitable for being loaded and executed by the processor 701 to implement the corresponding method processes provided by the embodiments of the present application. It should be noted that the computer readable storage medium herein can be a high-speed RAM memory, or a non-volatile memory such as at least one disk memory; optionally, it can also be at least one computer readable storage medium located away from the aforementioned processor.

[0170] In one embodiment, the computer program stored in the computer readable storage medium can be loaded and executed by the processor to implement the corresponding steps in the method embodiments shown in the above Figure 2 or Figure 4 In one embodiment, the computer program stored in the computer readable storage medium can be loaded and executed by the processor to implement the corresponding steps in the method embodiments shown in the above

[0171] obtaining N groups of electroencephalogram signals collected by a detection object in a brain region corresponding to a target electroencephalogram acquisition rule according to the target electroencephalogram acquisition rule; N is a positive integer;

[0172] pre-processing each group of electroencephalogram signals respectively to obtain a plurality of source signals of each group of electroencephalogram signals in the N groups of electroencephalogram signals;

[0173] determining a target source signal associated with an attention detection task from the plurality of source signals of each group of electroencephalogram signals respectively based on signal characteristics of the plurality of source signals;

[0174] According to the target source signal corresponding to each group of electroencephalogram signals, the attention detection operation corresponding to the attention detection task is performed to obtain a task processing result of the attention detection task; the task processing result includes an attention concentration level.

[0175] In an implementation, the processor 701 can be specifically configured to:

[0176] filtering the i-th group of electroencephalogram signals to obtain an intermediate signal corresponding to the i-th group of electroencephalogram signals; wherein i ∈ [1, N];

[0177] performing blind source separation processing on the intermediate signal corresponding to the i-th group of electroencephalogram signals to obtain a plurality of source signals of the i-th group of electroencephalogram signals;

[0178] The processor 701 can be specifically configured to:

[0179] performing feature conversion processing on the plurality of source signals of the i-th group of electroencephalogram signals to obtain a plurality of signal features of the i-th group of electroencephalogram signals;

[0180] determining the target source signal associated with the attention detection task from the plurality of source signals of the i-th group of electroencephalogram signals according to the plurality of signal features of the i-th group of electroencephalogram signals;

[0181] The feature conversion processing includes differential entropy linear dynamic system smoothing processing, and the corresponding signal feature is a differential entropy linear dynamic system smoothing feature;

[0182] Or the feature conversion processing includes differential entropy moving average processing, and the corresponding signal feature is a differential entropy moving average feature;

[0183] Or the feature conversion processing includes power spectral density linear dynamic system smoothing processing, and the corresponding signal feature is a power spectral density linear dynamic system smoothing feature;

[0184] Or the feature conversion processing includes power spectral density moving average processing, and the corresponding signal feature is a power spectral density moving average feature.

[0185] In an implementation, the processor 701 can be specifically configured to:

[0186] performing feature extraction processing on the target source signal corresponding to each group of electroencephalogram signals to obtain an electroencephalogram signal feature of each group of electroencephalogram signals;

[0187] The attention detection model is called to perform attention analysis on the EEG signal features of each set of EEG signals to obtain an attention evaluation value at the current collection time point; the current collection time point is a collection time point of the N sets of EEG signals, and the attention evaluation value is used to indicate a probability of attention concentration;

[0188] Based on the attention evaluation value at the current collection time point and the historical attention evaluation value determined at the target period before the current collection time point, a task processing result of the attention detection task is generated.

[0189] In an implementation, when the processor 701 is used to call the attention detection model to perform attention analysis on the EEG signal features of each set of EEG signals to obtain an attention evaluation value at the current collection time point, the processor 701 can be specifically used to:

[0190] The attention detection model is called to determine a feature similarity between each reference feature data in the reference feature data set and the target feature data; the target feature data includes the EEG signal features of the N sets of EEG signals, and the reference feature data set is determined in a training process of the attention detection model and includes the EEG signal features of the N sets of EEG signals for training;

[0191] According to the feature similarity, a target number of reference feature data are selected from the reference feature data set; the feature similarity between any selected reference feature data and the target feature data is greater than or equal to the feature similarity between other unselected reference feature data and the target feature data;

[0192] The attention labels corresponding to each selected reference feature data are obtained, and a statistical number of reference feature data corresponding to each type of attention label is counted; the types of attention labels include a label type used to represent attention concentration and a label type used to represent attention dispersion;

[0193] According to the statistical number of each type of attention label, the attention evaluation value at the current collection time point is determined.

[0194] In an implementation, when the processor 701 is used to generate a task processing result of the attention detection task based on the attention evaluation value at the current collection time point and the historical attention evaluation value determined at the target period before the current collection time point, the processor 701 can be specifically used to:

[0195] The historical attention evaluation values of a plurality of historical collection time points of the target period are obtained;

[0196] According to the attention evaluation value at the current collection time point and the historical attention evaluation values at the plurality of historical collection time points, a time length proportion of the attention evaluation value greater than the attention threshold value is determined;

[0197] According to the time length proportion, an attention concentration level is determined, and a task processing result of the attention detection task is generated based on the determined attention concentration level; wherein the attention concentration degree indicated by the determined attention concentration level is positively correlated with the time length proportion.

[0198] In an implementation, the processor 701 is further configured to:

[0199] Obtain a sample feature data set; wherein the sample feature data set includes a plurality of sample feature data, and the sample feature data set is determined in a training process of the attention detection model and includes brain electrical signal features of the N groups of brain electrical signals for training;

[0200] Perform H times of data division processing on the sample feature data set to obtain H data division results; wherein one data division result includes one training data set and one validation data set divided from the sample feature data set, the sample feature data in the training data set is used as training data, and the sample feature data in the validation data set is used as validation data, and H is a positive integer;

[0201] According to the M reference quantities and the feature similarity between the validation data in the validation data set and the training data in the training data set in the H data division results, determine a prediction effect evaluation value of predicting the attention label based on different data division results under each reference quantity; M is a positive integer;

[0202] Average process the prediction effect evaluation values of predicting the attention label based on different data division results under each reference quantity to obtain an average prediction effect evaluation value under each reference quantity;

[0203] Take the reference quantity corresponding to the target average prediction effect evaluation value as a target quantity; wherein the target average prediction effect evaluation value is the optimal value among the average prediction effect evaluation values.

[0204] In an implementation, before the processor 701 is configured to obtain the N groups of brain electrical signals collected by the detection object in the brain region corresponding to the target brain electrical acquisition rule according to the target brain electrical acquisition rule, the processor 701 is further configured to:

[0205] In response to a task request about the attention detection task, determine the target brain electrical acquisition rule corresponding to the attention detection task according to the correspondence between the task and the brain electrical acquisition rule;

[0206] The EEG acquisition rule is used to indicate a brain region from which an EEG signal needs to be acquired, and different tasks correspond to different EEG acquisition rules.

[0207] In an embodiment, the N groups of EEG signals are acquired from the brain region of the target object; the processor 701 is further configured to:

[0208] Upon receiving the task request, the first object feature and the second object feature of the target object are acquired; the first object feature and the second object feature include any one or more of a facial feature, a heart rate feature, and a blood oxygen feature of the target object, and the first object feature and the second object feature are different;

[0209] The attention concentration level is predicted based on the first object feature and the second object feature of the target object, respectively;

[0210] If the attention concentration levels predicted respectively are different, a step of determining a target EEG acquisition rule corresponding to the attention detection task according to the correspondence between the task and the EEG acquisition rule is triggered to be executed.

[0211] In an embodiment, the N groups of EEG signals are acquired from the brain region of the target object; the processor 701 is further configured to:

[0212] The working scenario of the target object is determined;

[0213] If the working scenario is a mobile control scenario under the first control strategy, it is determined whether the attention concentration degree indicated by the attention concentration level in the task processing result satisfies a switching condition;

[0214] If the switching condition is satisfied, the control strategy of the mobile object is switched from the first control strategy to the second control strategy, and the mobile object is controlled by using the second control strategy;

[0215] The attention concentration degree indicated by the attention concentration level in the task processing result satisfies the switching condition, which includes that the attention concentration degree is lower than that indicated by an attention concentration level threshold.

[0216] In the embodiments of the present application, N groups of brain electrical signals collected by the detection object in the corresponding brain region of the target object can be obtained according to the target brain electrical acquisition rule corresponding to the attention detection task, and then the task processing result containing the attention concentration level can be determined based on the brain electrical signals, and the attention concentration level can indicate the attention concentration degree of the target object. The attention detection is performed by collecting the brain electrical signals of the target object, so that the attention detection process is not affected by environmental factors such as light, and the accuracy of the attention concentration level obtained by the attention detection is high, that is, the accuracy of the attention detection is high, the stability is high, and the application range is wide. Moreover, in the process of attention detection based on the brain electrical signals, the target source signal associated with the attention detection task can be extracted from the obtained brain electrical signals, and then the attention detection operation corresponding to the attention detection task can be performed according to the extracted target source signal associated with the attention detection task, and the task processing result containing the attention concentration level is obtained. Compared with the task processing result obtained by directly performing the attention detection operation corresponding to the attention detection task according to the obtained brain electrical signals, the accuracy of the task processing result can be improved, that is, the accuracy of the obtained attention concentration level can be improved, and the accuracy of the attention detection can be improved.

[0217] The embodiments of the present application provide a computer program product, which comprises 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 the processor executes the computer program to enable the computer device to perform the method embodiments shown in the above Figure 2 or Figure 4 It should be understood that the above only discloses the preferred embodiments of the present application, and of course cannot limit the scope of the rights of the present application, so the equivalent changes made according to the claims of the present application still fall within the scope of the present application.

Claims

1. A signal processing method, characterized by, The method comprises the following steps: According to the target electroencephalogram acquisition rule, N groups of electroencephalogram signals collected by the detection object in the brain region corresponding to the target electroencephalogram acquisition rule are acquired; N is a positive integer; Each group of electroencephalogram signals is preprocessed respectively to obtain a plurality of source signals of each group of electroencephalogram signals in the N groups of electroencephalogram signals; Based on the signal characteristics of the plurality of source signals, the target source signal associated with the attention detection task is determined from the plurality of source signals of each group of electroencephalogram signals respectively; According to the target source signal corresponding to each group of electroencephalogram signals, the attention detection operation corresponding to the attention detection task is performed to obtain the task processing result of the attention detection task; The task processing result includes: attention concentration level.

2. The method of claim 1, wherein, The method comprises the following steps: Filtering processing is performed on the i-th group of electroencephalogram signals in the N groups of electroencephalogram signals to obtain the intermediate signal corresponding to the i-th group of electroencephalogram signals; wherein i∈[1,N]; Blind source separation processing is performed on the intermediate signal corresponding to the i-th group of electroencephalogram signals to obtain a plurality of source signals of the i-th group of electroencephalogram signals; The method comprises the following steps: The plurality of signal characteristics of the i-th group of electroencephalogram signals are obtained by performing feature conversion processing on the plurality of source signals of the i-th group of electroencephalogram signals respectively; According to the plurality of signal characteristics of the i-th group of electroencephalogram signals, the target source signal associated with the attention detection task is determined from the plurality of source signals of the i-th group of electroencephalogram signals; The feature conversion processing includes differential entropy linear dynamic system smoothing processing, and the corresponding signal characteristic is differential entropy linear dynamic system smoothing characteristic; Or the feature conversion processing includes differential entropy moving average processing, and the corresponding signal characteristic is differential entropy moving average characteristic; Or the feature conversion processing includes power spectral density linear dynamic system smoothing processing, and the corresponding signal characteristic is power spectral density linear dynamic system smoothing characteristic; Or the feature conversion processing includes power spectral density moving average processing, and the corresponding signal characteristic is power spectral density moving average characteristic.

3. The method of claim 1, wherein, The method comprises the following steps: The electroencephalogram signal characteristics of each group of electroencephalogram signals are obtained by performing feature extraction processing on the target source signal corresponding to each group of electroencephalogram signals respectively; The attention analysis is performed on the electroencephalogram signal characteristics of each group of electroencephalogram signals by calling the attention detection model to obtain the attention evaluation value of the current collection time point; wherein the current collection time point is the collection time point of the N groups of electroencephalogram signals, and the attention evaluation value is used to indicate the probability of attention concentration; Based on the attention evaluation value of the current collection time point and the historical attention evaluation value determined in the target period before the current collection time point, the task processing result of the attention detection task is generated.

4. The method of claim 3, wherein, The calling attention detection model performs attention analysis on the electroencephalogram signal features of each group of electroencephalogram signals to obtain an attention evaluation value at the current acquisition time point, including: The calling attention detection model determines the feature similarity between the target feature data and each reference feature data in the reference feature data set; wherein the target feature data includes the electroencephalogram signal features of the N groups of electroencephalogram signals, and the reference feature data set is determined in the training process of the attention detection model and includes the electroencephalogram signal features of the N groups of electroencephalogram signals for training; According to the feature similarity, a target number of reference feature data are selected from the reference feature data set; wherein the feature similarity between any selected reference feature data and the target feature data is greater than or equal to the feature similarity between other unselected reference feature data and the target feature data; The attention labels corresponding to each selected reference feature data are obtained, and the statistical number of reference feature data corresponding to each type of attention label is counted; wherein the types of attention labels include a label type for indicating attention concentration and a label type for indicating attention dispersion. According to the statistical number of each type of attention label, the attention evaluation value at the current acquisition time point is determined.

5. The method of claim 3, wherein, The task processing result of the attention detection task is generated based on the attention evaluation value at the current acquisition time point and the historical attention evaluation values determined in a target time period before the current acquisition time point, including: The historical attention evaluation values of a plurality of historical acquisition time points in the target time period are obtained; According to the attention evaluation value at the current acquisition time point and the historical attention evaluation values of the plurality of historical acquisition time points, a duration proportion of the attention evaluation value greater than an attention threshold value is determined; According to the duration proportion, an attention concentration level is determined, and the task processing result of the attention detection task is generated based on the determined attention concentration level; wherein the attention concentration degree indicated by the determined attention concentration level is positively correlated with the duration proportion.

6. The method of claim 4, wherein, The method further includes: Obtaining a sample feature data set; wherein the sample feature data set includes a plurality of sample feature data, and the sample feature data set is determined in the training process of the attention detection model and includes the electroencephalogram signal features of the N groups of electroencephalogram signals for training; The sample feature data set is subjected to H times of data division processing to obtain H data division results; wherein one data division result includes one training data set and one validation data set divided from the sample feature data set, the sample feature data in the training data set is used as training data, and the sample feature data in the validation data set is used as validation data, and H is a positive integer; According to M reference quantities and the feature similarity between the validation data in the validation data set and the training data in the training data set in the H data division results, a prediction effect evaluation value is determined for predicting attention labels based on different data division results under each reference quantity; M is a positive integer. Average the prediction effect evaluation values obtained when predicting the attention label based on different data division results under each reference quantity to obtain an average prediction effect evaluation value under each reference quantity; The reference quantity corresponding to the target average prediction effect evaluation value is taken as the target quantity; wherein the target average prediction effect evaluation value is the optimal value among the average prediction effect evaluation values.

7. The method of claim 1, wherein, Before acquiring the N sets of brain electrical signals collected by the target object in the brain region corresponding to the target brain electrical acquisition rule, the method further comprises: In response to a task request for an attention detection task, determining the target brain electrical acquisition rule corresponding to the attention detection task according to the correspondence between the task and the brain electrical acquisition rule; Wherein, the brain electrical acquisition rule is used to indicate the brain region where the brain electrical signal needs to be acquired, and different tasks correspond to different brain electrical acquisition rules.

8. The method of claim 7, wherein, The N sets of brain electrical signals are collected in the brain region of the target object; the method further comprises: Upon receiving the task request, acquiring the first object feature and the second object feature of the target object; wherein the first object feature and the second object feature include any one or more of the following: facial features of the target object, heart rate features of the target object, and blood oxygen features of the target object, and the first object feature and the second object feature are different; Predicting the attention concentration level based on the first object feature of the target object and the second object feature of the target object, respectively; If the attention concentration levels predicted respectively are different, triggering the step of determining the target brain electrical acquisition rule corresponding to the attention detection task according to the correspondence between the task and the brain electrical acquisition rule.

9. The method of claim 1, wherein, The N sets of brain electrical signals are collected in the brain region of the target object; the method further comprises: Determining the working scenario of the target object; If the working scenario is a mobile control scenario under a first control strategy, determining whether the attention concentration degree indicated by the attention concentration level in the task processing result meets the switching condition; If the switching condition is met, switching the control strategy of the mobile object from the first control strategy to a second control strategy, and controlling the mobile object using the second control strategy; Wherein, the attention concentration degree indicated by the attention concentration level in the task processing result meets the switching condition, including: the attention concentration degree is lower than the attention concentration degree indicated by the attention concentration level threshold.

10. A signal processing device, characterized by It comprises: A communication unit for communication interaction; A processing unit for acquiring N sets of brain electrical signals collected by the detection object in the brain region corresponding to the target brain electrical acquisition rule according to the target brain electrical acquisition rule; N is a positive integer; Preprocess each set of brain electrical signals to obtain multiple source signals of each set of brain electrical signals in the N sets of brain electrical signals; based on the signal characteristics of the multiple source signals, determine the target source signal associated with the attention detection task from the multiple source signals of each set of brain electrical signals, respectively; According to the target source signal corresponding to each group of brain electrical signals, an attention detection operation corresponding to the attention detection task is performed to obtain a task processing result of the attention detection task. The task processing result includes an attention concentration level.

11. A computer device comprising an input interface and an output interface, characterized in that, Further comprising: a processor and a computer readable storage medium; the computer readable storage medium is used to store a computer program; the processor is used to run the computer program, and realize the signal processing method in any one of claims 1-9.

12. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a computer program, and the computer program is adapted to be loaded and executed by the processor to realize the signal processing method in any one of claims 1-9.

13. A computer program product, characterised in that, The computer program product includes a computer program, and the computer program is adapted to be loaded and executed by the processor to realize the signal processing method in any one of claims 1-9.

Images