Identity recognition method, controller and guard device

Abstract

The application relates to an identity recognition method, a controller and a guard device. The method is applied to a controller of a guard device, the guard device further comprises a laser radar and a microphone array, and the method comprises the following steps: acquiring a plurality of point cloud data frames collected by the laser radar, and performing mobile detection processing on the point cloud data frames to determine a mobile target; performing sound tracking on the mobile target by using the microphone array to obtain an audio beam signal; and performing biological feature recognition processing on the audio beam signal by using a preset identity recognition model to obtain an identity recognition result. By adopting the method, the false positive rate can be effectively reduced, and the accuracy of the identity recognition method is improved.

Description

Technical Field

[0001] This application relates to the field of intelligent security technology, and in particular to an identity recognition method, controller and guard device. Background Technology

[0002] With the development of smart home technology, home security systems are receiving increasing attention. Home security systems are primarily used to monitor moving targets within a protected area and identify them to determine the security of the area.

[0003] Traditional identification methods rely on devices such as fixed cameras, fingerprint sensors, voice sensors, or infrared motion detectors to collect corresponding feature signals of moving targets, and then perform identification and classification processing on these feature signals to determine the identity of the moving target.

[0004] However, current identity verification methods suffer from a high false alarm rate in home nighttime security applications. Summary of the Invention

[0005] Therefore, it is necessary to provide an identity recognition method, controller, and guard device that can reduce the false alarm rate in home nighttime security application scenarios to address the above-mentioned technical problems.

[0006] In a first aspect, this application provides an identity recognition method applied to the controller of a guard device, the guard device further including a lidar and a microphone array, the method comprising:

[0007] Acquire multiple point cloud data frames collected by the lidar, and perform motion detection processing on the point cloud data frames to determine the moving target;

[0008] A microphone array is used to track the sound of a moving target and obtain the audio beam signal;

[0009] The audio beam signal is processed for biometric identification using a pre-defined identity recognition model to obtain the identity recognition result.

[0010] In one embodiment, motion detection processing is performed on the point cloud data frame to determine the moving target, including:

[0011] Perform differential analysis on at least two consecutive point cloud data frames to obtain at least one moving object;

[0012] Filter at least one moving object to determine the moving target.

[0013] In one embodiment, a microphone array is used to perform sound tracking on a moving target to obtain an audio beam signal, including:

[0014] Based on the moving target, determine the first azimuth and first distance of the moving target relative to the guarding equipment;

[0015] The directional parameters of the microphone array are adjusted according to the first azimuth angle and the first distance, and multiple sound signals collected by the microphone array are acquired.

[0016] Multiple sound signals are superimposed and enhanced to obtain an audio beam signal.

[0017] In one embodiment, a preset identity recognition model is used to perform biometric recognition processing on the audio beam signal to obtain an identity recognition result, including:

[0018] The audio beam signal is categorized to obtain the audio types;

[0019] When the audio type is footsteps, the identity recognition model is used to extract features from the audio beam signal to obtain the audio features corresponding to the audio beam signal.

[0020] The similarity calculation of audio features is performed using a pre-defined member database to obtain the membership probability of audio features and each tag in the member database.

[0021] The identity recognition result is determined based on the membership probability and the preset recognition threshold.

[0022] In one embodiment, the method further includes:

[0023] If the identification result is unsuccessful, the second azimuth angle and second distance of the moving target relative to the guarding device are determined based on the moving target.

[0024] Based on the second azimuth angle and the second distance, control the guarding equipment to stand up, turn to the direction corresponding to the moving target, and visually confirm the moving target to obtain the visual confirmation result;

[0025] If the visual confirmation result is unsuccessful, an alarm message will be generated based on the visual confirmation result.

[0026] In one embodiment, the method further includes:

[0027] If the visual confirmation result is satisfactory, the guard device is switched to listening mode. In listening mode, the lidar is controlled to collect point cloud data frames at a preset scanning frequency, and the microphone array is controlled to perform omnidirectional sound reception.

[0028] In one embodiment, the method further includes:

[0029] If the visual confirmation result is satisfactory, the audio beam signal is input into the identity recognition model to extract the corresponding voiceprint feature vector.

[0030] The corresponding identity tag of the moving target is obtained, and the voiceprint feature vector is associated with the identity tag and stored in the member database to obtain the updated member database. The updated member database is used to provide an identity comparison benchmark for the subsequently acquired audio beam signals.

[0031] Secondly, this application also provides a controller for performing the identity recognition method as described in the first aspect.

[0032] Thirdly, this application also provides a guard device, including a lidar, a microphone array, and a controller as described in the second aspect.

[0033] In one embodiment, the guarding device further includes: a night vision camera, a torque servo motor, and multiple joints;

[0034] Torque servo motors and multiple joints are used to move based on motion control signals sent by the controller, enabling the guarding device to stand up and turn in the direction of the moving target;

[0035] Night vision cameras are used to collect visual information of moving targets and transmit it to the controller;

[0036] The controller is also used to obtain visual confirmation results based on visual information.

[0037] The aforementioned identification method, controller, and guarding equipment acquire point cloud data frames corresponding to the guarded area through lidar. The lidar detection process does not depend on external lighting conditions, overcoming the limitations of traditional vision-based identification methods under low-light conditions. By using a microphone array to perform directional acoustic tracking of moving targets determined based on point cloud data frames and generating audio beam signals for the moving targets, the system can suppress background noise interference from non-target directions, effectively reducing the false alarm rate and improving the accuracy of the identification method. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0039] Figure 1 This is a schematic diagram of the module structure of the guard device in one embodiment;

[0040] Figure 2 This is a flowchart illustrating an identity recognition method in one embodiment;

[0041] Figure 3 This is a flowchart illustrating the steps for obtaining an identity recognition result in one embodiment;

[0042] Figure 4 This is a flowchart illustrating the identity recognition method in another embodiment;

[0043] Figure 5 This is a structural block diagram of an identity recognition device in one embodiment;

[0044] Figure 6 This is a schematic diagram of a guard device in one embodiment;

[0045] Figure 7 This is an internal structural diagram of a computer device in one embodiment.

[0046] Explanation of reference numerals in the attached figures:

[0047] 10. Guard equipment; 11. Controller; 12. LiDAR; 13. Microphone array; 14. User terminal. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0049] It should be noted that the terms "first," "second," etc., used in this application can be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "comprising" and "having," and any variations thereof, used in this application, are intended to cover non-exclusive inclusion. The term "multiple" used in this application refers to two or more. The term "and / or" used in this application refers to one of the embodiments, or any combination of multiple embodiments.

[0050] The identity recognition method provided in this application embodiment can be applied to, for example... Figure 1 The controller 11 of the guard device 10 shown is included. The guard device 10 includes a lidar 12 and a microphone array 13.

[0051] The lidar 12 and microphone array 13 are mounted on the main body of the guard device 10. The lidar 12 continuously emits laser pulses and receives laser signals reflected back from objects in the environment to actively scan the surrounding environment. Based on the time-of-flight and scanning angle information of the laser pulses, the lidar 12 acquires point cloud data reflecting the current spatial distribution of the environment in real time, forms point cloud data frames, and sends them to the controller 11 of the guard device 10. The microphone array 13 includes multiple microphone units, which are arranged in a specific geometric pattern, such as a linear arrangement or a circular arrangement.

[0052] The controller 11 is connected to the lidar 12 and the microphone array 13, respectively, to acquire multiple point cloud data frames collected by the lidar 12, perform motion detection processing on the point cloud data frames to determine the moving target, use the microphone array 13 to track the moving target by sound to obtain an audio beam signal, and use a preset identity recognition model to perform biometric recognition processing on the audio beam signal to obtain the identity recognition result. The controller 11 is also connected to the user terminal 14 to transmit the identity recognition result to the user terminal 14.

[0053] The controller 11 may be, but is not limited to, various embedded computing units or edge computing devices; the guard device 10 may be, but is not limited to, various mobile devices used to perform guarding, patrolling, monitoring or surveillance tasks, such as robots, robot dogs or drones.

[0054] The application scenarios of the guard device 10 in this application embodiment may include, but are not limited to, home security, corporate park security, facility security, nursing home monitoring, border patrol, and wildlife monitoring. The following description uses the scenario of home security as an example.

[0055] In one exemplary embodiment, such as Figure 2 As shown, an identity recognition method is provided, which can be applied to... Figure 1 Taking the guard device 10 as an example, the explanation includes the following steps 202 to 206. Wherein:

[0056] Step 202: Acquire multiple point cloud data frames collected by the lidar 12, and perform motion detection processing on the point cloud data frames to determine the moving target.

[0057] Motion detection processing refers to detecting point cloud data in point cloud data frames that have undergone significant changes in spatial location or state of existence, and aggregating these changed point cloud data to combine multiple dynamic points that are spatially adjacent and have consistent change characteristics into a moving target, so as to transform discrete point cloud data into an overall target with spatial scale and positional characteristics.

[0058] In some embodiments, after acquiring point cloud data frames, the point cloud data frames are preprocessed to remove obviously abnormal point cloud data; the spatial range of the point cloud data is limited to reduce the impact of environmental noise on the detection results.

[0059] Step 204: Use microphone array 13 to perform sound tracking on the moving target to obtain audio beam signals.

[0060] Sound tracking refers to adjusting the sound receiving direction of each microphone unit in the microphone array 13 in real time according to the location of the moving target, so that the microphone array 13 points in the direction of the moving target and performs directional acoustic reception to form an audio beam signal. The audio beam signal has the strongest signal strength in the direction of the moving target, and the signal strength in other directions is weaker.

[0061] Step 206: Use a preset identity recognition model to perform biometric recognition processing on the audio beam signal to obtain the identity recognition result.

[0062] The identity recognition model can be a machine learning model trained based on historical training samples. The historical training samples include multiple audio beam signals collected within a historical time interval.

[0063] For example, the identity recognition model can be a voiceprint recognition model, which analyzes the spectral distribution, vocal habits and vocal tract characteristics in the audio beam signal, compares the currently detected audio beam signal with the registered voiceprint template, and thus determines whether the moving target belongs to a known individual.

[0064] For example, the identity recognition model can be a fusion recognition model based on language features, breathing features, or footstep features, which can still identify the moving target based on the sound patterns generated by its walking or activities, even when the moving target does not make explicit speech.

[0065] In the above-mentioned identity recognition method, the point cloud data frame corresponding to the guarded area is obtained by the lidar 12. The detection process of the lidar 12 does not depend on the external lighting conditions, overcoming the limitations of the traditional vision-based identity recognition method under low light conditions. The microphone array 13 performs directional acoustic tracking on the moving target determined based on the point cloud data frame and generates an audio beam signal for the moving target. This can suppress background noise interference from non-target directions, effectively reduce the false alarm rate, and improve the accuracy of the identity recognition method.

[0066] In one exemplary embodiment, motion detection processing is performed on point cloud data frames to determine a moving target, including: performing differential analysis on at least two consecutive point cloud data frames to obtain at least one moving object; and performing filtering processing on the at least one moving object to determine the moving target.

[0067] Differential analysis refers to comparing two consecutive frames of point cloud data and extracting the point clusters whose positions have changed, which are the moving objects.

[0068] Filtering refers to filtering out moving objects whose positional changes and sizes are less than a preset threshold, in order to filter out misjudgments caused by minor measurement errors, slow speeds, or small sizes, such as swaying curtains, pets, or insects.

[0069] In some embodiments, before differential analysis, the point cloud data frames are spatially aligned so that point cloud data frames acquired at different times are under a unified coordinate reference. Based on this, point cloud data that have undergone significant changes in spatial position or existence state between point cloud data frames are identified.

[0070] In this embodiment of the application, by performing differential analysis and filtering on point cloud data frames, objects that exhibit stable displacement trends in multiple consecutive time frames are identified as valid moving targets, which can effectively filter out interference data and reduce the false judgment rate.

[0071] In an exemplary embodiment, the microphone array 13 is used to perform sound tracking on a moving target to obtain an audio beam signal, including: determining a first azimuth angle and a first distance of the moving target relative to the guard device 10 based on the moving target; adjusting the directional parameters of the microphone array 13 according to the first azimuth angle and the first distance, and acquiring multiple sound signals collected by the microphone array 13; and superimposing and enhancing the multiple sound signals to obtain an audio beam signal.

[0072] The first azimuth angle is used to characterize the spatial position of the moving target relative to the guard device 10 in the horizontal direction, while the first distance is used to characterize the spatial interval between the moving target and the guard device 10. The controller 11 adjusts the directional parameters of the microphone array 13, controlling the microphone array 13 to turn towards the direction of the moving target, so that the microphone array 13 collects sound signals in that direction.

[0073] Superposition and enhancement processing refers to introducing different time delays and weights into each sound signal, so that the sound wave from the moving target is enhanced during superposition, while sound waves from other directions cancel each other out or are suppressed during superposition, thus forming a directional auditory beam with direction selectivity in space, i.e., an audio beam signal. The direction of the moving target is the main lobe direction of the audio beam signal, and the other directions are the side lobe directions of the audio beam signal.

[0074] For example, the half-power beamwidth of the main lobe is about 30°, which enables the microphone array 13 to maintain stable acoustic reception while ensuring high sensitivity to sound in the target direction.

[0075] For example, for the first azimuth angle θ, the audio beam signal y(t) can be expressed as:

[0076] y(t)=∑m=1Mw m ·x m (t-Δm(θ))y(t)=∑m=1Mwm·x m (t-Δm(θ)),

[0077] Where M is the number of microphones, x m (t) is the input signal of the m-th microphone, w m Δm(θ) is the weighting coefficient, and Δm(θ) is the time delay compensation introduced by the m-th microphone to align the sound waves from the first azimuth angle θ.

[0078] In this embodiment, the directional parameters of the microphone array 13 are dynamically adjusted according to the first azimuth angle and the first distance of the moving target, so that the receiving main lobe of the audio beam signal is always facing the current direction of the moving target, thereby achieving accurate and reliable directional sound acquisition and significantly improving the accuracy of identity recognition based on audio beam signals.

[0079] In one exemplary embodiment, such as Figure 3 As shown, the process of using a preset identity recognition model to perform biometric recognition processing on audio beam signals to obtain identity recognition results includes:

[0080] Step 302: Perform category filtering on the audio beam signal to obtain the audio types.

[0081] Among them, the category selection process can utilize machine learning algorithms, such as SVM (Support Vector Machine), random forest, or deep learning, to obtain the audio category by observing the characteristics of the audio beam signal, such as short-time energy, zero-crossing rate, and spectral peak.

[0082] Step 304: When the audio type is footsteps, use an identity recognition model to extract features from the audio beam signal to obtain the audio features corresponding to the audio beam signal.

[0083] Footsteps are a type of biometric signal, containing unique information such as stride length, stride frequency, landing force, and foot shape. Audio features can include temporal features, frequency domain features, and temporal patterns.

[0084] For example, the identity recognition model can be a Hidden Markov Model (HMM). Temporal features can include the short-time energy, zero-crossing rate, and amplitude envelope of the footstep signal; frequency domain features can be characterized by MFCCs (Mel-scale Frequency Cepstral Coefficients); and temporal patterns can include the rhythm, intervals, and sequence patterns of the footsteps. During the research and development phase, feature extraction is performed using a large number of footstep data samples from many people in home settings, and an initial recognition model is trained. The objective function of the training is the model's ability to distinguish the footsteps of different people, resulting in the identity recognition model after training.

[0085] Step 306: Use the preset member database to perform similarity calculation on the frequency scale to obtain the membership probability of the audio features and each label in the member database.

[0086] Each family member corresponds to a label, and the member database includes the feature vector corresponding to each label. The similarity calculation process refers to calculating the similarity between the audio feature and the feature vector corresponding to each label, and then calculating the posterior probability of the audio feature belonging to each label based on the similarity as the membership probability.

[0087] In some embodiments, similarity can also be calculated by euclidean distance, cosine distance or Mahalanobis distance between audio features and feature vectors corresponding to each tag, and the distance values ​​are mapped to similarity.

[0088] Step 308: Determine the identity recognition result based on the membership probability and the preset recognition threshold.

[0089] Specifically, if the probability of affiliation is greater than or equal to the recognition threshold, the identity recognition result is determined to be successful, that is, the moving target is identified as a family member; if all the probabilities of affiliation are less than the recognition threshold, the identity recognition result is determined to be unsuccessful, that is, the moving target is identified as a non-family member.

[0090] In this embodiment of the application, by performing biometric recognition on footsteps, it is possible to identify the identity of a moving target based solely on footsteps in nighttime or low-visibility environments. The false alarm rate of the identity recognition method in this embodiment of the application is less than 5%, which is far lower than the false alarm rate of infrared motion detectors in the prior art.

[0091] In an exemplary embodiment, the method further includes: if the identification result is that the identification fails, determining a second azimuth angle and a second distance of the moving target relative to the guard device 10 based on the moving target; controlling the guard device 10 to stand up and turn to the direction corresponding to the moving target according to the second azimuth angle and the second distance, and visually confirming the moving target to obtain a visual confirmation result; if the visual confirmation result is that the identification fails, generating alarm information according to the visual confirmation result.

[0092] The process of controlling the guard device 10 to stand up may include: calculating the difference between the direction corresponding to the moving target and the current direction of the vision device, determining the rotation angle, decomposing the rotation angle into a rotation sequence of the motion module of the guard device 10, controlling the motion module to move according to the rotation sequence, and realizing the smooth turn of the guard device 10.

[0093] The visual verification process may include: using deep learning object detection algorithms, such as YOLO (You Only Look Once) or SSD (Single Shot MultiBox Detector), to analyze video frames captured by the visual device, identify and classify target objects in the video frames, and obtain a visual verification result. If the visual verification result is unsuccessful, it indicates the possible presence of an intruder, and the visual verification result and video frames are transmitted to the user's mobile app or security center via a communication module.

[0094] For example, the communication module can adopt any one or more of Wi-Fi 6, Bluetooth, ZigBee, 4G / 5G.

[0095] In some embodiments, if the visual confirmation result is unsuccessful, the guard device 10 can also be controlled to issue an audible and visual alarm, broadcast a preset warning message, or release a safety aerosol.

[0096] In some embodiments, during the movement of the guard device 10, the second azimuth angle and second distance of the moving target relative to the guard device 10 are continuously updated, and the movement direction of the guard device 10 is controlled according to the updated second azimuth angle and second distance to ensure that the vision device can be aligned with the moving target, thereby achieving visual confirmation of the moving target and obtaining a visual confirmation result. The vision device can be an infrared camera, which can work normally in a completely dark environment, overcoming the limitations of traditional vision systems in low-light conditions.

[0097] In this embodiment, video data is not collected during the regular monitoring phase. Visual verification is only initiated when the identification result obtained using the audio beam signal fails, which can effectively protect the privacy of family members. By controlling the guard device 10 to actively stand up and turn to the direction of the moving target and perform visual verification, active deterrence measures such as sound and light alarms can be integrated, which greatly improves the security effect.

[0098] In one possible implementation, if the visual confirmation result is satisfactory, the audio beam signal is input into the identity recognition model to extract the corresponding voiceprint feature vector; the corresponding identity tag of the moving target is obtained, and the voiceprint feature vector and the identity tag are associated and stored in the member database to obtain the updated member database; the updated member database is used to provide an identity comparison benchmark for subsequently acquired audio beam signals.

[0099] In one possible implementation, if the visual confirmation result is passed, the guard device 10 is switched to the listening mode. In the listening mode, the lidar 12 is controlled to collect point cloud data frames at a preset scanning frequency, and the microphone array 13 is controlled to perform omnidirectional sound reception.

[0100] For example, in listening mode, the guard device 10 returns to a preset guard point, such as a doorway or the end of a corridor, under the control of the controller 11, and the lidar 12 can scan at a scanning frequency of 5Hz.

[0101] In this embodiment, the guard device 10 maintains a low-power state before detecting a potential threat, and only gradually activates the more energy-intensive footstep recognition or visual confirmation module in the controller 11 when there is suspicious activity, thereby extending battery life.

[0102] In one exemplary embodiment, such as Figure 4 As shown, an identity recognition method is provided. This method is applied to the controller 11 of a guard device 10, which also includes a lidar 12 and a microphone array 13. The method includes:

[0103] Step 401: Acquire multiple point cloud data frames collected by the lidar 12, perform differential analysis on at least two consecutive point cloud data frames to obtain at least one moving object.

[0104] Step 402: Filter at least one moving object to determine the moving target.

[0105] Step 403: Based on the moving target, determine the first azimuth angle and the first distance of the moving target relative to the guarding device 10.

[0106] Step 404: Adjust the directional parameters of the microphone array 13 according to the first azimuth angle and the first distance, and acquire multiple sound signals collected by the microphone array 13.

[0107] Step 405: Superimpose and enhance multiple audio signals to obtain an audio beam signal.

[0108] Step 406: Use a preset identity recognition model to perform biometric recognition processing on the audio beam signal to obtain the identity recognition result.

[0109] The process involves: classifying the audio beam signal to obtain the audio type; extracting features from the audio beam signal using an identity recognition model when the audio type is footsteps; calculating the similarity of the audio features using a preset member database to obtain the membership probability of the audio features with each tag in the member database; and determining the identity recognition result based on the membership probability and a preset recognition threshold.

[0110] Step 407: If the identity recognition result is that the recognition fails, determine the second azimuth angle and the second distance of the moving target relative to the guard device 10 based on the moving target.

[0111] Step 408: Based on the second azimuth angle and the second distance, control the guard device 10 to stand up, turn to the direction corresponding to the moving target, and visually confirm the moving target to obtain the visual confirmation result.

[0112] Step 409: If the visual confirmation result is unsuccessful, generate an alarm message based on the visual confirmation result.

[0113] Step 410: If the visual confirmation result is satisfactory, add the audio beam signal to the audio database.

[0114] Step 411: Train the identity recognition model using the audio database to obtain an updated identity recognition model. The updated identity recognition model is used to perform biometric recognition processing on the subsequently acquired audio beam signals.

[0115] Step 412: If the visual confirmation result is passed, switch the guard device 10 to the listening mode. In the listening mode, control the lidar 12 to collect point cloud data frames at a preset scanning frequency, and control the microphone array 13 to perform omnidirectional sound reception.

[0116] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.

[0117] Based on the same inventive concept, this application also provides an identity recognition device for implementing the aforementioned identity recognition method. The solution provided by this device is similar to the implementation described in the above method; therefore, the specific limitations in one or more identity recognition device embodiments provided below can be found in the limitations of the identity recognition method described above, and will not be repeated here.

[0118] In one exemplary embodiment, such as Figure 5 As shown, an identity recognition device is provided, including: a motion detection module 502, a voice tracking module 504, and a feature recognition module 506, wherein:

[0119] The motion detection module 502 is used to acquire multiple point cloud data frames collected by the lidar 12, and to perform motion detection processing on the point cloud data frames to determine the moving target.

[0120] The sound tracking module 504 is used to track the sound of a moving target using the microphone array 13 and obtain an audio beam signal;

[0121] The feature recognition module 506 is used to perform biometric recognition processing on the audio beam signal using a preset identity recognition model to obtain the identity recognition result.

[0122] In one embodiment, the motion detection module 502 is further configured to perform differential analysis on at least two consecutive point cloud data frames to obtain at least one moving object; and to perform filtering processing on the at least one moving object to determine the moving target.

[0123] In one embodiment, the sound tracking module 504 is further configured to determine a first azimuth angle and a first distance of the moving target relative to the guard device 10 based on the moving target; adjust the directional parameters of the microphone array 13 according to the first azimuth angle and the first distance, and acquire multiple sound signals collected by the microphone array 13; and perform superposition and enhancement processing on the multiple sound signals to obtain an audio beam signal.

[0124] In one embodiment, the feature recognition module 506 is further configured to perform category filtering processing on the audio beam signal to obtain the audio type; when the audio type is footsteps, the feature of the audio beam signal is extracted using an identity recognition model to obtain the audio feature corresponding to the audio beam signal; the similarity calculation of the audio feature is performed using a preset member database to obtain the membership probability of the audio feature and each tag in the member database; and the identity recognition result is determined based on the membership probability and a preset recognition threshold.

[0125] In one embodiment, the device further includes a visual confirmation module, which, in the case of an identity recognition failure, determines a second azimuth angle and a second distance of the moving target relative to the guard device 10; controls the guard device 10 to stand up and turn to the direction corresponding to the moving target based on the second azimuth angle and the second distance, and performs visual confirmation on the moving target to obtain a visual confirmation result; and generates an alarm message based on the visual confirmation result if the visual confirmation result is unsuccessful.

[0126] In one embodiment, the device further includes a mode switching module, which switches the guard device 10 to a listening mode when the visual confirmation result is passed. In the listening mode, the lidar 12 is controlled to collect point cloud data frames at a preset scanning frequency, and the microphone array 13 is controlled to perform omnidirectional sound reception.

[0127] In one embodiment, the sound tracking module 504 is further configured to input the audio beam signal into the identity recognition model and extract the corresponding voiceprint feature vector when the visual confirmation result is passed; obtain the corresponding identity tag of the moving target, and associate the voiceprint feature vector with the identity tag and store it in the member database to obtain the updated member database; the updated member database is used to provide an identity comparison benchmark for the subsequently acquired audio beam signals.

[0128] Each module in the aforementioned identity recognition device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the operations corresponding to each module.

[0129] In one exemplary embodiment, a controller 11 is provided for executing the identity recognition method described in the above embodiments.

[0130] In one exemplary embodiment, a guard device 10 is provided, which includes a lidar 12, a microphone array 13, and a controller 11 as described in the above embodiment.

[0131] The lidar 12 in this embodiment can be a 360° two-dimensional lidar with a detection radius ≥12 meters and a scanning frequency ≥15Hz. The microphone array 13 can be a ring array composed of 4 MEMS microphones with an array diameter of 8cm, supporting beamforming and sound source localization.

[0132] In one possible implementation, the guard device 10 further includes: a night vision camera, a torque servo motor, and multiple joints; the torque servo motor and multiple joints are used to move based on motion control signals sent by the controller 11, so that the guard device 10 stands up and turns in the direction of the moving target; the night vision camera is used to collect visual information corresponding to the moving target and transmit it to the controller 11; the controller 11 is also used to obtain a visual confirmation result based on the visual information.

[0133] In one possible implementation, the guard device 10 is... Figure 6 The robot dog shown has an integrated infrared night vision camera in its head, which can capture video streams in completely dark environments.

[0134] For example, the robot dog's motion module can adopt a quadrupedal bionic design, wheeled or tracked, and is equipped with a high-torque servo motor and multiple precision joints. The controller 11 calculates the required rotation angle of each joint based on an inverse kinematics algorithm, and controls each joint to achieve smooth standing and turning movements through the lower-level controller 11 of the motion module.

[0135] In some embodiments, the guard device 10 also includes a power system, which may employ a 6000mAh, 7.4V lithium polymer battery that supports fast charging and hot-swapping.

[0136] In one exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 7As shown, the computer device includes a processor, memory, input / output interfaces, a communication interface, a display unit, and an input device. The processor, memory, and input / output interfaces are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interfaces. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The input / output interfaces are used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, Near Field Communication (NFC), or other technologies. When executed by the processor, the computer program implements an identification method. The display unit is used to form a visually visible image and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0137] Those skilled in the art will understand that Figure 7 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0138] In one exemplary embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0139] In one exemplary embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above-described method embodiments.

[0140] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0141] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0142] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, database, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0143] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0144] The above embodiments are merely illustrative of several implementation methods of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of this application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. An identity recognition method, characterized in that, The method is applied to a controller of a guard device, the guard device further including a lidar and a microphone array, and the method includes: Multiple point cloud data frames acquired by the lidar are obtained, and motion detection processing is performed on the point cloud data frames to determine the moving target; The moving target is tracked by sound using the microphone array to obtain an audio beam signal; The audio beam signal is processed using a preset identity recognition model to obtain the identity recognition result.

2. The method according to claim 1, characterized in that, The motion detection processing of the point cloud data frame to determine the moving target includes: Perform differential analysis on at least two consecutive point cloud data frames to obtain at least one moving object; The at least one moving object is filtered to determine the moving target.

3. The method according to claim 1, characterized in that, The step of using the microphone array to perform sound tracking on the moving target and obtain an audio beam signal includes: Based on the moving target, determine the first azimuth angle and the first distance of the moving target relative to the guarding device; The directional parameters of the microphone array are adjusted according to the first azimuth angle and the first distance, and multiple sound signals collected by the microphone array are acquired. The audio beam signal is obtained by superimposing and enhancing multiple sound signals.

4. The method according to claim 1, characterized in that, The step of performing biometric recognition processing on the audio beam signal using a preset identity recognition model to obtain identity recognition results includes: The audio beam signal is subjected to category filtering processing to obtain the audio types; When the audio type is footsteps, the identity recognition model is used to extract features from the audio beam signal to obtain the audio features corresponding to the audio beam signal. The similarity calculation of audio features is performed using a preset member database to obtain the membership probability of the audio features and each tag in the member database; The identity recognition result is determined based on the membership probability and the preset recognition threshold.

5. The method according to claim 4, characterized in that, The method further includes: If the identity recognition result is that the recognition fails, based on the moving target, determine the second azimuth angle and the second distance of the moving target relative to the guarding device; Based on the second azimuth angle and the second distance, the guard device is controlled to stand up, turn towards the direction corresponding to the moving target, and visually confirm the moving target to obtain a visual confirmation result; If the visual confirmation result is unsuccessful, an alarm message is generated based on the visual confirmation result.

6. The method according to claim 5, characterized in that, The method further includes: If the visual confirmation result is successful, the guard device is switched to monitoring mode. In monitoring mode, the lidar is controlled to collect the point cloud data frames at a preset scanning frequency, and the microphone array is controlled to perform omnidirectional sound reception.

7. The method according to claim 5, characterized in that, The method further includes: If the visual confirmation result is satisfactory, the audio beam signal is input into the identity recognition model to extract the corresponding voiceprint feature vector; The corresponding identity tag of the moving target is obtained, and the voiceprint feature vector is associated with the identity tag and stored in the member database to obtain the updated member database; the updated member database is used to provide an identity comparison benchmark for subsequently acquired audio beam signals.

8. A controller, characterized in that, The controller is used to perform the identity recognition method as described in any one of claims 1-7.

9. A guard device, characterized in that, The guarding device includes a lidar, a microphone array, and the controller as described in claim 8.

10. The guarding device according to claim 9, characterized in that, The guarding equipment also includes: a night vision camera, a torque servo motor, and multiple joints; The torque servo motor and the plurality of joints are used to move based on motion control signals sent by the controller, so that the guarding device stands up and turns in the direction of the moving target; The night vision camera is used to collect visual information corresponding to the moving target and transmit it to the controller; The controller is also used to obtain a visual confirmation result based on the visual information.

Images