A method and device for identifying the identity of an occupant in a vehicle

Abstract

This invention relates to a method and device for identifying in-vehicle occupants, belonging to the field of in-vehicle occupant monitoring technology. The method includes: collecting facial features, face center coordinates, sound source coordinates, voiceprint features, and in-vehicle ambient noise intensity corresponding to the target seat position in the vehicle; obtaining a comprehensive occupant feature score based on the facial features, face center coordinates, sound source coordinates, voiceprint features, and in-vehicle ambient noise intensity; and determining whether the occupant at the target seat position is an adult or a child based on the comprehensive occupant feature score. This invention achieves child identification through dynamic weighted fusion of multi-dimensional sensor information, which can improve the accuracy of in-vehicle child identification, thereby restricting children's vehicle control operation permissions in the vehicle and effectively improving user safety.

Description

Technical Field

[0001] This invention belongs to the field of vehicle occupant monitoring technology, specifically relating to a method and device for identifying the identity of vehicle occupants. Background Technology

[0002] With the development of automotive intelligence, in-vehicle child safety interaction has become a key focus for the industry. The primary task of in-vehicle child safety interaction is child identification. Traditional in-vehicle child identification solutions mainly rely on single-camera or voiceprint recognition solutions. 1. Camera-based identification solutions rely on continuous video monitoring of the in-vehicle environment to identify dangerous child behaviors. Their drawbacks include: the need for real-time video stream processing, resulting in high computational and memory consumption; the complexity of the in-vehicle environment, children's faces being easily obscured by clothing and headwear, and significant influence from lighting and shooting angles, leading to low accuracy and unreliable identification of children relying solely on cameras. 2. Voiceprint-based identification solutions pre-record children's voiceprint features and match them during voice interaction to restrict children's vehicle control operations. However, these solutions have significant drawbacks: children are in the vocal cord development stage, and their voiceprint features are unstable, easily affected by emotions, health status, and age, resulting in low recognition rates; environmental noise such as wind and tire noise in the vehicle further interferes with voiceprint recognition; and the requirement to pre-record the voiceprints of all possible children riding in the vehicle is impractical in real-world applications.

[0003] In summary, existing technologies that rely solely on cameras or voiceprint recognition cannot accurately identify children in complex in-vehicle scenarios such as dynamic seat changes, facial occlusion, and environmental noise. This allows children's interaction rights to be unrestricted, which can easily lead to car accidents and results in low safety. Summary of the Invention

[0004] To address this issue, the present invention provides a method and device for identifying the identity of in-vehicle occupants, thereby solving the problem that solutions relying solely on cameras or voiceprint recognition cannot accurately identify children in complex in-vehicle scenarios such as dynamic seat changes, facial occupancy, and environmental noise. This leads to unrestricted children's interactive rights, which can easily cause vehicle accidents and result in low safety.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, the present invention provides a method for identifying the identity of occupants in a vehicle, comprising: Collect feature data corresponding to the target seat position inside the vehicle; the feature data includes facial features, face center coordinates, sound source coordinates, voiceprint features, and in-vehicle ambient noise intensity; Based on the aforementioned feature data, a comprehensive score for occupant characteristics is obtained; The occupant identity at the target seat position is determined based on the comprehensive score of the occupant characteristics; the occupant identity is either an adult or a child.

[0006] Furthermore, obtaining the comprehensive score of occupant characteristics based on the feature data includes: Facial feature values ​​are obtained based on the facial features, voiceprint feature values ​​are obtained based on the voiceprint features, and position matching values ​​are obtained based on the face center coordinates and sound source coordinates. Environmental noise interference values ​​are obtained based on the in-vehicle environmental noise intensity. The comprehensive score of the occupant characteristics is obtained based on the facial feature value, voiceprint feature value, location matching value, and environmental noise interference value.

[0007] Furthermore, the facial features are occupant facial images captured by an in-vehicle camera at the target seat position, and facial feature values ​​are obtained based on the occupant facial images, adult standard facial geometric feature reference values, and child standard facial geometric feature reference values. The voiceprint feature is the voice signal of the occupant in the target seat position collected by the vehicle microphone array, and the voiceprint feature value is obtained based on the voice signal, the adult standard voiceprint feature benchmark value and the child standard voiceprint feature benchmark value. The overall matching score is obtained based on the preset child seat coordinates, face center coordinates, and sound source coordinates, and the position matching score is obtained based on the overall matching score. The environmental noise interference value is obtained based on the ambient noise intensity when the vehicle is stationary and no one is speaking inside the vehicle, the ambient noise intensity when the vehicle is traveling at a preset speed with all four windows open to the maximum, and the ambient noise intensity inside the vehicle collected in real time.

[0008] Further, determining the occupant identity at the target seat position based on the comprehensive score of occupant characteristics includes: If the comprehensive score of the occupant characteristics is greater than or equal to the preset child characteristic threshold, then the occupant at the target seat position is a child; If the comprehensive score of the occupant characteristics is less than the preset child characteristic threshold, then the occupant at the target seat position is an adult.

[0009] Furthermore, the method also includes: If the comprehensive score of the occupant features is within a fuzzy preset range, the identity of the occupant in the vehicle is identified once every first preset time interval and the corresponding comprehensive score of the occupant features is obtained, until the number of identifications reaches a preset number, and the average of the comprehensive scores of the occupant features of the preset number of identifications is taken as the target comprehensive score of the occupant features. If the comprehensive score of the target occupant's characteristics is greater than or equal to the preset child characteristic threshold, then the occupant at the target seat position is a child. If the comprehensive score of the target occupant characteristics is less than the preset child characteristic threshold, then the occupant at the target seat position is an adult.

[0010] Furthermore, the method also includes: Feature data corresponding to the target seat position is collected every second preset time interval. If invalid feature data is collected for a consecutive preset number of times, the occupant identity of the target seat position is set to adult. If any of the facial features, face center coordinates, sound source coordinates, voiceprint features, and in-vehicle ambient noise intensity in the collected feature data is empty, the corresponding feature data is invalid.

[0011] Furthermore, the method also includes: If the occupant of the target seat position is a child, then the vehicle control operation corresponding to the target seat position is prohibited; If the occupant at the target seat position is an adult, then the vehicle control operation corresponding to the target seat position is permitted.

[0012] Furthermore, the method also includes: When the first door is detected to have completed the opening and closing operations within a third preset time period, the pressure value of the seat closest to the first door is collected. If the pressure value is greater than or equal to the preset pressure value, the position of the seat closest to the first door is set as the target seat position. At the same time, the identity of the occupants in the vehicle is identified, and the collected feature data, pressure value and occupant identity are stored as historical data of the target seat position.

[0013] Furthermore, the method also includes: When a passenger is detected speaking inside the vehicle, the coordinates of the sound source are collected and calculated. The seat position closest to the sound source coordinates is set as the target seat position. Pressure values ​​and historical data of the target seat position are collected. If the difference between the pressure value collected at the current target seat position and the pressure value in the historical data of the target seat position is greater than or equal to a preset pressure value, the passenger's identity is identified, and the collected feature data, pressure value, and passenger identity are stored as historical data of the target seat position. If the difference between the pressure value collected at the current target seat position and the pressure value in the historical data of the target seat position is less than the preset pressure value, the voiceprint features of the collected target seat position are compared with the voiceprint features in the historical data. If they are the same, the pressure value in the historical data is modified to the pressure value collected this time. If they are different, the passenger's identity is identified, and the collected feature data, pressure value, and passenger identity are stored as historical data of the target seat position.

[0014] Secondly, the present invention also provides a vehicle occupant identification device, the device comprising: The data acquisition module is used to collect feature data corresponding to the position of the target seat inside the vehicle; the feature data includes facial features, face center coordinates, sound source coordinates, voiceprint features, and the noise intensity of the in-vehicle environment; The comprehensive score calculation module is used to obtain the comprehensive score of occupant characteristics based on the feature data. The identification module is used to determine the identity of the occupant at the target seat position based on the comprehensive score of the occupant characteristics; the occupant identity is an adult or a child.

[0015] The present invention, by adopting the above technical solution, has at least the following beneficial effects: This invention provides a method and device for identifying in-vehicle occupants. It collects facial features, face center coordinates, sound source coordinates, voiceprint features, and in-vehicle ambient noise intensity corresponding to the target seat position. Based on these factors, a comprehensive occupant feature score is obtained. The score is then used to determine whether the occupant at the target seat position is an adult or a child. This invention uses dynamic weighted fusion of multi-dimensional sensor information to determine child identity, improving the accuracy of in-vehicle child identification and thus restricting children's vehicle control permissions, effectively enhancing user safety.

[0016] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit the invention. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating an exemplary embodiment of the present invention of a method for identifying the identity of a vehicle occupant; Figure 2 This is a flowchart illustrating an exemplary embodiment of the present invention for obtaining a comprehensive score of occupant characteristics based on feature data; Figure 3 This is a flowchart illustrating the determination of passenger identity as shown in an exemplary embodiment of the present invention; Figure 4 This is a schematic block diagram illustrating an in-vehicle occupant identification device according to an exemplary embodiment of the present invention.

[0019] The present invention will be further described below with reference to the accompanying drawings and specific embodiments. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be described in detail below. Obviously, the described embodiments are merely some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other implementation methods obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0021] This invention provides a method and device for identifying the identity of occupants in a vehicle. By using multiple dimensions such as facial features, face center coordinates, sound source coordinates, voiceprint features, and in-vehicle environmental noise intensity, the method can identify children in the vehicle, thereby improving the accuracy of child identification and restricting children's vehicle control operation permissions, effectively improving user safety.

[0022] The methods and apparatus of the present invention will be described below through specific embodiments.

[0023] Please see Figure 1 , Figure 1 This is a flowchart illustrating a method for identifying the identity of vehicle occupants according to an exemplary embodiment of the present invention. See also... Figure 1 The method includes: Step S11: Collect feature data corresponding to the target seat position inside the vehicle; the feature data includes facial features, face center coordinates, sound source coordinates, voiceprint features, and in-vehicle ambient noise intensity; Step S12: Obtain the comprehensive score of occupant characteristics based on the feature data; Step S13: Determine the occupant identity of the target seat position based on the comprehensive score of occupant characteristics; the occupant identity is either an adult or a child.

[0024] It should be noted that the technical solution provided in this embodiment can be used in practice as a mini-program or a plugin loaded into existing in-vehicle systems or applications, or as a standalone application that implements in-vehicle occupant identification through external interfaces. Applicable scenarios include, but are not limited to, in-vehicle child identification.

[0025] Specifically, the in-vehicle data acquisition equipment includes a camera, a microphone array, and a seat pressure sensor. The camera is used to collect the facial features of the occupants and determine the coordinates of the center of the face. The microphone array distributed in the vehicle is used to collect the voiceprint features of the occupants. Combined with the TDOA (Time Difference of Arrival) algorithm, the coordinates of the sound source are obtained to achieve in-vehicle sound source localization. The microphone array is also used to collect ambient noise signals, and the seat pressure sensor is used to collect the pressure values ​​of the corresponding seat.

[0026] It is understood that the method provided in this embodiment collects facial features, face center coordinates, sound source coordinates, voiceprint features, and in-vehicle ambient noise intensity corresponding to the target seat position in the vehicle. Based on the facial features, face center coordinates, sound source coordinates, voiceprint features, and in-vehicle ambient noise intensity, a comprehensive score of occupant features is obtained. Based on the comprehensive score of occupant features, the identity of the occupant at the target seat position is determined as an adult or a child. By dynamically weighting and fusing multi-dimensional sensor information to achieve child identification, the accuracy of child identification in the vehicle can be improved, thereby restricting the child's vehicle control operation permissions in the vehicle and effectively improving the user's vehicle safety.

[0027] In practice, the feature data in step S11, "collecting feature data corresponding to the target seat position in the vehicle", includes facial features, face center coordinates, sound source coordinates, voiceprint features, and in-vehicle environmental noise intensity.

[0028] Specifically, when the first door is detected to have completed opening and closing operations within a third preset time period, the pressure value of the seat closest to the first door is collected. If the pressure value is greater than or equal to a preset pressure value, the seat position closest to the first door is set as the target seat position. Simultaneously, the identity of the occupants is identified, and the collected feature data, pressure values, and occupant identities are stored as historical data for the target seat position. When occupant speech is detected, the sound source coordinates are collected and calculated. The seat position closest to the sound source coordinates is set as the target seat position. The pressure value and historical data of the target seat position are collected. If the pressure value collected at the current target seat position is greater than or equal to the target seat position... If the difference between the pressure values ​​in the historical data of the target seat position and the preset pressure value is greater than or equal to the preset pressure value, then the identity of the occupant in the vehicle is identified, and the collected feature data, pressure values, and occupant identity are stored as historical data of the target seat position. If the difference between the pressure value collected at the current target seat position and the pressure value in the historical data of the target seat position is less than the preset pressure value, then the voiceprint features collected at the target seat position are compared with the voiceprint features in the historical data. If they are the same, the pressure value in the historical data is modified to the pressure value collected this time. If they are different, the identity of the occupant in the vehicle is identified, and the collected feature data, pressure values, and occupant identity are stored as historical data of the target seat position.

[0029] It should be noted that the third preset duration and preset pressure value are both set according to specific business needs. The pressure value in the historical data has an initial value, which is set according to specific business needs.

[0030] It should be noted that in Scenario 1: when the opening and closing of the first door is detected within a third preset time period, it is determined whether the pressure value collected by the pressure sensor at the corresponding seat position of the first door is greater than or equal to a preset pressure value (unit: kilopascals). If the collected pressure value is greater than or equal to the preset pressure value, the acquisition module is triggered to collect the facial features, face center coordinates, voiceprint features, sound source coordinates, in-vehicle ambient noise, seat position information, and corresponding pressure value of the occupant at the corresponding seat position of the first door. The collected data is then quantified and normalized. A dynamic weighted fusion algorithm is used to calculate a comprehensive score of child characteristics to determine whether the occupant in that seat position is a child. The system records the occupant's relevant information, including occupant identity (child / adult), facial features (face width / length ratio, eye distance / face width ratio, head shape dimensions), voiceprint features (fundamental frequency, effective speech duration, speech spectral width, formant distribution), seat position information collected by seat pressure sensors (driver's seat, front passenger seat, left rear seat, right rear seat) and corresponding pressure values. This data is stored in the system as historical data for that seat position. If the collected pressure value is less than a preset pressure value, the historical data for the seat position corresponding to the first door remains unchanged, and data collection is not triggered.

[0031] It should be noted that in scenario 2: when a passenger is detected speaking inside the vehicle, the target seat position is determined based on the passenger's sound source coordinates. The difference between the pressure value collected by the pressure sensor at the target seat position and the pressure value recorded in the historical data for that position is greater than or equal to a preset pressure value. If it is greater than or equal to the preset pressure value, the system will collect the passenger's facial features, face center coordinates, voiceprint features, sound source coordinates, in-vehicle ambient noise, seat position information, and corresponding pressure value. The collected data will be quantified and normalized, and a dynamic weighted fusion algorithm will be used to calculate a comprehensive score of child features to determine whether the passenger at that seat position is a child. The historical data for that position will then be updated. If the pressure value is less than the preset value, it is determined whether the voiceprint features collected at the current seat position are the same as the voiceprint features recorded at that position in the historical data. If they are the same, only the pressure value is updated in the historical data for the current seat position. If they are different, the facial features, face center coordinates, voiceprint features, sound source coordinates, in-vehicle ambient noise, seat position information, and corresponding pressure value of the occupant at the current seat position are collected. The collected data are quantified and normalized, and a dynamic weighted fusion algorithm is used to calculate the comprehensive score of the child features to determine whether the occupant at that seat position is a child, and the historical data for that position is updated.

[0032] It is understood that the technical solution provided in this embodiment can accurately identify which passengers are children when the passengers in the vehicle dynamically adjust their seats.

[0033] Please see Figure 2 , Figure 2This is a flowchart illustrating an exemplary embodiment of the present invention for obtaining a comprehensive score of occupant characteristics based on feature data. See also: Figure 2 Step S12, “obtaining the comprehensive score of occupant features based on feature data”, includes: Step S21, obtaining facial feature values ​​based on facial features, obtaining voiceprint feature values ​​based on voiceprint features, obtaining position matching values ​​based on the center coordinates of the face and the coordinates of the sound source, and obtaining environmental noise interference values ​​based on the ambient noise intensity inside the vehicle; Step S22, obtaining the comprehensive score of occupant features based on facial feature values, voiceprint feature values, position matching values, and environmental noise interference values.

[0034] Specifically, the occupant feature comprehensive score is obtained based on facial feature values, voiceprint feature values, location matching values, and environmental noise interference values. This includes: obtaining the occupant feature comprehensive score using a first formula based on facial feature values, voiceprint feature values, location matching values, and environmental noise interference values, where the first formula is: C score = w1×V face + w2×V voice + w3×S pos –α×N noise Among them, C score V is a comprehensive score for occupant characteristics. face V represents facial feature values. voice S is the voiceprint feature value; pos For position matching values; N noise This represents the environmental noise interference value; w 1、 w2 and w3 are the weighting coefficients for facial features, voiceprint features, and location matching, respectively; α is the noise correction coefficient for environmental noise interference.

[0035] It should be noted that w 1、 w2 and w3 are weighting coefficients that can be dynamically adjusted according to the actual in-vehicle scenario, and satisfy w1+w2+w3=100%; α is a noise correction coefficient that can be dynamically adjusted according to the ambient noise intensity.

[0036] Specifically, the facial features are the occupant's facial image at the target seat position, captured by an in-vehicle camera. Facial feature values ​​are obtained based on the occupant's facial image, adult standard facial geometric feature benchmarks, and child standard facial geometric feature benchmarks. This includes: preprocessing the occupant's facial image to obtain the target occupant's facial image; extracting the target occupant's facial geometric features using a pre-trained convolutional neural network model based on the target occupant's facial image; using the adult standard facial geometric feature benchmark as the first lower limit and the child standard facial geometric feature benchmark as the first upper limit; and obtaining the facial feature values ​​using a min-max normalization algorithm based on the target facial geometric features, the first lower limit, and the first upper limit.

[0037] It should be noted that, after capturing occupant facial images at the target seat position using an in-vehicle camera and performing image preprocessing (including grayscale conversion, noise reduction, lighting compensation, and face region cropping), a pre-trained CNN (Convolutional Neural Network) model is used to accurately extract key geometric features of the occupant's face, including the face width / face length ratio, eye distance / face width ratio, and head shape contour parameters. For the extracted multiple facial geometric features, a child facial feature discrimination model is constructed. The baseline values ​​of adult standard facial geometric features are used as the lower limit of the comparison, and the baseline values ​​of child standard facial geometric features are used as the upper limit of the comparison. Through a min-max normalization algorithm, the comprehensive score of multiple facial features is linearly mapped to the [0,1] interval. The normalized value is the facial feature value; where V... face The closer the value is to 1, the more the occupant's facial features resemble those of a child. face The closer the value is to 0, the more the occupant's facial features match those of an adult.

[0038] Specifically, the voiceprint feature is the voice signal of the occupant in the target seat position collected by the vehicle microphone array; the voiceprint feature value is obtained based on the voice signal, the adult standard voiceprint feature benchmark value, and the child standard voiceprint feature benchmark value, including: preprocessing the voice signal to obtain the target voice signal; obtaining the voiceprint bottom layer feature based on the target voice signal using the Mel frequency cepstral coefficient algorithm; using the adult standard voiceprint feature benchmark value as the second lower limit of comparison, and the child standard voiceprint feature benchmark value as the second upper limit of comparison, and obtaining the voiceprint feature value based on the voiceprint bottom layer feature, the second lower limit of comparison, and the second upper limit of comparison using the min-max normalization algorithm.

[0039] It should be noted that, after collecting the voice signal of the occupant at the target seat position through an in-vehicle microphone array and completing preprocessing (including pre-emphasis, frame-by-frame windowing, dereverberation, and separation of human voice from noise), the MFCC (Mel-Frequency Cepstral Coefficients) algorithm is used to extract the low-level features of the voiceprint, including the fundamental frequency of speech, effective phonation duration, speech spectral width, and formant distribution parameters. For the extracted multiple voiceprint features, a child voiceprint feature discrimination model is constructed. The lower limit of the comparison is based on the adult standard voiceprint feature benchmark, and the upper limit is based on the child standard voiceprint feature benchmark. The min-max normalization algorithm linearly maps the comprehensive score of the multiple voiceprint features to the [0,1] interval. The normalized value is the voiceprint feature value; where V... voice The closer the value is to 1, the more the occupant's voiceprint characteristics match those of a child. voice The closer the value is to 0, the more the occupant's voiceprint characteristics match those of an adult.

[0040] Specifically, a comprehensive matching degree value is obtained based on preset child seat coordinates, face center coordinates, and sound source coordinates. A location matching value is obtained based on the comprehensive matching degree value. This includes: obtaining preset child seat coordinates; obtaining a first deviation value based on face center coordinates and child seat coordinates; obtaining a second deviation value based on sound source coordinates and child seat coordinates; obtaining a comprehensive location matching degree based on the first deviation value and the second deviation value; and obtaining a location matching value based on the comprehensive location matching degree using a min-max normalization algorithm.

[0041] It should be noted that the preset child seat coordinates are set according to specific business needs. The system collects the voice signal of the occupant at the target seat position using an in-vehicle microphone array and combines it with the TDOA (Time Difference of Arrival) algorithm to obtain the sound source coordinates of the occupant. Simultaneously, the system obtains the center coordinates of the occupant's face using an in-vehicle camera. The system pre-stores the adult and child coordinate intervals for each seat (driver's seat, front passenger seat, left rear, right rear). The system calculates the deviation between the sound source coordinates and the corresponding child seat coordinates, and the deviation between the face center coordinates and the corresponding child seat coordinates. Combining these two deviations, the system calculates the overall matching degree of the sound source coordinates, face center coordinates, and the preset child seat coordinates. Using a min-max normalization algorithm, the overall matching degree is linearly mapped to the [0,1] interval. The normalized value is the position matching value; where S... pos The closer the value is to 1, the higher the overlap between the sound source coordinates, the face center coordinates, and the corresponding preset child coordinate range of the seat. pos The closer the value is to 0, the lower the overlap between the sound source coordinates, the center coordinates of the face, and the corresponding child's preset coordinate range of the seat.

[0042] Specifically, the environmental noise interference value is obtained based on the environmental noise intensity when the vehicle is stationary and no one is speaking inside the vehicle, the environmental noise intensity when the vehicle is traveling at a preset speed with all four windows open to the maximum, and the currently collected in-vehicle environmental noise intensity. This includes: using the environmental noise intensity when the vehicle is stationary and no one is speaking inside the vehicle as the third lower limit of comparison, using the environmental noise intensity when the vehicle is traveling at a preset speed with all four windows open to the maximum limit of comparison as the third upper limit of comparison, and obtaining the environmental noise interference value through a min-max normalization algorithm based on the in-vehicle environmental noise intensity, the third lower limit of comparison, and the third upper limit of comparison.

[0043] It should be noted that the preset speed is set according to specific business needs, and is generally set to the maximum actual vehicle speed by default; the current ambient noise intensity is obtained by synchronously collecting ambient noise signals inside the vehicle through the vehicle's microphone array. Using the noise floor in a quiet state as the lower limit and the extreme interference noise of the vehicle at high speed with the windows open as the upper limit, a min-max normalization algorithm is used to linearly map the ambient noise intensity to the [0,1] interval. The normalized value is the ambient noise interference value; where N... noise The closer the value is to 1, the stronger the noise interference in the current in-vehicle environment, and the greater the interference with voice signals and voiceprint recognition. N noise The closer the value is to 0, the quieter the in-car environment and the less interference.

[0044] It is understood that the technical solution provided in this embodiment quantifies facial features, voiceprint features, location matching, and environmental noise interference, providing data support for subsequent child identification calculations.

[0045] Please see Figure 3 , Figure 3 This is a flowchart illustrating the determination of occupant identity as shown in an exemplary embodiment of the present invention. See also: Figure 3 In step S13, “determine the occupant identity of the target seat position based on the comprehensive score of occupant characteristics” includes: if the comprehensive score of occupant characteristics is greater than or equal to the preset child characteristic threshold, then the occupant identity of the target seat position is a child; if the comprehensive score of occupant characteristics is less than the preset child characteristic threshold, then the occupant identity of the target seat position is an adult.

[0046] It should be noted that the preset threshold for children's characteristics is set according to specific business needs, and is generally set to 0.6 by default.

[0047] Specifically, if the comprehensive score of occupant features is within a fuzzy preset range, the identity of the occupant in the vehicle is identified once every first preset time interval and the corresponding comprehensive score of occupant features is obtained, until the number of identifications reaches a preset number. Then, the average of the comprehensive scores of occupant features from the preset number of identifications is taken as the target comprehensive score of occupant features. If the target comprehensive score of occupant features is greater than or equal to the preset child feature threshold, the occupant in the target seat position is a child. If the target comprehensive score of occupant features is less than the preset child feature threshold, the occupant in the target seat position is an adult.

[0048] It should be noted that the preset number of times, the first preset duration, and the fuzzy preset range are all set according to specific business needs. The fuzzy preset range can be [0.45, 0.6), the first preset duration is 3 seconds by default, and the preset number of times is 3 by default.

[0049] It should be noted that if the comprehensive score of the occupant features falls within a preset fuzzy range in a single determination, the occupant identity will be identified a preset number of times. The average of the comprehensive scores of the occupant features will be used as the final comprehensive score of the occupant features. The occupant identity will be determined by comparing the final comprehensive score of the occupant features with the preset child feature threshold.

[0050] Specifically, the method also includes: collecting feature data corresponding to the target seat position every second preset time interval; if invalid feature data is collected for a consecutive preset number of times, the occupant identity of the target seat position is set to adult; if any of the facial features, face center coordinates, sound source coordinates, voiceprint features and in-vehicle ambient noise intensity in the collected feature data is empty, the corresponding feature data is invalid feature data.

[0051] It should be noted that the second preset duration and number of collections are set according to specific business needs. The second preset duration is 3 seconds by default, and the number of collections is 3 times by default. If no valid features can be obtained after a series of collections (such as the occupant's face being continuously obscured or no voice input), the occupant in that seat position will be determined to be an adult by default to avoid system lag, and the user will be prompted to collect additional valid information.

[0052] In practice, this also includes: if the occupant of the target seat is a child, then vehicle control operations corresponding to the target seat position are prohibited; if the occupant of the target seat is an adult, then vehicle control operations corresponding to the target seat position are permitted.

[0053] It should be noted that the vehicle control operations corresponding to the target seat position include: window control (the most core control), door control, seat adjustment operation, air conditioning and air vent control, as well as sunroof control, panoramic sunroof control, and sliding control (not to be opened when a child is in the seat).

[0054] In one specific embodiment, a scenario is implemented where a child is alone in a fixed seat in a vehicle, without needing to change seats or experiencing significant environmental noise. The specific implementation process is as follows: When a child is accompanied by a guardian and sits in the right rear seat, after the guardian closes the right rear door, the pressure value collected by the pressure sensor of the right rear seat is greater than the preset child characteristic threshold (e.g., 10 kPa), and the system detects that the "door opening-closing" operation is completed within time T (e.g., 10 seconds), the identification process of the occupant is triggered: the in-vehicle camera is activated, and the child's facial features are collected in the right rear seat area. Feature parameters such as face width / face length ratio, eye distance / face width ratio, and head shape outline size are extracted and quantified into facial feature values ​​(e.g., 0.7); the child issues a voice command (e.g., "open the front window, open the window"), and the vehicle microphone array collects the voiceprint features of the voice (voice fundamental frequency, effective speech duration, voice spectrum width, formant distribution), quantifies them into voiceprint feature values ​​(e.g., 0.65), and combines them with the TDOA (Time Difference of Arrival) algorithm to obtain the sound source coordinates of the occupant. Simultaneously, the in-vehicle camera acquires the center coordinates of the occupant's face; based on the sound source coordinates, the occupant is located in the right rear seat. Combining this with the pre-stored coordinate range of the right rear seat for children, the deviations between the sound source coordinates and the right rear seat child coordinate range, and between the face center coordinates and the right rear seat child coordinate range are calculated. These two deviations are then combined to calculate a position matching value (e.g., 0.9). The in-vehicle microphone array simultaneously collects in-vehicle ambient noise, quantifying it as an ambient noise interference value (e.g., 0.1). Based on the current ambient noise intensity, a noise correction coefficient α = 0.5 is set, and weighting coefficients w1 = 0.4, w2 = 0.3, and w3 = 0.3 are dynamically allocated (satisfying w1 + w2 + w3 = 100%). According to formula C... score = 0.4×0.7 + 0.3×0.65 + 0.3×0.9 - 0.5×0.1, calculate C score =0.695; preset child characteristic threshold M=0.6, due to C score =0.695>M, determining that the occupant in the right rear seat is a child, the main control module outputs a vehicle control restriction command to the vehicle control execution and feedback module, which then restricts the right rear seat occupant from using voice control to operate the windows. If the child bends down to play and obscures their face, the facial feature collected in that single instance is invalid, and the calculated C... score =0.5 (within the preset fuzzy range [0.45, 0.6)), triggering the fault tolerance mechanism, the acquisition module re-acquires the child's voiceprint and facial features within 3 seconds, and recalculates C. score The average of three consecutive calculations is taken as the final comprehensive score of occupant characteristics, and the child's identity determination result is stably output based on the final comprehensive score of occupant characteristics.

[0055] In one specific embodiment, a scenario where a passenger dynamically changes their seat position and the passenger's identity needs to be updated in real time, is implemented as follows: Initial state: An adult is in the left rear seat, and a child is in the right rear seat. The in-vehicle system has already recorded the relevant information of the occupant in the left rear seat (identity: adult, facial feature value = 0.3, voiceprint feature value = 0.2, seat position information (left rear), pressure value 25 kPa) and the relevant information of the occupant in the right rear seat (identity: child, facial feature value = 0.7, voiceprint feature value = 0.65, seat position information (right rear), pressure value 15 kPa) after processing in the previous embodiment. After a period of time, the adult in the left rear seat and the child in the right rear seat switch seats. After the switch, the pressure value collected by the pressure sensor of the left rear seat is 15 kPa, which is 10 kPa different from the historically stored left rear pressure value (25 kPa), exceeding the preset threshold Y (e.g., 5 kPa). At this time, the child in the left rear seat issues a voice command (e.g., "open the front window, open the window"). The system locates the passenger speaking in the left rear seat based on the sound source coordinates, and the pressure difference in the left rear seat is greater than the threshold Y, triggering the identity recognition process in scenario 2. The in-vehicle camera is activated to capture the facial features of the left rear seat occupant (child), and the requantization value is set to 0.7. Simultaneously, the voiceprint features are captured and quantized to a value of 0.68. The sound source coordinates are located to the left rear, with a position matching value of 0.92 and an environmental noise interference value of 0.12. Weighting coefficients w1=0.45, w2=0.25, w3=0.3, and α=0.5 are adjusted. Calculate C. score =0.45×0.7 + 0.25×0.68 + 0.3×0.92 - 0.5×0.12=0.701, which is greater than the threshold of 0.6, so the left rear seat occupant is determined to be a child, and the left rear seat occupant's voice control of the windows is restricted. Simultaneously, the pressure value collected by the right rear seat pressure sensor is 25 kPa, which differs from the historically stored right rear pressure value (15 kPa) by 10 kPa, exceeding the threshold Y. Furthermore, an adult in the right rear seat issues a voice command, triggering the identity recognition process; the camera is activated to collect the facial features of the right rear seat occupant (adult), V. face =0.3, the collected voiceprint features are quantized to V voice =0.2, S pos =0.9, N noise =0.12, with the weighting coefficients unchanged, calculate V. voice =0.45×0.3 + 0.25×0.2 + 0.3×0.9 - 0.5×0.12=0.395, which is less than the threshold of 0.6 and not within the preset fuzzy range. Therefore, the occupant of the right rear seat is determined to be an adult, and vehicle control permissions for the right rear seat are granted. The relevant information of the occupants of the left and right rear seats is updated to ensure the accuracy of subsequent identity recognition.

[0056] In one specific embodiment, for scenarios with high environmental noise, such as high-speed driving, the specific implementation process is as follows: The occupant, seated in the left rear seat, has completed initial identification and had their information recorded. They have been identified as a child, and vehicle control access for the left rear seat has been restricted. The vehicle is traveling at high speed, generating significant wind and tire noise inside the vehicle. The environmental noise interference value N collected by the onboard microphone array... noise =0.4. If the original weighting coefficient is still used at this point, it will lead to increased interference from voiceprint features, affecting the accuracy of recognition. According to the ambient noise intensity, the weighting coefficient is dynamically adjusted, reducing the weight of voiceprint features (from 0.3 to 0.15), increasing the weight of facial features (from 0.4 to 0.5) and the weight of location matching value (from 0.3 to 0.35), and adjusting the noise correction coefficient α to 0.65. The left rear passenger (child) issues a voice command, and the facial feature value = 0.92, voiceprint feature value = 0.75, and location matching value S are collected and calculated. pos =0.93, calculate C according to the formula. score =0.5×0.92 + 0.15×0.75 + 0.35×0.93 - 0.65×0.4=0.638; C score If the noise level is greater than 0.6 (threshold M), the vehicle will still stably identify the occupant of the left rear seat as a child, avoiding misjudgments caused by environmental noise and ensuring the continuity and safety of vehicle control restrictions; when the vehicle decelerates, the environmental noise level decreases (N). noise =0.15), the main control module automatically adjusts the weight coefficients to restore them to normal values, ensuring a balance between recognition efficiency and accuracy.

[0057] Please see Figure 4 , Figure 4 This is a schematic block diagram illustrating an in-vehicle occupant identification device according to an exemplary embodiment of the present invention. See also: Figure 4 The occupant identification device 100 includes: The acquisition module 101 is used to acquire feature data corresponding to the position of the target seat in the vehicle; the feature data includes facial features, face center coordinates, sound source coordinates, voiceprint features, and in-vehicle ambient noise intensity; The comprehensive score calculation module 102 is used to obtain the comprehensive score of occupant characteristics based on the feature data; The identification module 103 is used to determine the identity of the occupant at the target seat position based on the comprehensive score of the occupant characteristics; the occupant identity is an adult or a child.

[0058] It should be noted that the technical solution provided in this embodiment is applicable to scenarios including but not limited to: in-vehicle child recognition.

[0059] Specifically, the in-vehicle data acquisition equipment includes a camera, a microphone array, and a seat pressure sensor. The camera is used to collect the facial features of the occupants and determine the coordinates of the center of the face. The microphone array distributed in the vehicle is used to collect the voiceprint features of the occupants. Combined with the TDOA (Time Difference of Arrival) algorithm, the coordinates of the sound source are obtained to achieve in-vehicle sound source localization. The microphone array is also used to collect ambient noise signals, and the seat pressure sensor is used to collect the pressure values ​​of the corresponding seat.

[0060] It is understood that the device provided in this embodiment collects facial features, face center coordinates, sound source coordinates, voiceprint features, and in-vehicle ambient noise intensity corresponding to the target seat position in the vehicle. Based on the facial features, face center coordinates, sound source coordinates, voiceprint features, and in-vehicle ambient noise intensity, it obtains a comprehensive score of occupant features. Based on the comprehensive score of occupant features, it determines whether the occupant at the target seat position is an adult or a child. By dynamically weighting and fusing multi-dimensional sensor information to achieve child identification, it can improve the accuracy of child identification in the vehicle, thereby restricting children's vehicle control operation permissions in the vehicle and effectively improving user safety.

[0061] 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, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0062] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0063] It should also 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 display, data used for analysis, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0064] The various embodiments in this specification are described in a related manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0065] 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 specification.

[0066] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. 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 patent application should be determined by the appended claims.

Claims

1. A method for identifying the identity of a vehicle occupant, characterized in that, include: Collect feature data corresponding to the target seat position inside the vehicle; the feature data includes facial features, face center coordinates, sound source coordinates, voiceprint features, and in-vehicle ambient noise intensity; Based on the aforementioned feature data, a comprehensive score for occupant characteristics is obtained; The occupant identity at the target seat position is determined based on the comprehensive score of the occupant characteristics; the occupant identity is either an adult or a child.

2. The method for identifying the identity of vehicle occupants according to claim 1, characterized in that, The process of obtaining a comprehensive score for occupant characteristics based on the feature data includes: Facial feature values ​​are obtained based on the facial features, voiceprint feature values ​​are obtained based on the voiceprint features, position matching values ​​are obtained based on the face center coordinates and sound source coordinates, and environmental noise interference values ​​are obtained based on the in-vehicle environmental noise intensity; the occupant feature comprehensive score is obtained based on the facial feature values, voiceprint feature values, position matching values, and environmental noise interference values.

3. The method for identifying the identity of vehicle occupants according to claim 2, characterized in that, The facial features are the occupant's facial image at the target seat position captured by the in-vehicle camera, and the facial feature value is obtained based on the occupant's facial image, adult standard facial geometric feature reference value and child standard facial geometric feature reference value; The voiceprint feature is the voice signal of the occupant in the target seat position collected by the vehicle microphone array, and the voiceprint feature value is obtained based on the voice signal, the adult standard voiceprint feature benchmark value and the child standard voiceprint feature benchmark value. The overall matching score is obtained based on the preset child seat coordinates, face center coordinates, and sound source coordinates, and the position matching score is obtained based on the overall matching score. The environmental noise interference value is obtained based on the ambient noise intensity when the vehicle is stationary and no one is speaking inside the vehicle, the ambient noise intensity when the vehicle is traveling at a preset speed with all four windows open to the maximum, and the ambient noise intensity inside the vehicle collected in real time.

4. The method for identifying the identity of vehicle occupants according to claim 1, characterized in that, Determining the occupant identity at the target seat position based on the comprehensive score of occupant characteristics includes: If the comprehensive score of the occupant characteristics is greater than or equal to the preset child characteristic threshold, then the occupant at the target seat position is a child; If the comprehensive score of the occupant characteristics is less than the preset child characteristic threshold, then the occupant at the target seat position is an adult.

5. The method for identifying the identity of vehicle occupants according to claim 1, characterized in that, The method further includes: If the comprehensive score of the occupant features is within a fuzzy preset range, the identity of the occupant in the vehicle is identified once every first preset time interval and the corresponding comprehensive score of the occupant features is obtained, until the number of identifications reaches a preset number, and the average of the comprehensive scores of the occupant features of the preset number of identifications is taken as the target comprehensive score of the occupant features. If the comprehensive score of the target occupant's characteristics is greater than or equal to the preset child characteristic threshold, then the occupant at the target seat position is a child. If the comprehensive score of the target occupant characteristics is less than the preset child characteristic threshold, then the occupant at the target seat position is an adult.

6. The method for identifying the identity of vehicle occupants according to claim 1, characterized in that, The method further includes: Feature data corresponding to the target seat position is collected every second preset time interval. If invalid feature data is collected for a consecutive preset number of times, the occupant identity of the target seat position is set to adult. If any of the facial features, face center coordinates, sound source coordinates, voiceprint features, and in-vehicle ambient noise intensity in the collected feature data is empty, the corresponding feature data is invalid.

7. The method for identifying the identity of a vehicle occupant according to any one of claims 4-6, characterized in that, The method further includes: If the occupant of the target seat position is a child, then the vehicle control operation corresponding to the target seat position is prohibited; If the occupant at the target seat position is an adult, then the vehicle control operation corresponding to the target seat position is permitted.

8. The method for identifying the identity of a vehicle occupant according to claim 1, characterized in that, The method further includes: When the first door is detected to have completed the opening and closing operations within a third preset time period, the pressure value of the seat closest to the first door is collected. If the pressure value is greater than or equal to the preset pressure value, the position of the seat closest to the first door is set as the target seat position. At the same time, the identity of the occupants in the vehicle is identified, and the collected feature data, pressure value and occupant identity are stored as historical data of the target seat position.

9. The method for identifying the identity of a vehicle occupant according to claim 8, characterized in that, The method further includes: When a passenger is detected speaking inside the vehicle, the coordinates of the sound source are collected and calculated. The seat position closest to the sound source coordinates is set as the target seat position. Pressure values ​​and historical data of the target seat position are collected. If the difference between the pressure value collected at the current target seat position and the pressure value in the historical data of the target seat position is greater than or equal to a preset pressure value, the passenger's identity is identified, and the collected feature data, pressure value, and passenger identity are stored as historical data of the target seat position. If the difference between the pressure value collected at the current target seat position and the pressure value in the historical data of the target seat position is less than the preset pressure value, the voiceprint features of the collected target seat position are compared with the voiceprint features in the historical data. If they are the same, the pressure value in the historical data is modified to the pressure value collected this time. If they are different, the passenger's identity is identified, and the collected feature data, pressure value, and passenger identity are stored as historical data of the target seat position.

10. A vehicle occupant identification device, characterized in that, The device includes: The data acquisition module is used to collect feature data corresponding to the position of the target seat inside the vehicle; the feature data includes facial features, face center coordinates, sound source coordinates, voiceprint features, and the noise intensity of the in-vehicle environment; The comprehensive score calculation module is used to obtain the comprehensive score of occupant characteristics based on the feature data. The identification module is used to determine the identity of the occupant at the target seat position based on the comprehensive score of the occupant characteristics; the occupant identity is an adult or a child.

Images