Non-contact child presence detection method and system based on distributed vehicle-mounted acoustic devices

Abstract

The application discloses a non-contact child existence detection method and system based on a distributed vehicle-mounted acoustic device, and the method comprises the following steps: based on a ZC sequence signal sent by a distributed multi-speaker of a vehicle, CIR signals of each microphone received by a multi-microphone system of the vehicle are calculated; the CIR signals of two microphones located at the same position are combined to enhance dynamic components, and according to the signal variance after the dynamic component enhancement, a first detection result of whether a child body movement exists is obtained; in the case that the first detection result is that the child body movement does not exist, after a self-correlation coefficient is used to select a breathing candidate path from a path set of the distributed multi-speaker-distributed multi-microphone, sound signals on all the breathing candidate paths are combined and enhanced, and according to the signals after the combination and enhancement, a second detection result of whether a child breathes is obtained. The application can effectively detect the existence of a child, and has the advantages of full coverage, low cost and high sensitivity.

Description

Technical Field

[0001] This invention belongs to the field of non-contact intelligent sensing, and specifically relates to a non-contact method and system for detecting the presence of children based on distributed vehicle-mounted acoustic devices. Background Technology

[0002] More and more new energy vehicles and autonomous vehicles are now equipped with audio systems containing a large number of acoustic components. Currently, the audio systems in smart cockpits are mainly used for voice control and in-vehicle entertainment, such as playing music and movies and providing a 3D immersive surround sound experience. However, this invention proposes that the distributed acoustic components in smart cockpits can not only be used for traditional voice-related functions, but also that in-vehicle sound wave perception can become a new type of perception method that can be directly implemented in new energy vehicles and autonomous vehicles.

[0003] A review of existing research on in-vehicle acoustic sensing revealed that most existing methods monitor driver fatigue by placing a mobile phone on the dashboard facing the driver, but their sensing coverage is limited and they struggle to monitor rear-seat passengers. Leveraging the unique advantages of distributed acoustic sensing (high deployment density and zero hardware cost), this invention proposes a novel method for detecting the presence of children, a fundamental function required for future autonomous vehicles. In summer, car interiors become extremely hot, and children's body temperature rises much faster than adults; when their body temperature reaches 40 degrees Celsius, they face a dangerous situation.

[0004] Existing child presence detection methods largely rely on various types of sensors, such as thermal and pressure sensors mounted on the car seat. When a child is in a car seat, the performance of these sensors degrades because they are not in direct contact with the child and the child's weight is small. Camera-based solutions are commonly used for driver fatigue detection; however, for child presence detection, a single camera cannot cover all areas, resulting in blind spots. Furthermore, camera-based solutions are easily affected by lighting conditions and can raise privacy concerns. Radio frequency-based solutions include the use of UWB and millimeter-wave radar for sensing; however, single radar devices still face blind spot problems, and using multiple devices is too costly. Therefore, this invention proposes a non-contact child presence detection system based on the existing distributed audio system in the vehicle, achieving full coverage detection of child presence within the vehicle without any additional hardware costs. Summary of the Invention

[0005] This invention addresses the technical problem of providing a non-contact child presence detection method and system based on distributed in-vehicle acoustic devices, requiring no additional hardware. The system utilizes distributed acoustic devices within a smart cockpit to detect subtle body movements and vital signs (breathing) of children left inside the vehicle. It proposes a multi-microphone signal combination technique and a multipath enhancement algorithm. By leveraging the abundant multipath signals within the car cockpit, multiple weak multipath signals are combined and enhanced using a multipath signal combination algorithm, achieving accurate and reliable non-contact detection of child movements and breathing within the smart cockpit to determine the child's presence. Furthermore, this method can effectively detect children in different locations within the vehicle, across different vehicle models, and in various scenarios. The designed system offers advantages such as full coverage, low cost, and high sensitivity.

[0006] To achieve the above objectives, the technical solution of the present invention includes the following contents.

[0007] A non-contact child presence detection method based on distributed vehicle-mounted acoustic devices includes:

[0008] Based on the ZC sequence signal transmitted by the vehicle-mounted distributed multi-speaker system, calculate the CIR signal of each microphone received by the vehicle-mounted multi-microphone system;

[0009] The CIR signals from two microphones located at the same position are combined for dynamic component enhancement, and the first detection result of whether there is a child's physical movement is obtained based on the variance of the enhanced signal.

[0010] If the first detection result indicates that there is no child movement, then the candidate breathing path is selected from the path set of distributed multi-speaker-distributed multi-microphone using the autocorrelation coefficient. The sound signals on all candidate breathing paths are then combined and enhanced. Based on the combined and enhanced signal, the second detection result indicating whether there is child breathing is obtained.

[0011] Furthermore, the calculation of the CIR signal of each microphone received by the in-vehicle multi-microphone system based on the ZC sequence signal transmitted by the in-vehicle distributed multi-speaker system includes:

[0012] The ZC sequence is used as the baseband signal for the loudspeaker, and the baseband signal is modulated to the high-frequency part before being transmitted.

[0013] After bandpass filtering the microphone's received signal, cross-correlation is performed between the filtered received signal of each frame and the signal transmitted by the speaker to obtain the CIR signal of that frame.

[0014] By combining CIR signals from consecutive frames, a distance-time spectrum is created for subsequent signal processing to determine the presence of a child inside the vehicle.

[0015] Furthermore, the received signal Where M represents the total number of speakers in the system, m = 1, 2, ..., M represents the speaker number, P m and i m Let represent the number of paths and the path number of the signal emitted by the m-th speaker to the microphone, respectively. and Representing path i respectively m The attenuation coefficient, path length, and phase shift caused by propagation / reflection, The received signal represents the delay between the received and transmitted signals, c represents the speed of sound, and f represents the speed of sound. s Zc represents the system sampling rate. T This indicates the signal transmitted by the loudspeaker after being modulated to a high frequency.

[0016] Furthermore, the CIR signal of the microphone

[0017] Furthermore, the CIR signals from the two microphones located at the same position are dynamically amplified, and a first detection result regarding the presence of child movement is obtained based on the variance of the amplified signal, including:

[0018] The microphone's CIR signal is normalized.

[0019] An enhanced dynamic signal is generated based on the CIR signals from two microphones at the same location.

[0020] Based on the enhanced dynamic signal, the signal variance of each path at each microphone position is calculated using a specific time window, and then all signal variances are summed to obtain the final signal variance.

[0021] The final signal variance is compared with a first set threshold to obtain a first detection result indicating whether a child's physical movement exists.

[0022] Furthermore, based on the CIR signals from two microphones at the same location, an enhanced dynamic signal is generated, including:

[0023] Align the CIR signals of two microphones at the same location;

[0024] The interference signal caused by other static reflective objects in the environment is eliminated from the microphone CIR signal by using a loop filter;

[0025] The enhanced dynamic signal is obtained by multiplying the CIR signals of two microphones at the same location by their dynamic components.

[0026] Furthermore, the alignment of the two microphone CIR signals at the same position includes:

[0027] Select two frames of signal from the distance-time spectrum composed of CIR signals from two microphones at the same location;

[0028] The time delay offset between these two frames of signals is calculated using cross-correlation.

[0029] The two frames are aligned by compensating for the delay in one of the microphone CIR signals.

[0030] Furthermore, after selecting candidate breathing paths from the distributed multi-speaker-distributed multi-microphone path set using autocorrelation coefficients, the sound signals on all candidate breathing paths are combined and enhanced, and a second detection result of whether a child is breathing is obtained based on the combined and enhanced signal, including:

[0031] The presence of respiratory signals on each path is determined by using the autocorrelation coefficient to obtain candidate respiratory paths;

[0032] By combining respiratory signals from multiple candidate respiratory pathways, a new respiratory signal is obtained;

[0033] Perform an autocorrelation operation on the new respiratory signal to calculate the respiratory rate;

[0034] The respiratory rate is compared with a second set threshold to obtain a second detection result indicating whether a child is breathing.

[0035] Furthermore, the combination of respiratory signals from multiple respiratory candidate pathways to obtain a new respiratory signal includes:

[0036] The dynamic phase change and signal energy of the respiratory signal on each candidate respiratory path are normalized.

[0037] Select a respiratory signal on the complex plane as a reference signal, and then rotate other respiratory signals to be in the same direction as the reference signal to obtain the same initial phase;

[0038] All adjusted respiratory signals are superimposed to obtain a new respiratory signal.

[0039] A non-contact child presence detection system based on distributed vehicle-mounted acoustic devices includes:

[0040] The CIR signal calculation module is used to calculate the CIR signal of each microphone received by the vehicle multi-microphone system based on the ZC sequence signal sent by the vehicle distributed multi-speaker.

[0041] The first detection module is used to combine the CIR signals of two microphones located at the same position for dynamic component enhancement, and obtain the first detection result of whether there is a child's body movement based on the variance of the signal after dynamic component enhancement.

[0042] The second detection module is used to select a breathing candidate path from the path set of distributed multi-speaker-distributed multi-microphone using the autocorrelation coefficient when the first detection result is that there is no child's physical movement. Then, it combines and enhances the sound signals on all breathing candidate paths and obtains a second detection result of whether a child is breathing based on the combined and enhanced signal.

[0043] The advantages of this invention compared to existing technologies are as follows: This invention utilizes distributed in-vehicle acoustic devices to send and receive sound wave signals, and achieves non-contact child presence detection through signal analysis and processing of the echoes. This invention fully utilizes the distributed acoustic devices already deployed in the vehicle, requiring no additional hardware, enabling convenient and efficient non-contact sensing applications. Simultaneously, the multi-microphone signal combination technology proposed in this invention fully utilizes signal similarity, effectively increasing the phase change of dynamic components; the multipath combination enhancement algorithm combines multiple weak multipath signals to obtain a signal with a higher signal-to-noise ratio, effectively solving the child presence detection problems in some complex scenarios (such as child safety seats obstructing the view, children covered with blankets, etc.). The child presence detection system proposed in this invention is effective in any location within the vehicle where a child might be present, achieving blind-spot-free detection, and has been verified in different vehicle models and scenarios, demonstrating the system's accurate perception capabilities. This invention is easy to promote and applicable to fields such as smart cockpits and intelligent sensing. Attached Figure Description

[0044] Figure 1 This is a flowchart corresponding to the system of the present invention.

[0045] Figure 2 This diagram shows the deployment of speaker and microphone components inside the vehicle.

[0046] Figure 3 This is a child simulator used for testing in an embodiment of the present invention.

[0047] Figure 4 This is the web interface for a non-contact child presence detection system based on distributed vehicle-mounted acoustic devices. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0049] The present invention discloses a non-contact child presence detection method based on distributed vehicle-mounted acoustic devices. First, a ZC sequence signal is transmitted using a distributed multi-speaker system and received by a distributed multi-microphone system. The received signal is then bandpass filtered to remove noise outside the desired range. Next, the transmitted and filtered received signals are cross-correlated to obtain the channel impulse response (CIR). Then, the CIR signals from two microphones located at the same position are dynamically amplified, and the variance of the amplified signal is calculated to determine the presence of child movement. If a child is detected, the system reports "child detected." If no child is detected, respiratory detection is further performed to determine the child's presence. Specifically, candidate respiratory paths are first selected using autocorrelation coefficients, and then a multipath enhancement algorithm is used to enhance the signal-to-noise ratio of the candidate respiratory signals. Finally, the respiratory frequency of the amplified signal is estimated. If the frequency is within the normal respiratory frequency range of a child, the system reports "child detected."

[0050] like Figure 1 As shown, this system mainly consists of three steps: signal preprocessing, child movement detection, and child breathing detection. The first step, signal preprocessing, mainly includes bandpass filtering of the received signal and CIR signal estimation. The second step, child movement detection, mainly includes signal normalization, dynamic signal enhancement, and signal variance calculation to determine the presence of child movement. The third step, child breathing detection, mainly includes candidate breathing path selection, multipath combination enhancement algorithm, and respiratory rate estimation.

[0051] The following will describe each step of the system in detail.

[0052] The implementation steps of this invention are as follows:

[0053] Step 1: Signal preprocessing.

[0054] The ZC (Zadoff-Chu) sequence is used as the baseband signal for transmission. The expression for generating the ZC sequence is as follows:

[0055]

[0056] Where u is the root sequence, u∈{1,2,…,(N)} zc -1)},N zc Let u and N be the length of the ZC sequence. zc The condition must be coprime, i.e., gcd(u,N) zc ) = 1, n represents the sequence position, n = 0, 1, ..., N zc -1, q is a positive integer constant, and j is the imaginary unit of the complex signal.

[0057] To prevent noise interference to children during the sound signal detection process, this invention modulates the baseband signal to a high frequency range of 18kHz-20kHz before transmission. The transmitted signal is denoted as zc. T [n].

[0058] Based on the characteristics of an in-vehicle distributed multi-speaker, multi-microphone system, this invention models the microphone's received signal as follows:

[0059]

[0060] Where M represents the total number of speakers in the system, m = 1, 2, ..., M represents the speaker number, P m and i m Let represent the number of paths and the path number of the signal emitted by the m-th speaker to the microphone, respectively. and Representing path i respectively m The attenuation coefficient, path length, and phase shift caused by propagation / reflection, This represents the delay of the received signal relative to the transmitted signal (c represents the speed of sound, f...). s (This represents the system sampling rate).

[0061] To obtain channel information, namely the channel impulse response (CIR) for sensing, this invention first performs bandpass filtering on the received signal, and then cross-correlates the filtered received signal with the transmitted signal for each frame. Finally, the CIRs of consecutive frames are combined to create a distance-time spectrum.

[0062] The mathematical expression of the CIR signal in a multi-speaker system is as follows:

[0063]

[0064] The above formula shows that the path length from the signal emitted by the m-th speaker to the microphone is... Thus at the sampling point An impulse response is generated. Since the ZC signal is limited by the frequency gate function, the corresponding time-domain CIR signal is the convolution of the sinc function and the impulse response.

[0065] Step 2: Child's physical activity assessment.

[0066] This invention notes that in distributed multi-microphone systems, at least two microphones are close to each other, and the signals received by these microphones are very similar. Therefore, this invention proposes a multi-microphone signal joint technique to enhance dynamic signals, the specific steps of which are as follows:

[0067] Step 1: CIR Signal Alignment. First, select two frames from the distance-time spectrum composed of CIR signals from two microphones at the same location. Calculate the time delay offset between these two frames using cross-correlation, and then align the two frames by compensating for the time delay of one of the microphones. Each time the system restarts, the offset only needs to be calculated once, and this offset is used to uniformly align all subsequent data.

[0068] Step 2: Static Interference Cancellation. The interior space of a vehicle is relatively small, and many reflective objects in the environment can affect the overall received signal. Therefore, it is necessary to pass the CIR signal through a loop filter to eliminate interference signals caused by other static reflective objects in the environment.

[0069] Step 3: CIR signal multiplication. This invention observes that the dynamic phase changes of the aligned CIR signals from the two microphones exhibit the same pattern (i.e., simultaneously reaching peaks / troughs). By multiplying these two dynamic components, the signal phase change caused by target motion can be increased. Let the two dynamic components from the two microphones be... and This invention can obtain CIR signal multiplication H CMM The result for (n,t):

[0070]

[0071] Among them, A d1 and A d2 These represent the attenuation coefficients of the two dynamic components, respectively. and Let represent the dynamic phases respectively. The above equation shows that the dynamic phase change of the new signal after multiplication becomes the sum of the dynamic phase changes of the original two microphones. More specifically, by combining the CIR signals of microphones placed in the same position to amplify the phase change, the ability to detect subtle target movements is further enhanced.

[0072] Based on the above-mentioned multi-microphone signal combination technology, the present invention can perform children's physical movement detection.

[0073] Compared to a scenario where there are no children in the car, a child's body movements (such as spontaneous body movements and arm / leg movements) will significantly alter the propagation channel of the sound signal. Therefore, a body movement detection threshold can be directly set. When the signal variance exceeds the threshold, a child is detected in the car. The detailed steps for body movement detection are as follows:

[0074] Step 1: CIR signal normalization. Considering the impact of various complex application environments (such as noisy parking lots, underground parking lots, etc.) on signal energy, the signal needs to be normalized before detecting body motion.

[0075] Step 2: Dynamic Signal Enhancement. The multi-microphone signal enhancement technique described above amplifies the impact of the child's subtle movements on the signal, resulting in a more significant difference in signal variance compared to when the child is stationary (sleeping) or when there are no children in the car.

[0076] Step 3: Signal Variance Calculation. Based on the enhanced dynamic signal, the signal variance for each path at each microphone location is further calculated using a specific time window (i.e., all 46 samples within 2 seconds). To fully utilize the signal characteristics of multipath and multiple microphones, this invention sums all signal variances, and the final signal variance is defined as:

[0077]

[0078] Where δ(l,c,t) represents the time-varying signal variance of the c-th path of the microphone at position l, L represents the number of different positions of the microphone, and C represents the number of propagation paths of the sound signal received by each microphone.

[0079] To select an appropriate threshold, this invention collected a large number of CIR signal samples from empty vehicles and vehicles with children moving inside. Calculation results show that the variance distribution in both cases is normally distributed, with no overlap between them. According to the 3σ rule, 99.73% of the data falls within three standard deviations of the mean (i.e., the range (μ-3σ, μ+3σ), where μ represents the mean and σ represents the standard deviation). To reduce the false alarm rate and improve the detection rate, this invention selects five times the upper boundary of the 3σ range (μ+3σ) as the threshold.

[0080] If the signal variance is greater than the threshold, it is assumed that a child is moving inside the vehicle, and the system issues an alarm "Child detected." Otherwise, the system proceeds to the next step, using respiratory detection to determine the presence of a child.

[0081] Step 3: Child's breathing test.

[0082] Compared to relatively large body movements, the slight chest rise and fall caused by a child's breathing is much weaker. Therefore, the reflected signal from the chest cavity is easily affected by environmental noise. This is especially true in some highly challenging scenarios (such as when a child's chest cavity is obstructed by a car seat). Therefore, this invention proposes a multipath signal combination enhancement algorithm to address these challenging detection cases. The detailed steps of the multipath signal combination enhancement algorithm are as follows:

[0083] Step 1: Breathing Path Selection. The reflection signals generated by chest cavity movements during a child's breathing are periodic. Therefore, the autocorrelation coefficient is used to determine whether a breathing signal exists on each path. When the autocorrelation coefficient exceeds a set threshold, the signal on that path is selected as a candidate breathing signal.

[0084] Step 2: Multipath Combination Enhancement Algorithm. Due to the confined space inside a vehicle, when a child's chest cavity is obstructed by other objects (car seats, blankets, etc.), the breathing signal will be very weak, or even drowned out by noise. Therefore, this invention proposes a multipath combination enhancement algorithm to enhance the signal-to-noise ratio and restore the breathing waveform by combining multiple weak breathing candidate signals. The specific algorithm steps are mainly divided into three steps:

[0085] (1) Phase normalization: Normalize the dynamic phase change and signal energy of each respiratory candidate signal;

[0086] (2) Phase alignment: Select a multipath signal as a reference in the complex plane, and then rotate other signals to be in the same direction as the reference signal, so as to obtain the same initial phase;

[0087] (3) Multipath superposition: Superimpose all multipath signals to achieve the effect of signal energy enhancement and noise reduction;

[0088] By using multipath combination enhancement algorithms, the intensity of dynamic signals can be improved and the impact of environmental noise can be reduced without changing the target motion information, thereby obtaining a signal with a higher signal-to-noise ratio.

[0089] Step 3: Respiratory Rate Estimation. An autocorrelation operation is performed on the new signal obtained from the multipath combination enhancement algorithm to calculate the respiratory rate. If this respiratory rate is within the normal respiratory rate range for a child (i.e., 12 BPM-40 BPM), the system issues a "Child Detected" alarm and reports the respiratory rate to help confirm the child's current health status. Only if neither Step 2 (Child Movement Detection) nor Step 3 (Child Respiratory Detection) detects a child will the system report "Child Not Detected".

[0090] Example Description

[0091] Embodiments of the present invention employ acoustic devices embedded in a real automobile to implement the system of the present invention. These acoustic devices include six Bose speakers and eight microphones. For example... Figure 2 As shown, in a typical deployment of automotive acoustic devices, six speakers are located on the dashboard under the windshield (2), the A-pillars (2), and the B-pillars (2). Eight microphones are located in three positions: a four-microphone linear array with a 1cm spacing is positioned above the rearview mirror, and the other microphones are located on the B-pillars near the speakers (two on each side). All speakers and microphones are connected to a sound card (PXUA216MB-DL2-M / XMOS) for playing and capturing sound data (transmit signal parameters: f). c=19kHz, B=2kHz, T=0.0427s). The sound card is then connected to a laptop to receive the audio signal in real time at a sampling rate of 48kHz and further process the audio signal using MATLAB. Figure 3 As shown, this invention uses two child simulators for experiments, both preset with a fixed respiratory rate. These high-fidelity simulators possess realistic neonatal characteristics and can provide a lifelike clinical response. Detection rate and false alarm rate are used as indicators to evaluate the overall performance of the child presence detection system, while mean absolute error is used to evaluate the accuracy of the child respiratory rate estimation.

[0092] After performing steps 1-3 above, the processing results are displayed on a web-based front-end interface, such as... Figure 4 As shown, the interface displays the breathing waveform and frequency (times / minute) extracted using the method of the present invention. The test scene is displayed in the upper left corner of the interface, the system detection results (child detected / child not detected) are displayed in the lower left corner of the interface, and the status of the child in the vehicle (breathing / body movement) is displayed in the lower right corner.

[0093] This invention collected a large amount of data on vehicles with and without children, displaying the detection rate and false alarm rate, respectively. The data collection process involved turning off the car's power and locking the doors, then activating distributed acoustic devices to collect sensor data, with each monitoring session lasting 1-2 minutes. This invention recorded approximately 130 hours of data in different types of vehicles (including SUVs and sedans) and in different scenarios (such as roadside parking lots and underground parking lots).

[0094] This invention demonstrates, through the application of child presence detection, that a child's presence can be detected in all locations within a vehicle cabin using their limb movements and even subtle breathing (when the child is asleep), without any blind spots. The invention proves that the proposed system can operate stably in various vehicle cabins, achieving an average detection rate of 97% and a false alarm rate consistently below 2% in diverse scenarios (including challenging scenarios such as those obstructed by blankets or car seats).

[0095] The above embodiments are provided merely for the purpose of describing the present invention and are not intended to limit the scope of the invention. The scope of the invention is defined by the appended claims. Various equivalent substitutions and modifications made without departing from the spirit and principles of the invention should be covered within the scope of the invention.

Claims

1. A non-contact method for detecting the presence of children based on distributed vehicle-mounted acoustic devices, characterized in that, The method includes: Based on the ZC sequence signal transmitted by the vehicle-mounted distributed multi-speaker system, calculate the CIR signal of each microphone received by the vehicle-mounted multi-microphone system; The CIR signals from two microphones located at the same position are combined for dynamic component enhancement, and the first detection result of whether there is a child's physical movement is obtained based on the variance of the enhanced signal. If the first detection result is that there is no child movement, the breathing candidate path is selected from the path set of distributed multi-speaker-distributed multi-microphone using the autocorrelation coefficient. The sound signals on all breathing candidate paths are combined and enhanced. Based on the combined and enhanced signal, the second detection result of whether there is child breathing is obtained. The calculation of the CIR signal of each microphone received by the in-vehicle multi-microphone system based on the ZC sequence signal transmitted by the in-vehicle distributed multi-speaker system includes: The ZC sequence is used as the baseband signal for the loudspeaker, and the baseband signal is modulated to the high-frequency part before being transmitted. After bandpass filtering the microphone's received signal, cross-correlation is performed between the filtered received signal of each frame and the signal transmitted by the speaker to obtain the CIR signal of that frame. By combining CIR signals from consecutive frames, a distance-time spectrum is created for subsequent signal processing to determine the presence of a child inside the vehicle; Wherein, the received signal ;in, This indicates the total number of speakers in the system. Indicates the speaker's serial number. and They represent the first The number of paths and path numbers from which a signal emitted by a speaker reaches a microphone. , and Representing paths The attenuation coefficient, path length, and phase shift caused by propagation / reflection, This indicates the delay of the received signal relative to the transmitted signal. Indicates the speed of sound. Indicates the system sampling rate. This indicates the signal transmitted by the loudspeaker after being modulated to a high frequency; Specifically, after selecting candidate breathing paths from the distributed multi-speaker-distributed multi-microphone path set using autocorrelation coefficients, the sound signals on all candidate breathing paths are combined and enhanced. Based on the combined and enhanced signals, a second detection result is obtained regarding whether a child is breathing, including: The presence of respiratory signals on each path is determined by using the autocorrelation coefficient to obtain candidate respiratory paths; By combining respiratory signals from multiple candidate respiratory pathways, a new respiratory signal is obtained; Perform an autocorrelation operation on the new respiratory signal to calculate the respiratory rate; The respiratory rate is compared with a second set threshold to obtain a second detection result indicating whether a child is breathing.

2. The method according to claim 1, characterized in that, The microphone's CIR signal .

3. The method according to claim 1, characterized in that, The CIR signals from two microphones located at the same position are dynamically amplified, and a first detection result of whether a child's body movement exists is obtained based on the variance of the amplified signal, including: The microphone's CIR signal is normalized. An enhanced dynamic signal is generated based on the CIR signals from two microphones at the same location. Based on the enhanced dynamic signal, the signal variance of each path at each microphone position is calculated using a specific time window, and then all signal variances are summed to obtain the final signal variance. The final signal variance is compared with a first set threshold to obtain a first detection result indicating whether a child's physical movement exists.

4. The method according to claim 3, characterized in that, An enhanced dynamic signal is generated based on the CIR signals from two microphones at the same location, including: Align the CIR signals of two microphones at the same location; The interference signal caused by other static reflective objects in the environment is eliminated from the microphone CIR signal by using a loop filter; The enhanced dynamic signal is obtained by multiplying the CIR signals of two microphones at the same location by their dynamic components.

5. The method according to claim 4, characterized in that, The two microphone CIR signals aligned at the same position include: Select two frames of signal from the distance-time spectrum composed of CIR signals from two microphones at the same location; The time delay offset between these two frames of signals is calculated using cross-correlation. The two frames are aligned by compensating for the delay in one of the microphone CIR signals.

6. The method according to claim 1, characterized in that, The combination of respiratory signals from multiple candidate respiratory pathways yields a new respiratory signal, including: The dynamic phase change and signal energy of the respiratory signal on each candidate respiratory path are normalized. Select a respiratory signal on the complex plane as a reference signal, and then rotate other respiratory signals to be in the same direction as the reference signal to obtain the same initial phase; All adjusted respiratory signals are superimposed to obtain a new respiratory signal.

7. A non-contact child presence detection system based on distributed vehicle-mounted acoustic devices, characterized in that, The system includes: The CIR signal calculation module is used to calculate the CIR signal of each microphone received by the vehicle multi-microphone system based on the ZC sequence signal sent by the vehicle distributed multi-speaker. The first detection module is used to combine the CIR signals of two microphones located at the same position for dynamic component enhancement, and obtain the first detection result of whether there is a child's body movement based on the variance of the signal after dynamic component enhancement. The second detection module is used to select a breathing candidate path from the path set of distributed multi-speaker-distributed multi-microphone using the autocorrelation coefficient when the first detection result is that there is no child's body movement. Then, it combines and enhances the sound signals on all breathing candidate paths and obtains a second detection result of whether there is a child breathing based on the combined and enhanced signal. The calculation of the CIR signal of each microphone received by the in-vehicle multi-microphone system based on the ZC sequence signal transmitted by the in-vehicle distributed multi-speaker system includes: The ZC sequence is used as the baseband signal for the loudspeaker, and the baseband signal is modulated to the high-frequency part before being transmitted. After bandpass filtering the microphone's received signal, cross-correlation is performed between the filtered received signal of each frame and the signal transmitted by the speaker to obtain the CIR signal of that frame. By combining CIR signals from consecutive frames, a distance-time spectrum is created for subsequent signal processing to determine the presence of a child inside the vehicle; Wherein, the received signal ;in, This indicates the total number of speakers in the system. Indicates the speaker's serial number. and They represent the first The number of paths and path numbers from which a signal emitted by a speaker reaches a microphone. , and Representing paths The attenuation coefficient, path length, and phase shift caused by propagation / reflection, This indicates the delay of the received signal relative to the transmitted signal. Indicates the speed of sound. Indicates the system sampling rate. This indicates the signal transmitted by the loudspeaker after being modulated to a high frequency; Specifically, after selecting candidate breathing paths from the distributed multi-speaker-distributed multi-microphone path set using autocorrelation coefficients, the sound signals on all candidate breathing paths are combined and enhanced. Based on the combined and enhanced signals, a second detection result is obtained regarding whether a child is breathing, including: The presence of respiratory signals on each path is determined by using the autocorrelation coefficient to obtain candidate respiratory paths; By combining respiratory signals from multiple candidate respiratory pathways, a new respiratory signal is obtained; Perform an autocorrelation operation on the new respiratory signal to calculate the respiratory rate; The respiratory rate is compared with a second set threshold to obtain a second detection result indicating whether a child is breathing.

Images