A millimeter-wave radar-based auxiliary interrogation system

Abstract

This application discloses an auxiliary interrogation system based on millimeter-wave radar, relating to the field of interrogation auxiliary analysis technology. The system includes an audio acquisition module, a millimeter-wave radar monitoring module, a text generation module, a time synchronization module, an anomaly analysis module, a correlation analysis module, and a result generation module. The audio acquisition module collects interrogation audio data; the millimeter-wave radar monitoring module non-contactly detects the interrogated person and generates respiratory data and / or heart rate data; the text generation module generates interrogation text data containing time information; the time synchronization module establishes the time correspondence between audio, text, and physiological sign data; the anomaly analysis module identifies time periods of abnormal physiological states; the correlation analysis module determines the corresponding interrogation content; and the result generation module generates auxiliary interrogation analysis results. This reduces interference from contact monitoring and improves the efficiency of correlation analysis between physiological abnormalities and interrogation content.

Description

Technical Field

[0001] This application relates to the field of interrogation support analysis technology, and in particular to an interrogation support system based on millimeter-wave radar. Background Technology

[0002] With the continuous improvement of information technology and intelligence in law enforcement and case handling, audio and video recording, text transcription, and data retention during interrogation have gradually become important auxiliary means for case handling and interrogation management. During interrogation, the language content, hesitation in responses, emotional changes, and physiological changes of the interrogated person can all provide reference for interrogators to judge the focus of questions and answers, adjust interrogation strategies, and conduct post-interrogation analysis. Therefore, how to continuously collect, record, and analyze relevant status information during interrogation without significantly interfering with the interrogation process has become an important research direction in the field of auxiliary interrogation technology.

[0003] In existing technologies, interrogation processes typically rely primarily on audio and video recording equipment to document the questions and answers, with interrogators assessing the interrogated person's condition based on on-site observations or post-interrogation review. For monitoring physiological indicators such as respiration and heart rate, traditional methods often require contact sensors such as electrodes, finger clips, or chest straps. These methods generally suffer from lengthy preparation times, cumbersome operation, and the potential to elicit resistance or psychological suggestion from the interrogated person, potentially affecting the naturalness of the interrogation process. Furthermore, interrogation audio recordings, written records, and physiological monitoring data are usually collected and stored separately, lacking a unified time correlation mechanism. This necessitates manual comparison of specific questions and answers with physiological fluctuations during subsequent analysis, resulting in low efficiency and a high risk of error. Moreover, interrogators struggle to continuously and accurately monitor subtle or instantaneous physiological changes during interrogation; the detection and marking of abnormal states rely heavily on manual experience, making it difficult to promptly generate analytical materials corresponding to specific questions and answers.

[0004] Therefore, in the process of assisting interrogation, contact physiological monitoring is prone to interfering with the natural state of interrogation, has poor temporal correlation between interrogation content and physiological data, and has low efficiency in identifying abnormal physiological states and generating related materials, which have become problems that urgently need to be solved. Summary of the Invention

[0005] This application provides an auxiliary interrogation system based on millimeter-wave radar, aiming to solve the problems in the existing technology during auxiliary interrogation, such as the easy interference of contact physiological monitoring with the natural state of interrogation, the poor temporal correlation between interrogation content and physiological data, and the low efficiency of abnormal physiological state identification and related material generation.

[0006] This application provides an auxiliary interrogation system based on millimeter-wave radar, the system comprising: The audio acquisition module is used to collect interrogation audio data during the interrogation process; A millimeter-wave radar monitoring module is used to perform non-contact radar detection on the interrogated person and generate physiological sign data of the interrogated person based on the obtained radar echo signal. The physiological sign data includes respiratory data and / or heart rate data. The text generation module is used to perform speech recognition processing on the interrogation audio data to generate interrogation text data corresponding to the interrogation audio data. The interrogation text data includes text content and its corresponding time information. The time synchronization module is used to perform time synchronization processing on the interrogation audio data, the interrogation text data and the physiological sign data based on a unified time reference, establish the time correspondence between the interrogation audio data, the interrogation text data and the physiological sign data, and obtain synchronized interrogation data; Anomaly analysis module is used to analyze the physiological sign data based on preset physiological state discrimination conditions and identify the time period of abnormal physiological state of the interrogated person. The correlation analysis module is used to determine the target interrogation content corresponding to the abnormal physiological state time period from the synchronous interrogation data based on the abnormal physiological state time period and the time correspondence. The target interrogation content includes interrogation audio segments and / or interrogation text segments. The result generation module is used to generate auxiliary interrogation analysis results based on the abnormal physiological state time period, the physiological sign data corresponding to the abnormal physiological state time period, and the target interrogation content.

[0007] Optionally, in the above scheme, the millimeter-wave radar monitoring module is used for: The system transmits millimeter-wave radar signals to the target detection area where the interrogated person is located and receives radar echo signals reflected by the interrogated person. The radar echo signal is down-converted and analog-to-digital converted to obtain a radar digital signal. Range dimension analysis is performed on the radar digital signal to determine the target range unit corresponding to the interrogated person; The thoracic cavity micromotion signal is extracted based on the target distance unit, and the physiological sign data is generated based on the thoracic cavity micromotion signal.

[0008] Optionally, in the above scheme, the millimeter-wave radar monitoring module is used for: The quadrature I / Q signal is extracted from the target range cell, and the quadrature I / Q signal is de-DC processed to obtain the target quadrature I / Q signal; The target orthogonal I / Q signal is subjected to phase demodulation processing to obtain a phase-change signal; The phase change signal is unwrapped to obtain the thoracic displacement waveform; The thoracic displacement waveform is denoised and subjected to spectral analysis to obtain respiratory rate data and / or heart rate data, and the respiratory rate data and / or heart rate data are used as the physiological sign data.

[0009] Optionally, in the above scheme, the time synchronization module is used for: A unified time reference is initialized at the start of the interrogation to obtain the interrogation timeline; Based on the interrogation timeline, audio timestamps are configured for the audio frames in the interrogation audio data to obtain time-stamped audio data; Based on the interrogation timeline, physiological timestamps are configured for the physiological data points in the physiological sign data to obtain time-stamped physiological sign data. Based on the interrogation timeline, a text timestamp is configured for the text segments in the interrogation text data to obtain text data with timestamps; The time correspondence is established based on the audio timestamp, the physiological timestamp, and the text timestamp.

[0010] Optionally, in the above scheme, the text generation module is used for: Based on the audio timestamp, the interrogation audio data is divided into speech segments to obtain multiple audio segments to be identified; Speech recognition processing is performed on each of the audio segments to be identified to obtain the text content corresponding to each of the audio segments to be identified. Based on the start and end positions of each audio segment to be identified on the interrogation timeline, determine the start time information and end time information corresponding to each text content; The interrogation text data is obtained by binding the text content, the start time information, and the end time information.

[0011] Optionally, in the above scheme, the anomaly analysis module is used for: The physiological sign data are smoothed and filtered to obtain the physiological sign data to be judged; Determine respiratory and / or heart rate variation characteristics based on the physiological signs data to be identified; The respiratory change characteristics and / or the heart rate change characteristics are compared with the preset physiological state discrimination conditions to obtain the abnormality discrimination result; Abnormal physiological data points are determined based on the abnormality discrimination results, and the time period of the abnormal physiological state is determined based on the continuous abnormal physiological data points.

[0012] Optionally, in the above scheme, the anomaly analysis module is used for: The respiratory abnormality discrimination result is obtained by comparing the respiratory change characteristics with the respiratory rate threshold condition and / or the respiratory change amplitude threshold condition. Based on the comparison between the heart rate change characteristics and the heart rate frequency threshold conditions and / or heart rate change amplitude threshold conditions, the heart rate abnormality discrimination result is obtained; Based on the respiratory abnormality discrimination results and / or the heart rate abnormality discrimination results, determine the abnormality type corresponding to the abnormal physiological data point; Anomaly marker information is generated based on the anomaly type, the time of occurrence of the anomaly, and the duration of the anomaly.

[0013] Optionally, in the above scheme, the correlation analysis module is used for: The time range for associated searches is determined based on the time period of the abnormal physiological state. Based on the associated retrieval time range and the time correspondence, the corresponding interrogation audio segment is retrieved from the synchronous interrogation data to obtain the target audio segment; Based on the associated retrieval time range and the time correspondence, the corresponding interrogation text fragment is retrieved from the synchronous interrogation data to obtain the target text fragment; The abnormal physiological state time period, the abnormal marker information, the physiological sign data corresponding to the abnormal physiological state time period, the target audio segment, and the target text segment are bound together to obtain an abnormal event package.

[0014] Optionally, in the above scheme, the result generation module is used to: The abnormal event packets are sorted in chronological order to obtain an abnormal event sequence; An abnormal physiological state record is generated based on the abnormal event sequence, and the abnormal physiological state record includes at least one of the following: abnormal occurrence time, abnormal duration, abnormal type, and abnormal physiological sign data. The abnormal physiological state records, the corresponding physiological sign curve segments, and the corresponding target interrogation content are correlated to obtain the interrogation correlation analysis content. The auxiliary interrogation analysis results are generated based on the interrogation correlation analysis content. The auxiliary interrogation analysis results include at least one of the following: abnormal physiological state time period, abnormality type, corresponding physiological sign curve segment, target audio segment, and target text segment.

[0015] Optionally, the above solution may also include a human-computer interaction module and an installation support module; The human-computer interaction module is used to obtain the abnormal physiological state time period output by the abnormal analysis module, generate early warning display information, and synchronously display the early warning display information, the continuous change curve of the physiological sign data, and the target interrogation content corresponding to the abnormal physiological state time period to obtain real-time auxiliary interrogation prompt results. The mounting module is used to carry the millimeter-wave radar monitoring module and install the millimeter-wave radar monitoring module at a preset position in the interrogation site, so that the millimeter-wave radar monitoring module can perform non-contact radar detection towards the target detection area where the interrogated person is located.

[0016] Compared with the prior art, this application has at least the following beneficial effects: Based on further analysis and research of existing technical problems, this application recognizes that existing technologies, in assisting interrogation, suffer from issues such as contact-based physiological monitoring easily interfering with the natural state of the interrogation, poor temporal correlation between interrogation content and physiological data, and low efficiency in identifying abnormal physiological states and generating related materials. This application addresses these problems by setting up an audio acquisition module to collect interrogation audio data and using a millimeter-wave radar monitoring module to perform non-contact radar detection on the interrogated person to generate respiratory and / or heart rate data. This allows for the synchronous acquisition of interrogation questions and answers and the interrogated person's physiological characteristics without the need for contact sensors, thereby reducing the interference of contact-based acquisition on the natural state of the interrogation. Furthermore, a text generation module converts the interrogation audio data into interrogation text data containing text content and corresponding time information, and a time synchronization module establishes the time correspondence between the interrogation audio data, interrogation text data, and physiological characteristic data based on a unified time benchmark. This system establishes a relationship that allows previously independent audio recordings, textual content, and physiological changes to correspond on the same timeline. Based on this, an anomaly analysis module identifies abnormal physiological state time periods according to preset physiological state discrimination conditions. Then, a correlation analysis module determines the corresponding interrogation audio and / or text segments based on these abnormal physiological state time periods and their temporal correspondences, enabling the automatic location of abnormal physiological changes within specific interrogation content. Finally, a results generation module generates auxiliary interrogation analysis results by combining the abnormal physiological state time periods, corresponding physiological sign data, and target interrogation content. This reduces reliance on manual observation and comparison, improves the efficiency of abnormal physiological state identification, interrogation content correlation, and analysis material generation, and addresses the problems of contact-based physiological monitoring easily interfering with the natural state of interrogation, poor temporal correlation between interrogation content and physiological data, and low efficiency in abnormal physiological state identification and correlation material generation during auxiliary interrogation. Attached Figure Description

[0017] Figure 1 A block diagram of the module architecture of an auxiliary interrogation system based on millimeter-wave radar provided in one embodiment of this application; Figure 2 A block diagram of the module architecture of another millimeter-wave radar-based auxiliary interrogation system provided in one embodiment of this application; Figure 3 A schematic diagram of a physiological sign monitoring module provided in one embodiment of this application; Figure 4 This is a schematic diagram illustrating the workflow of the recording and vital signs association module provided in one embodiment of this application. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0019] In one embodiment, such as Figure 1 As shown, the millimeter-wave radar-based auxiliary interrogation system includes an audio acquisition module, a millimeter-wave radar monitoring module, a text generation module, a time synchronization module, an anomaly analysis module, a correlation analysis module, and a result generation module. These modules can be integrated into the same processing device or disposed in different hardware units, and data transmission can be achieved via wired or wireless communication. The processing device may include a central processing unit, a digital signal processor, a microcontroller, a field-programmable gate array, a graphics processor, an edge computing device, or a combination thereof.

[0020] An audio acquisition module is used to collect interrogation audio data during the interrogation process. Specifically, the audio acquisition module may include one or more microphones, or a directional microphone array. The audio acquisition module can be placed on an interrogation table, wall, ceiling, recording terminal, or other location where interrogation questions and answers can be effectively captured. After acquiring analog audio signals, the audio acquisition module can convert them into digital audio streams through an audio encoder, analog-to-digital converter, or audio acquisition card, and then send the digital audio streams as interrogation audio data to the processing device.

[0021] The millimeter-wave radar monitoring module is used for non-contact radar detection of the interrogated person. Specifically, the millimeter-wave radar monitoring module can be placed in front of, to the side of, on a wall, on a table, or other locations that face the interrogated person's chest area. The millimeter-wave radar monitoring module can use frequency-modulated continuous wave millimeter-wave radar or other millimeter-wave radars capable of detecting micro-movements in the human chest cavity. The millimeter-wave radar monitoring module extracts information about the micro-movements in the interrogated person's chest cavity by transmitting millimeter-wave radar signals and receiving radar echo signals reflected by the interrogated person, and generates physiological sign data accordingly. The physiological sign data includes respiratory data and / or heart rate data, and may further include respiratory rate, respiratory waveform, heart rate rate, heart rate waveform, respiratory variation amplitude, heart rate variation amplitude, signal quality indicators, etc.

[0022] The text generation module performs speech recognition processing on the interrogation audio data to generate corresponding interrogation text data. Specifically, the text generation module can call an automatic speech recognition model or speech recognition engine to perform endpoint detection, speech enhancement, acoustic feature extraction, speech recognition, and text post-processing on the interrogation audio data to obtain the interrogation text data. The interrogation text data includes text content and its corresponding time information. The time information can include the start time, end time, duration, or timestamp corresponding to the interrogation timeline of the text segment.

[0023] The time synchronization module is used to synchronize interrogation audio data, interrogation text data, and physiological characteristic data based on a unified time reference. Specifically, the unified time reference can be provided by a local high-precision timer, crystal oscillator clock, network time synchronization clock, or other time synchronization units. When the system starts an interrogation task, the time synchronization module initializes the interrogation time axis and maps audio frames from the interrogation audio data, physiological data points from the physiological characteristic data, and text segments from the interrogation text data to this interrogation time axis, thereby establishing the time correspondence between the interrogation audio data, interrogation text data, and physiological characteristic data, and obtaining synchronized interrogation data.

[0024] The anomaly analysis module analyzes physiological sign data based on preset physiological state discrimination conditions to identify time periods of abnormal physiological states in the interrogated person. These preset physiological state discrimination conditions may include thresholds for respiratory rate, heart rate, respiratory variability, heart rate variability, duration of continuous abnormalities, short-term abrupt changes, or a discrimination model based on historical data or individual baseline data. The anomaly analysis module can perform real-time analysis of physiological sign data or offline analysis after the interrogation concludes.

[0025] The correlation analysis module is used to determine the target interrogation content corresponding to the abnormal physiological state time period from the synchronous interrogation data based on the time correspondence between the abnormal physiological state time period and the time. The target interrogation content includes interrogation audio clips and / or interrogation text clips. Specifically, when the anomaly analysis module identifies a certain abnormal physiological state time period, the correlation analysis module can expand a preset time range forward and / or backward with the abnormal physiological state time period as the center to obtain the correlation retrieval time range, and retrieve the interrogation audio clips and interrogation text clips within this time range from the synchronous interrogation data.

[0026] The results generation module generates auxiliary interrogation analysis results based on the time period of the abnormal physiological state, the corresponding physiological sign data, and the target interrogation content. These auxiliary analysis results can include the time period of the abnormal physiological state, the type of abnormality, the duration of the abnormality, the corresponding respiratory curve segment, the corresponding heart rate curve segment, the interrogation audio segment, the interrogation text segment, timeline markers, and related explanations. The auxiliary interrogation analysis results can be output in the form of tables, timelines, graphs, event lists, report files, or a human-computer interaction interface.

[0027] This embodiment can collect respiratory and / or heart rate data using a millimeter-wave radar monitoring module without requiring the interrogated person to wear contact sensors, reducing interference from contact data collection on the natural state of the interrogation; it establishes a time correspondence between interrogation audio data, interrogation text data, and physiological sign data through a time synchronization module, so that changes in physiological signs can be correlated with specific interrogation content; it automatically identifies abnormal physiological state time periods and associates them with target interrogation content through an anomaly analysis module and a correlation analysis module, reducing the workload of manual observation and comparison; and it outputs auxiliary interrogation analysis results through a result generation module, improving the efficiency of data processing and abnormal state analysis during the interrogation process.

[0028] In one embodiment, the millimeter-wave radar monitoring module first transmits millimeter-wave radar signals towards the target detection area where the interrogator is located. The target detection area can be the space containing the interrogator's chest cavity or a detection range including the upper body of the interrogator. The millimeter-wave radar monitoring module can employ a linear frequency modulated continuous wave radar, whose transmitting antenna transmits radio frequency signals according to a preset frequency modulation slope; after the radar signal reaches the interrogator's body, it is reflected by the chest cavity, clothing, and body surface to form radar echo signals, which are received by the receiving antenna.

[0029] Upon receiving the radar echo signal, the millimeter-wave radar monitoring module performs down-conversion and analog-to-digital conversion on the radar echo signal to obtain a radar digital signal. Specifically, the millimeter-wave radar monitoring module can mix the received signal with the transmitted reference signal using a mixer to obtain an intermediate frequency (IF) signal; then, it samples the IF signal using an analog-to-digital converter to obtain a radar digital signal suitable for digital processing. To improve the stability of subsequent analysis, gain control, bandpass filtering, static clutter suppression, or abnormal sampling point removal can be applied to the radar digital signal.

[0030] Subsequently, the millimeter-wave radar monitoring module performs range dimension analysis on the radar digital signal to determine the target range cell corresponding to the interrogator. Specifically, a fast Fourier transform can be performed on the radar digital signal to obtain the reflection intensity distribution on different range cells; based on the peak reflection intensity, preset seating position, target distance range, or human target detection results, the target range cell where the interrogator is located is determined. If there are static reflective objects such as tables, chairs, and walls in the interrogation area, the influence of irrelevant reflections on the selection of the target range cell can be reduced through background modeling, static clutter filtering, or range gate screening.

[0031] After determining the target range cell, the millimeter-wave radar monitoring module extracts the chest cavity micromotion signal based on the target range cell. Since respiration and heartbeat cause minute displacements in the chest cavity, the phase information in the target range cell changes with these micromotions. The millimeter-wave radar monitoring module can continuously track phase changes, amplitude changes, or complex echo changes within the target range cell and use these changes as the chest cavity micromotion signal. Subsequently, the system can generate physiological sign data based on the chest cavity micromotion signal.

[0032] This embodiment further clarifies the processing chain of the millimeter-wave radar monitoring module, from radar signal transmission, echo reception, down-conversion and analog-to-digital conversion, to range dimension analysis and extraction of chest cavity micromotion signals, providing a clear signal processing foundation for the non-contact vital sign acquisition process. This implementation method enables the system to acquire chest cavity micromotion information related to respiration and heart rate without contact with the person being questioned, improving the naturalness and continuity of physiological sign data acquisition.

[0033] In one embodiment, the millimeter-wave radar monitoring module extracts orthogonal I / Q signals from the target range cell. The orthogonal I / Q signals can include in-phase signals I and quadrature signals Q, both of which characterize the amplitude and phase changes of the target range cell echo. Since radar hardware, environmental static reflections, and DC bias can affect the orthogonal I / Q signals, the system can first perform DC removal processing on the orthogonal I / Q signals to obtain the target orthogonal I / Q signals. DC removal processing can be implemented using methods such as mean subtraction, high-pass filtering, sliding window mean removal, or background reference signal subtraction.

[0034] After obtaining the target's orthogonal I / Q signals, the millimeter-wave radar monitoring module performs phase demodulation processing on the target's orthogonal I / Q signals to obtain the phase change signal. Specifically, the phase value can be calculated using an arctangent phase demodulation algorithm, for example, by calculating the phase angle based on the in-phase signal I and the orthogonal signal Q. Since the displacement caused by chest cavity micro-movements leads to periodic changes in the radar echo phase, the phase change signal can reflect the chest cavity micro-movement state of the person being interrogated.

[0035] Because the phase values ​​obtained from phase demodulation may exhibit periodic jumps, the millimeter-wave radar monitoring module performs phase unwrapping processing on the phase change signal to obtain a continuous chest cavity displacement waveform. Phase unwrapping processing can determine the phase difference between adjacent sampling points; when the phase difference exceeds a preset jump threshold, compensation is performed to restore the phase change signal to a continuously changing signal. Subsequently, based on the correspondence between phase change and displacement, the phase change signal can be converted into a chest cavity displacement waveform.

[0036] After obtaining the chest cavity displacement waveform, the millimeter-wave radar monitoring module performs denoising and spectral analysis on the waveform to obtain respiratory rate data and / or heart rate data. Denoising may include bandpass filtering, singular value decomposition, empirical mode decomposition, wavelet denoising, moving average filtering, or adaptive filtering. Spectral analysis may include fast Fourier transform, power spectrum estimation, or time-frequency analysis. Because the respiratory and heart rate components differ in frequency range, the system can extract respiratory rate data in the low-frequency range and heart rate data in the relatively high-frequency range, using the extracted results as physiological sign data.

[0037] This implementation can reduce the impact of DC bias, environmental interference and phase jump on the extraction of vital signs, making the obtained respiratory data and / or heart rate data more stable, thereby providing a reliable data foundation for subsequent identification of abnormal physiological states.

[0038] In one embodiment, the time synchronization module initializes a unified time base at the start of the interrogation to obtain the interrogation timeline. The unified time base can be provided by a local high-precision crystal oscillator, system clock, network time clock, or dedicated timer. To ensure time consistency between different data streams, the audio acquisition module, millimeter-wave radar monitoring module, text generation module, anomaly analysis module, and correlation analysis module can all use the interrogation timeline as the time marker reference.

[0039] The time synchronization module configures audio timestamps for audio frames in the interrogation audio data according to the interrogation timeline, resulting in timestamped audio data. Specifically, the audio acquisition module can acquire audio data at a fixed sampling rate and write it to the buffer in units of audio frames, audio packets, or audio segments; the time synchronization module records the time value on the current interrogation timeline when an audio frame is generated or received, as the audio timestamp of that audio frame. The audio timestamp can be an absolute time or a relative time relative to the start time of the interrogation.

[0040] The time synchronization module assigns physiological timestamps to physiological data points in the physiological sign data based on the interrogation timeline, resulting in timestamped physiological sign data. For example, the millimeter-wave radar monitoring module can output respiratory rate data and / or heart rate data at preset intervals, and the time synchronization module records the corresponding physiological timestamp when the physiological sign data is output. Physiological data points may include a single respiratory rate value, a single heart rate value, a physiological curve segment within a time window, or a signal quality index.

[0041] The time synchronization module configures text timestamps for text segments in the interrogation text data according to the interrogation timeline, resulting in timestamped text data. Text timestamps can include the start and end timestamps of the text segment. A text segment can correspond to a sentence, a response, a continuous speech, or a fragment of speech recognition results. For real-time speech recognition scenarios, text timestamps can be determined by the character-level, word-level, or sentence-level time boundaries output by the speech recognition engine; for offline speech recognition scenarios, they can be determined based on the start and end times of the audio segment and the alignment information of the recognition results.

[0042] After obtaining the audio timestamp, physiological timestamp, and text timestamp respectively, the time synchronization module establishes a time correspondence based on these timestamps. Specifically, audio data, text data, and physiological characteristic data can be written into an index database, using the timestamp as the index field; alternatively, a unified timeline data structure can be constructed to map various types of data to the same timeline. Through this time correspondence, when the system determines the time of occurrence of a certain physiological abnormality, it can quickly retrieve interrogation audio data and interrogation text data within the same or adjacent time range.

[0043] This embodiment addresses the problem of difficulty in aligning separately collected and stored interrogation audio data, interrogation text data, and physiological sign data. This approach enables unified management of different data streams within the same interrogation timeline, thereby improving the accuracy and retrieval efficiency of associating subsequent abnormal physiological state time periods with target interrogation content.

[0044] In one embodiment, the text generation module divides the interrogation audio data into multiple audio segments to be identified based on audio timestamps. Specifically, the text generation module can divide continuous interrogation audio data into multiple audio segments to be identified based on voice activity detection results, silence intervals, speaker switching, preset duration windows, or interrogation question-and-answer rounds. Each audio segment to be identified retains its start and end times on the interrogation timeline.

[0045] The text generation module performs speech recognition processing on each audio segment to be recognized, obtaining the corresponding text content. Speech recognition processing may include noise suppression, echo cancellation, speech enhancement, acoustic feature extraction, acoustic model recognition, language model correction, and text post-processing. In interrogation scenarios involving multiple people in conversation, the text generation module can also perform speaker separation or speaker tagging to distinguish the speech content of the questioner and the interrogated.

[0046] The text generation module determines the start and end times of each text content based on the start and end positions of each audio segment to be recognized on the interrogation timeline. If the speech recognition engine can output word-level time information, more granular time information can be configured for characters, words, or phrases in the text content; if the speech recognition engine only outputs segment-level results, the start and end times of the audio segments to be recognized can be used as the time information of the corresponding text content.

[0047] After obtaining the text content, start time information, and end time information, the text generation module binds these three elements to obtain the interrogation text data. The interrogation text data can be stored in a structured data format, including, for example, text segment numbers, text content, start time, end time, speaker identifier, recognition confidence level, and corresponding audio segment identifier. This structured interrogation text data can be accessed by the time synchronization module and the correlation analysis module.

[0048] This implementation method enables the subsequent correlation analysis module to directly retrieve text content within the corresponding time range after determining the time period of abnormal physiological state, reducing the time for manual transcription and manual search, and improving the accuracy of the correspondence between interrogation content and abnormal physiological reactions.

[0049] In one embodiment, the anomaly analysis module first performs smoothing filtering on the physiological sign data to obtain the physiological sign data to be identified. The smoothing filtering can be performed separately on the respiratory data and heart rate data, or it can be performed on the corresponding original waveforms of both. Specifically, moving average filtering, median filtering, low-pass filtering, band-pass filtering, Kalman filtering, or adaptive filtering can be used to reduce the impact of random noise, short-time artifacts, and radar signal quality fluctuations on anomaly identification.

[0050] The anomaly analysis module determines respiratory and / or heart rate variability characteristics based on the physiological signs to be identified. Respiratory variability characteristics may include respiratory rate, respiratory cycle, respiratory amplitude, rate of change of respiratory rate, respiratory rhythm stability, and amplitude of respiratory abrupt changes. Heart rate variability characteristics may include heart rate, rate of change of heart rate, amplitude of heart rate fluctuations, amplitude of short-term heart rate increases, and heart rate stability. The anomaly analysis module can calculate these characteristics using a sliding time window or in real-time as each physiological data point is updated.

[0051] The anomaly analysis module compares respiratory and / or heart rate variability characteristics with preset physiological state discrimination criteria to obtain anomaly discrimination results. These preset physiological state discrimination criteria can be set according to a general normal range or dynamically determined based on the interrogated person's baseline physiological level within a preset time period before the interrogation begins. For example, baseline respiratory and heart rate data can be collected first, and then individualized discrimination criteria can be determined based on the baseline mean, standard deviation, fluctuation range, or historical stable interval. Anomaly discrimination results can be represented as normal, suspected abnormal, or obvious abnormal, or as a numerical anomaly score.

[0052] Based on the anomaly detection results, the anomaly analysis module identifies abnormal physiological data points and determines abnormal physiological state time periods based on consecutive abnormal physiological data points. Specifically, when one or more consecutive physiological data points meet the anomaly conditions in the preset physiological state detection criteria, these physiological data points can be marked as abnormal physiological data points; when the continuous duration of abnormal physiological data points reaches a preset duration threshold, they can be merged into one abnormal physiological state time period. For multiple abnormal segments with short adjacent intervals, they can be merged into the same abnormal physiological state time period according to preset merging rules.

[0053] This implementation method can continuously analyze respiratory data and / or heart rate data, automatically capture short-term, subtle or continuous abnormal fluctuations, and improve the real-time performance and stability of abnormal physiological state identification.

[0054] In one embodiment, the anomaly analysis module compares respiratory change characteristics with respiratory rate threshold conditions and / or respiratory change amplitude threshold conditions to obtain a respiratory anomaly discrimination result. The respiratory rate threshold conditions may include an upper limit of respiratory rate, a lower limit of respiratory rate, or the magnitude of respiratory rate deviation from the baseline; the respiratory change amplitude threshold conditions may include the amount of change in respiratory rate per unit time, the amount of change in respiratory waveform amplitude, the degree of respiratory rhythm disorder, or the magnitude of respiratory abrupt changes. When the respiratory change characteristics meet the corresponding threshold conditions, the anomaly analysis module can generate a respiratory anomaly discrimination result.

[0055] The anomaly analysis module compares heart rate variability characteristics with heart rate frequency threshold conditions and / or heart rate variability amplitude threshold conditions to obtain a heart rate anomaly detection result. Heart rate frequency threshold conditions can include the upper limit of heart rate, the lower limit of heart rate, or the magnitude of heart rate deviation from the baseline; heart rate variability amplitude threshold conditions can include the short-term rate of heart rate increase, the amplitude of heart rate fluctuations, the duration of sustained heart rate increases, or the magnitude of sudden heart rate changes. When the heart rate variability characteristics meet the corresponding threshold conditions, the anomaly analysis module can generate a heart rate anomaly detection result.

[0056] The anomaly analysis module determines the anomaly type corresponding to the abnormal physiological data point based on the results of respiratory anomaly discrimination and / or heart rate anomaly discrimination. Anomaly types can include respiratory anomalies, heart rate anomalies, respiratory and heart rate synchronization anomalies, short-term abrupt changes, and persistent anomalies. If both respiratory and heart rate anomaly discrimination results exist within the same time frame, the anomaly type can be identified as a composite anomaly; if only respiratory or heart rate anomaly discrimination results exist, they can be identified as a single anomaly type.

[0057] The anomaly analysis module generates anomaly marker information based on the anomaly type, occurrence time, and duration. The anomaly occurrence time can be determined based on the timestamp of the first anomaly data point in the abnormal physiological data points, and the anomaly duration can be determined based on the time span of consecutive abnormal physiological data points. Anomaly marker information may include anomaly number, anomaly type, anomaly start time, anomaly end time, anomaly duration, anomaly peak value, anomaly intensity, corresponding physiological data point identifier, and anomaly confidence level. Anomaly marker information can be stored in a database and accessed by the correlation analysis module and the result generation module.

[0058] This implementation method can distinguish different types of abnormalities, such as respiratory abnormalities, heart rate abnormalities, and combined abnormalities, and clearly record the time of occurrence and duration of the abnormality, providing clear data identification for the subsequent generation of abnormal event packages and the generation of auxiliary interrogation analysis results.

[0059] In one embodiment, the association analysis module determines the association retrieval time range based on the abnormal physiological state time period. The association retrieval time range can be completely consistent with the abnormal physiological state time period, or it can be extended forward by a first preset duration and backward by a second preset duration based on the abnormal physiological state time period. Based on the association retrieval time range and time correspondence, the association analysis module retrieves the corresponding interrogation audio segments from the synchronized interrogation data to obtain the target audio segment. Specifically, the association analysis module can retrieve audio frames or audio segments whose audio timestamps fall within the association retrieval time range from the timestamped audio data, and splice or truncate them to form the target audio segment. The target audio segment may include the questioning content before the anomaly occurred, the answering content when the anomaly occurred, and the short dialogue content after the anomaly occurred.

[0060] The correlation analysis module retrieves corresponding interrogation text fragments from synchronized interrogation data based on the correlation retrieval time range and time correspondence, thus obtaining the target text fragment. Specifically, the correlation analysis module can retrieve text segments from timestamped text data whose start time, end time, or time span overlaps with the correlation retrieval time range, and use the retrieved text segments as the target text fragment. The target text fragment may include the corresponding speaker identifier, text content, start time information, and end time information.

[0061] After obtaining the target audio and text segments, the correlation analysis module binds the abnormal physiological state time period, abnormal marker information, physiological sign data corresponding to the abnormal physiological state time period, the target audio segment, and the target text segment to obtain an abnormal event package. The abnormal event package can be stored in a structured data format, including event number, abnormal type, abnormal time range, respiratory curve segment, heart rate curve segment, audio segment path, text segment content, text time range, and search window parameters. The abnormal event package can serve as the basic unit for the results generation module to generate auxiliary interrogation analysis results.

[0062] This implementation method can generate abnormal event packages indexed by abnormal physiological states, thereby improving the efficiency of locating abnormal interrogation fragments and organizing interrogation materials.

[0063] In one embodiment, the result generation module sorts multiple abnormal event packets in chronological order to obtain an abnormal event sequence. The sorting can be based on the start time of the abnormality within the abnormal event packet, or on the peak time of the abnormality, the abnormal event number, or the position on the interrogation timeline. By sorting, multiple abnormal events can be presented in the chronological order of the interrogation process, facilitating the interrogator or case analyst to view the distribution of abnormal physiological states throughout the interrogation.

[0064] The results generation module generates abnormal physiological state records based on the abnormal event sequence. These records may include at least one of the following: abnormal occurrence time, abnormal duration, abnormal type, and abnormal physiological sign data. The abnormal physiological sign data may include respiratory rate data, heart rate data, respiratory curve segments, heart rate curve segments, abnormal peak values, abnormal amplitude of change, and abnormal confidence levels within the abnormal time period. The abnormal physiological state records can be generated in tabular, timeline, or event list format.

[0065] The results generation module correlates abnormal physiological state records, corresponding physiological sign curve segments, and corresponding target interrogation content to obtain interrogation correlation analysis content. Specifically, the interrogation correlation analysis content can be displayed on the same interface or in the same report section, showing the abnormal physiological state time period, abnormality type, physiological sign curve segment, target audio segment playback entry, and target text segment content. If the system has visualization capabilities, the abnormal location can also be displayed on the physiological curve using color, marker points, time axis intervals, or labels.

[0066] The results generation module generates auxiliary interrogation analysis results based on the interrogation correlation analysis content. These auxiliary analysis results can include at least one of the following: abnormal physiological state time period, abnormality type, corresponding physiological sign curve segment, target audio segment, and target text segment. The auxiliary analysis results can be output as an electronic report, a structured data file, a visual timeline, a list of interrogation events, or an interrogation auxiliary analysis interface. These results are used to assist in recording the interrogation process, reviewing abnormal states, and subsequent analysis; they do not directly replace the judgment of the interrogators or relevant personnel regarding the facts of the case.

[0067] This embodiment demonstrates the automated generation of auxiliary interrogation analysis results from abnormal event packages. This implementation can integrate scattered physiological sign data, interrogation audio clips, and interrogation text clips into structured, traceable analysis content, improving the efficiency of interrogation process review, abnormal clue location, and report compilation.

[0068] In one embodiment, the millimeter-wave radar-based assisted interrogation system further includes a human-computer interaction module and an installation and support module. The human-computer interaction module may include a display screen, interrogation terminal, tablet terminal, desktop client, web page, or dedicated control panel. The human-computer interaction module is used to acquire the time period of abnormal physiological states output by the anomaly analysis module and generate early warning display information. The early warning display information may include text prompts, color prompts, pop-up prompts, sound prompts, flashing icons, or timeline markers. For example, when the anomaly analysis module identifies that respiratory data or heart rate data exceeds preset discrimination conditions, the human-computer interaction module can generate a red early warning marker on the interface and display the anomaly type and the time of anomaly occurrence.

[0069] The human-computer interaction module also synchronously displays warning information, continuous change curves of physiological signs, and target interrogation content corresponding to the time period of abnormal physiological state, providing real-time auxiliary interrogation prompts. Specifically, the interface may include a real-time respiratory curve, a real-time heart rate curve, a scrolling display area for interrogation text, an abnormal event list area, and an audio playback area. Interrogators can click on abnormal markers to view corresponding physiological sign curve segments, target interrogation text segments, and target audio segments, thereby paying close attention to the questions and answers corresponding to abnormal physiological states during the interrogation process.

[0070] The mounting module is used to support the millimeter-wave radar monitoring module and install it in a predetermined location within the interrogation area. The mounting module can be installed using a fixed bracket, magnetic bracket, wall-mounted structure, desktop support structure, hanging structure, or embedded installation structure. The mounting module allows adjustment of the millimeter-wave radar monitoring module's installation height, elevation angle, horizontal angle, and orientation, enabling the module to perform non-contact radar detection towards the target detection area where the interrogator is located.

[0071] In one alternative implementation, the mounting module may employ a ferrous casing, an environmentally friendly textured coating, and a perforated structure to blend its appearance with decorative elements, monitoring equipment, or other fixed fixtures in the interrogation area. The mounting module may also be equipped with a strong magnetic base or adjustable supports for quick installation, removal, and angle adjustment. The millimeter-wave radar monitoring module may be housed within the mounting module, with the perforated structure allowing for the transmission and reception of millimeter-wave radar signals. This arrangement reduces the impact of exposed equipment on the psychological state of the interrogated person.

[0072] In another optional implementation, the system employs a camera-less design, acquiring chest cavity micro-motion signals only through a millimeter-wave radar monitoring module and interrogation audio data through an audio acquisition module, without collecting image information of the interrogated person. The system can also be configured with a local data storage unit to store interrogation audio data, interrogation text data, physiological sign data, abnormal event packets, and auxiliary interrogation analysis results on a local device or local server. The camera-less design and local data storage reduce privacy risks associated with image information acquisition and enhance the system's adaptability to different interrogation environments.

[0073] This embodiment not only displays abnormal physiological states, continuous change curves, and target interrogation content to interrogators in real time during interrogation, but also ensures that the millimeter-wave radar monitoring module is stably oriented towards the target detection area through the installation of a support module. This implementation enhances the real-time prompting capability and scene deployment capability of the auxiliary interrogation system, while reducing the impact of equipment installation and external exposure on the natural state of interrogation.

[0074] In one embodiment, such as Figure 2 As shown, this system mainly consists of a central processing unit (CPU), an input module, a physiological sign monitoring module, a voice recording to text module, a physiological sign analysis module, a voice recording and vital sign correlation module, and a human-computer interaction terminal.

[0075] Central Processing Unit (CPU): As the control core of the system, it is responsible for coordinating data acquisition, processing, storage, and instruction distribution among various modules. It receives raw radio frequency signals from the radar module and audio signals from the microphone, and schedules and analyzes the algorithm execution.

[0076] Input module: Includes a high-fidelity directional microphone array for acquiring audio signals throughout the interrogation process and converting them into digital audio streams for transmission to the central processing unit.

[0077] Physiological signs monitoring module: The core component is a millimeter-wave radar sensor (60GHz FMCW radar). This module transmits frequency-modulated continuous waves, receives echo signals reflected from the chest cavity micro-movements of the person being questioned, and calculates respiratory rate and heart rate data.

[0078] Audio-to-text module: It has a built-in speech recognition (ASR) engine that processes the incoming audio stream in real time or offline to generate text data with timestamps.

[0079] Physiological sign analysis module: It has a built-in signal processing algorithm library to filter and denoise the physiological data calculated by the radar, and compare it with the preset normal threshold model or early warning analysis model.

[0080] Recording and vital signs association module: Based on a unified high-precision system clock (NTP time synchronization or local crystal oscillator), it adds a unified timestamp to audio frames, text segments and physiological data points, and establishes an index database.

[0081] The specific implementation process of physiological sign monitoring is as follows: Figure 3 As shown, it includes: Signal transmission and reception: Millimeter-wave radar transmits linear frequency modulated continuous waves (FMCW), which are reflected after encountering the body of the person being questioned. Due to the micrometer-level vibrations of the chest cavity caused by breathing and heartbeat, the phase of the reflected wave undergoes a slight change.

[0082] Signal preprocessing: The received echo signal is down-converted to an intermediate frequency signal by a mixer, and then sampled into a digital signal by an analog-to-digital converter (ADC).

[0083] Distance-velocity mapping: The central processing unit performs a fast Fourier transform (FFT) on the digital signal to lock the distance cell where the person being questioned is located in the distance dimension, filtering out background static noise.

[0084] Phase demodulation and feature extraction: Extract the phase information of the target range cell signal. Since the phase change is proportional to the displacement, the chest cavity displacement waveform is reconstructed using an unwrapping algorithm.

[0085] Spectrum analysis: Perform FFT spectrum analysis on the displacement waveform again to separate the low-frequency respiratory component (0.1Hz-0.5Hz) and the high-frequency heart rate component (0.8Hz-2.5Hz) in the frequency domain, thereby calculating the respiratory rate and heart rate in real time.

[0086] Radar transmitting antenna → Human body reflection → Receiving antenna → Mixer → ADC sampling → FFT transformation → Phase demodulation → Respiration / heart rate extraction algorithm Recording and vital signs association module logic: When the system starts, the central processing unit initializes a global high-precision timer.

[0087] The input module marks the current timestamp Taudio for each frame of audio it captures.

[0088] The physiological signs monitoring module outputs a calculation result each time (e.g., once every 1 second) and marks it with a timestamp Tbio.

[0089] Each time the audio-to-text module generates a piece of text, it marks the start timestamp Tstart and the end timestamp Tend.

[0090] Association Logic: When all data is stored in the database, the timestamp TT is used as the primary key. When the physiological signs analysis module detects an abnormal heart rate at time Tx, the association module automatically retrieves data from the database where T∈[Tx]. Δt,Tx+Δt]T∈[Tx Audio segments within the range of [Δt, Tx+Δt] and their corresponding textual testimonies form an "abnormal event package".

[0091] System workflow diagram as follows Figure 4 As shown, it specifically includes: Start -> Activate radar and recording -> Data acquisition and synchronous tagging -> Real-time physiological analysis -> Determine if threshold exceeded? -> (Yes) Trigger red alert and mark time point -> (No) Continue monitoring -> Interrogation ends -> Generate related report -> End.

[0092] This embodiment introduces millimeter-wave radar into an assisted interrogation scenario, collecting physiological data such as respiration and heart rate from the interrogated person in a non-contact manner, replacing traditional contact-based monitoring methods such as skin conductance and finger-clip heart rate monitoring. This reduces psychological suggestion, resistance, or physiological interference caused by wearing sensors, ensuring the objectivity of physiological data collection and the natural state of the interrogation process. Simultaneously, a unified time reference is established, aligning the interrogation audio stream, speech-to-text stream, and radar physiological data stream along the timeline. This ensures that each physiological fluctuation can be located to a corresponding interrogation question, answer, or audio segment, resolving the disconnect between physiological data and interrogation content. Furthermore, the system detects… When physiological indicators such as respiration and heart rate exceed preset thresholds, anomaly markers are automatically triggered. Based on the abnormal time period, corresponding audio clips and text content are extracted to form a correlation analysis result containing the abnormal physiological state, corresponding audio, and corresponding text, reducing the workload of manual comparison and material organization afterward. In addition, the system can also output early warning prompts during interrogation and link abnormal warnings, continuous change curves of physiological signs, and corresponding interrogation content for display. This allows interrogators to pay attention to the questions and answers corresponding to abnormal physiological fluctuations in a timely manner, thus forming an auxiliary interrogation closed loop of "monitoring, early warning, and correlation analysis," improving the efficiency of capturing abnormal states, associating interrogation content, and generating analysis results during the interrogation process.

[0093] 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.

Claims

1. An auxiliary interrogation system based on millimeter-wave radar, characterized in that, The system includes: The audio acquisition module is used to collect interrogation audio data during the interrogation process; A millimeter-wave radar monitoring module is used to perform non-contact radar detection on the interrogated person and generate physiological sign data of the interrogated person based on the obtained radar echo signal. The physiological sign data includes respiratory data and / or heart rate data. The text generation module is used to perform speech recognition processing on the interrogation audio data to generate interrogation text data corresponding to the interrogation audio data. The interrogation text data includes text content and its corresponding time information. The time synchronization module is used to perform time synchronization processing on the interrogation audio data, the interrogation text data and the physiological sign data based on a unified time reference, establish the time correspondence between the interrogation audio data, the interrogation text data and the physiological sign data, and obtain synchronized interrogation data; Anomaly analysis module is used to analyze the physiological sign data based on preset physiological state discrimination conditions and identify the time period of abnormal physiological state of the interrogated person. The correlation analysis module is used to determine the target interrogation content corresponding to the abnormal physiological state time period from the synchronous interrogation data based on the abnormal physiological state time period and the time correspondence. The target interrogation content includes interrogation audio segments and / or interrogation text segments. The result generation module is used to generate auxiliary interrogation analysis results based on the abnormal physiological state time period, the physiological sign data corresponding to the abnormal physiological state time period, and the target interrogation content.

2. The millimeter-wave radar-based auxiliary interrogation system according to claim 1, characterized in that, The millimeter-wave radar monitoring module is used for: The system transmits millimeter-wave radar signals to the target detection area where the interrogated person is located and receives radar echo signals reflected by the interrogated person. The radar echo signal is down-converted and analog-to-digital converted to obtain a radar digital signal. Perform range dimension analysis on the radar digital signal to determine the target range unit corresponding to the interrogated person; The thoracic cavity micromotion signal is extracted based on the target distance unit, and the physiological sign data is generated based on the thoracic cavity micromotion signal.

3. The millimeter-wave radar-based auxiliary interrogation system according to claim 2, characterized in that, The millimeter-wave radar monitoring module is used for: The quadrature I / Q signal is extracted from the target range cell, and the quadrature I / Q signal is de-DC processed to obtain the target quadrature I / Q signal; The target orthogonal I / Q signal is subjected to phase demodulation processing to obtain a phase-change signal; The phase change signal is unwrapped to obtain the thoracic displacement waveform; The thoracic displacement waveform is denoised and subjected to spectral analysis to obtain respiratory rate data and / or heart rate data, and the respiratory rate data and / or heart rate data are used as the physiological sign data.

4. The millimeter-wave radar-based auxiliary interrogation system according to claim 1, characterized in that, The time synchronization module is used for: A unified time reference is initialized at the start of the interrogation to obtain the interrogation timeline; Based on the interrogation timeline, audio timestamps are configured for the audio frames in the interrogation audio data to obtain time-stamped audio data; Based on the interrogation timeline, physiological timestamps are configured for the physiological data points in the physiological sign data to obtain time-stamped physiological sign data. Based on the interrogation timeline, a text timestamp is configured for the text segments in the interrogation text data to obtain text data with timestamps; The time correspondence is established based on the audio timestamp, the physiological timestamp, and the text timestamp.

5. The millimeter-wave radar-based auxiliary interrogation system according to claim 4, characterized in that, The text generation module is used for: Based on the audio timestamp, the interrogation audio data is divided into speech segments to obtain multiple audio segments to be identified; Speech recognition processing is performed on each of the audio segments to be identified to obtain the text content corresponding to each of the audio segments to be identified. Based on the start and end positions of each audio segment to be identified on the interrogation timeline, determine the start time information and end time information corresponding to each text content; The interrogation text data is obtained by binding the text content, the start time information, and the end time information.

6. The millimeter-wave radar-based auxiliary interrogation system according to claim 1, characterized in that, The anomaly analysis module is used for: The physiological sign data are smoothed and filtered to obtain the physiological sign data to be judged; Determine respiratory and / or heart rate variation characteristics based on the physiological signs data to be identified; The respiratory change characteristics and / or the heart rate change characteristics are compared with the preset physiological state discrimination conditions to obtain the abnormality discrimination result; Abnormal physiological data points are determined based on the abnormality discrimination results, and the time period of the abnormal physiological state is determined based on the continuous abnormal physiological data points.

7. The millimeter-wave radar-based auxiliary interrogation system according to claim 6, characterized in that, The anomaly analysis module is used for: The respiratory abnormality discrimination result is obtained by comparing the respiratory change characteristics with the respiratory rate threshold condition and / or the respiratory change amplitude threshold condition. Based on the comparison between the heart rate change characteristics and the heart rate frequency threshold conditions and / or the heart rate change amplitude threshold conditions, the heart rate abnormality discrimination result is obtained; Based on the respiratory abnormality discrimination results and / or the heart rate abnormality discrimination results, determine the abnormality type corresponding to the abnormal physiological data point; Anomaly marker information is generated based on the anomaly type, the time of occurrence of the anomaly, and the duration of the anomaly.

8. The millimeter-wave radar-based auxiliary interrogation system according to claim 7, characterized in that, The correlation analysis module is used for: The time range for associated searches is determined based on the time period of the abnormal physiological state. Based on the associated retrieval time range and the time correspondence, the corresponding interrogation audio segment is retrieved from the synchronous interrogation data to obtain the target audio segment; Based on the associated retrieval time range and the time correspondence, the corresponding interrogation text fragment is retrieved from the synchronous interrogation data to obtain the target text fragment; The abnormal physiological state time period, the abnormal marker information, the physiological sign data corresponding to the abnormal physiological state time period, the target audio segment, and the target text segment are bound together to obtain an abnormal event package.

9. The millimeter-wave radar-based auxiliary interrogation system according to claim 8, characterized in that, The result generation module is used for: The abnormal event packets are sorted in chronological order to obtain an abnormal event sequence; An abnormal physiological state record is generated based on the abnormal event sequence, and the abnormal physiological state record includes at least one of the following: abnormal occurrence time, abnormal duration, abnormal type, and abnormal physiological sign data. The abnormal physiological state records, the corresponding physiological sign curve segments, and the corresponding target interrogation content are correlated to obtain the interrogation correlation analysis content. The auxiliary interrogation analysis results are generated based on the interrogation correlation analysis content. The auxiliary interrogation analysis results include at least one of the following: abnormal physiological state time period, abnormality type, corresponding physiological sign curve segment, target audio segment, and target text segment.

10. The millimeter-wave radar-based auxiliary interrogation system according to claim 1, characterized in that, It also includes a human-computer interaction module and an installation and support module; The human-computer interaction module is used to obtain the abnormal physiological state time period output by the abnormal analysis module, generate early warning display information, and synchronously display the early warning display information, the continuous change curve of the physiological sign data, and the target interrogation content corresponding to the abnormal physiological state time period to obtain real-time auxiliary interrogation prompt results. The mounting module is used to carry the millimeter-wave radar monitoring module and install the millimeter-wave radar monitoring module at a preset position in the interrogation site, so that the millimeter-wave radar monitoring module can perform non-contact radar detection towards the target detection area where the interrogated person is located.

Images