An electronic device detection method and system based on passive passive detection

Abstract

The application relates to an electronic device detection method and system based on passive passive detection, belonging to the technical field of electronic device detection, and applied to a handheld detection device. The detection method comprises the following steps: receiving an environmental electromagnetic signal and performing adaptive noise reduction processing; extracting a frequency band with a signal strength exceeding a preset strength threshold to generate a suspicious frequency band set; performing fast Fourier transform on each suspicious frequency band to generate an electromagnetic feature vector and an electromagnetic confidence weight factor; weighting and fusing multiple frames of environmental electromagnetic signals to generate an enhanced electromagnetic feature vector and perform spatial spectrum estimation to calculate the horizontal azimuth angle of a signal source; identifying the center coordinates of a temperature anomaly area; fusing the horizontal azimuth angle and the center coordinates of the temperature anomaly area to generate the planar position coordinates of a target device; and driving a linear motor array to generate an azimuth guide vibration and controlling an OLED projection module to display target distance information. The application can improve the accurate detection capability of electronic devices in a complex electromagnetic environment.

Description

Technical Field

[0001] This application relates to the field of electronic device detection technology, and in particular to an electronic device detection method and system based on passive detection. Background Technology

[0002] With the development of modern information society, various portable electronic devices are becoming increasingly popular. These devices generate varying degrees of electromagnetic radiation and heat release during daily operation. For example, illegally or covertly used electronic products such as mobile phones, eavesdropping devices, and drone controllers inevitably leak high-frequency electromagnetic waves and experience localized temperature rises during operation. Traditional methods for detecting electronic devices mainly include active radar detection, X-ray scanning, and millimeter-wave imaging. However, these methods generally suffer from problems such as requiring human intervention, being easily detected, being harmful to the human body, or only providing static structural information, making it difficult to meet the growing demands for concealment, real-time performance, and intelligence. Especially in applications with high security requirements, how to effectively identify and accurately locate potential threat sources without triggering alarms has become one of the key challenges that urgently needs to be overcome.

[0003] Currently, while some common passive detection systems attempt to infer the presence and approximate location of targets by receiving natural electromagnetic signals from the environment, they are often susceptible to a large amount of unrelated electromagnetic interference, resulting in high false alarm rates, frequent missed detections, and insufficient identification accuracy. Especially in densely populated urban areas, industrial zones, or other strong electromagnetic backgrounds, it is difficult to reliably distinguish the difference between the actual radiation signal from the target object and other background noise, thus affecting the overall practicality and reliability of the system and preventing the achievement of high-precision device positioning capabilities. Summary of the Invention

[0004] In order to improve the accuracy of electronic device detection in complex electromagnetic environments, this application provides an electronic device detection method and system based on passive detection.

[0005] Firstly, this application provides a method for detecting electronic devices based on passive detection, employing the following technical solution:

[0006] A method for detecting electronic devices based on passive detection is applied to a handheld detection device integrating an RF receiving module, a pyroelectric sensor, a linear motor array, and an OLED projection module. The detection method includes:

[0007] The radio frequency receiving module receives environmental electromagnetic signals.

[0008] The pre-stored noise feature library is used to perform adaptive noise reduction processing on the environmental electromagnetic signal, and the noise-reduced spectrum data is output.

[0009] Scan the noise-reduced spectrum data, extract frequency bands whose signal strength exceeds a preset strength threshold, and generate a set of suspicious frequency bands;

[0010] For each suspicious frequency band in the suspicious frequency band set, perform a fast Fourier transform to generate an electromagnetic feature vector, and calculate the normalized Euclidean distance between the electromagnetic feature vector and the preset electronic device feature database to generate an electromagnetic confidence weight factor.

[0011] Based on the electromagnetic confidence weighting factor, the multi-frame environmental electromagnetic signals are weighted and fused to generate an enhanced electromagnetic feature vector and perform spatial spectrum estimation to calculate the horizontal azimuth angle of the signal source.

[0012] Acquire the infrared temperature distribution data collected by the pyroelectric sensor and identify the center coordinates of the temperature anomaly area;

[0013] By combining the horizontal azimuth angle with the center coordinates of the temperature anomaly area, the planar position coordinates of the target device are generated;

[0014] The linear motor array is driven to generate azimuth-guided vibration based on the planar position coordinates, and the OLED projection module is controlled to display the target distance information.

[0015] By employing the aforementioned technical solution, the advantages of both electromagnetic induction and infrared sensing channels are fully utilized. Combined with advanced signal processing technology and intelligent fusion strategies, accurate identification and three-dimensional spatial positioning of electronic devices operating in a concealed state are achieved. A key feature of this application is that it requires no manual detection signals throughout the entire process, relying entirely on the physical field characteristics spontaneously generated by the external target for detection, thus possessing extremely high stealth capabilities and practical value. Simultaneously, through multi-level data cleaning, feature extraction, weight evaluation, and cross-modal information fusion, the limitations of single sensing methods are effectively overcome, significantly enhancing the stability and generalization ability of the entire system.

[0016] Optionally, the steps of weighted fusion of multiple frames of environmental electromagnetic signals based on the electromagnetic confidence weighting factor to generate an enhanced electromagnetic feature vector and perform spatial spectrum estimation, and calculating the horizontal azimuth angle of the signal source include:

[0017] Obtain the electromagnetic feature vector sequence and the corresponding electromagnetic confidence weight factor sequence of environmental electromagnetic signals in multiple consecutive frames;

[0018] For each electromagnetic feature vector in the electromagnetic feature vector sequence, multiply it by the electromagnetic confidence weight factor of the corresponding frame to generate a weighted feature vector sequence;

[0019] Perform time-domain averaging on the weighted feature vector sequence to generate enhanced electromagnetic feature vectors;

[0020] Construct the covariance matrix of the enhanced electromagnetic eigenvectors and perform eigenvalue decomposition to separate the signal subspace matrix and the noise subspace matrix;

[0021] A spatial spectrum function is constructed based on the noise subspace matrix, and the spatial spectrum function values ​​corresponding to different azimuth angles are calculated.

[0022] The azimuth angle corresponding to the maximum peak value of the spatial spectrum function is output as the horizontal azimuth angle of the signal source.

[0023] By employing the above technical solution, the ideas of time-domain weighted averaging and time-frequency joint processing are cleverly integrated, fully leveraging the advantages of multi-level information synergy guided by confidence. On the one hand, the dynamic weighting mechanism improves the utilization and representativeness of the original data; on the other hand, it relies on a mature subspace decomposition framework to achieve robust and efficient angle estimation. Compared with traditional single-frame snapshot or simple multi-frame overlay modes, this solution can not only adapt to the challenges of various uncertainties in complex electromagnetic environments, but also maintain good real-time performance and generalization ability.

[0024] Optionally, after the step of identifying the center coordinates of the temperature anomaly region, the following steps may also be included:

[0025] Obtain the coordinate sequence of the center of the temperature anomaly region in N consecutive frames, and calculate the displacement vector between adjacent coordinates;

[0026] Generate a velocity distribution histogram and curvature variation matrix of the motion trajectory based on the displacement vector sequence;

[0027] Extract the peak standard deviation and the entropy value of the curvature change matrix from the velocity distribution histogram to construct the trajectory feature vector;

[0028] Simultaneously acquire the temperature change rate range and thermal gradient variance of the temperature anomaly region within the corresponding time period;

[0029] The trajectory feature vector, the temperature change rate range, and the thermal gradient variance are combined into a multimodal feature vector;

[0030] Calculate the Mahalanobis distance between the multimodal feature vector and the preset device heat source pattern library;

[0031] When the Mahalanobis distance is less than the preset threshold, it is marked as a heat source of the device and the center coordinates of the abnormal temperature area are output; otherwise, it is marked as a human heat source and the coordinates are masked.

[0032] By adopting the above technical solutions, a highly reliable and sensitive intelligent identification scheme for close-carrying scenarios has been constructed, which enhances the ability of existing security inspection systems to combat concealed electronic cheating tools.

[0033] Optionally, after the step of marking the human body as a heat source and masking the coordinates, the following may also be included:

[0034] Detect the electromagnetic confidence weighting factor of the temperature abnormality region where the human body heat source is located;

[0035] If the electromagnetic confidence weight factor continues to be greater than the preset confidence threshold, then the coordinate shielding is removed and the device is re-marked as a heat source.

[0036] By adopting the above technical solutions, a composite identification scheme integrating thermodynamic behavior modeling, macroscopic motion law constraints, and microscopic electromagnetic radiation detection is constructed. This not only solves the technical bottleneck of traditional schemes in accurately identifying contraband carried close to the body, but also enhances the adaptability to different camouflage strategies in complex usage scenarios. It demonstrates particularly promising application prospects in areas such as security checks to prevent cheating and the management of confidential locations.

[0037] Optionally, the step of fusing the horizontal azimuth angle with the center coordinates of the temperature anomaly area to generate the planar position coordinates of the target device includes:

[0038] The horizontal azimuth of the receiving signal source and the center coordinates of the temperature anomaly area;

[0039] Construct a unit direction vector based on the horizontal azimuth angle;

[0040] Multiply the unit direction vector by the preset detection distance calibration value to generate electromagnetic positioning coordinates;

[0041] Calculate the weighted average of the center coordinates of the temperature anomaly area and the electromagnetic positioning coordinates, and output the weighted average result as the planar position coordinates of the target equipment.

[0042] By adopting the above technical solutions, combining electromagnetic induction direction finding technology and infrared thermal imaging positioning methods, and through data fusion strategies, the problems of susceptibility to interference and limited accuracy of single sensing modes are solved, thereby improving the positioning stability and accuracy in complex electromagnetic environments.

[0043] Optionally, the step of driving the linear motor array to generate azimuth-guided vibration based on the planar position coordinates and controlling the OLED projection module to display target distance information includes:

[0044] Calculate the azimuth angle based on the horizontal component of the planar position coordinates of the target device;

[0045] The azimuth angle is mapped to a preset motor partition index number, and the linear motor corresponding to the motor partition index number in the linear motor array is activated;

[0046] Calculate the length of the projected arrow based on the modulus of the plane position coordinates;

[0047] The OLED projection module is controlled to project an indicator arrow; wherein the direction of the indicator arrow is aligned with the azimuth angle and its length is the length of the projected arrow.

[0048] By adopting the above technical solution, the abstract spatial coordinate information is analyzed layer by layer and transformed into intuitive tactile and visual feedback, which not only improves the accuracy and real-time performance of positioning guidance, but also enhances the human-computer interaction experience.

[0049] Optionally, after the step of generating the set of suspicious frequency bands, the following steps are also included:

[0050] Obtain the mean temperature distribution, electromagnetic noise floor intensity, and ambient light intensity of the current environment, and generate an environmental parameter vector;

[0051] The environmental parameter vector is input into a pre-trained scene classification model, which outputs a scene identifier code.

[0052] The corresponding electromagnetic interference feature library is invoked according to the scene identification code; wherein, the electromagnetic interference feature library stores typical interference frequency bands and probability distributions;

[0053] Calculate the matching probability of each frequency band in the suspected frequency band set with all frequency bands in the electromagnetic interference feature library;

[0054] If there is a frequency band with the highest matching probability exceeding a preset dynamic threshold, then the frequency band is identified as an electromagnetic interference frequency band and is eliminated.

[0055] An optimized set of suspicious frequency bands is constructed based on the remaining frequency bands in the set of suspicious frequency bands after elimination, and the interference confidence factor of each frequency band is calculated according to the matching probability.

[0056] By adopting the above technical solution, effective purification and refined management of the initially screened suspicious frequency band set are achieved. Compared with traditional static rule filtering methods, this technical solution can not only significantly improve the ability to suppress false alarms, but also ensure that truly valuable abnormal signals are not missed, thereby greatly enhancing the applicability and stability of the system in diverse and complex scenarios while ensuring security.

[0057] Secondly, this application provides an electronic device detection system based on passive detection, which adopts the following technical solution:

[0058] The signal receiving module is used to receive ambient electromagnetic signals through the radio frequency receiving module;

[0059] The noise reduction processing module is used to call a pre-stored noise feature library to perform adaptive noise reduction processing on the environmental electromagnetic signal and output noise reduction spectrum data;

[0060] The suspicious frequency band extraction module is used to scan the noise-reduced spectrum data, extract frequency bands whose signal strength exceeds a preset strength threshold, and generate a suspicious frequency band set.

[0061] The electromagnetic feature extraction module is used to perform a fast Fourier transform on each suspicious frequency band in the suspicious frequency band set to generate an electromagnetic feature vector;

[0062] The confidence weight calculation module is used to calculate the normalized Euclidean distance between the electromagnetic feature vector and the preset electronic device feature database, and generate the electromagnetic confidence weight factor.

[0063] The horizontal azimuth angle calculation module is used to perform weighted fusion of multiple frames of environmental electromagnetic signals based on the electromagnetic confidence weighting factor, generate an enhanced electromagnetic feature vector and perform spatial spectrum estimation to calculate the horizontal azimuth angle of the signal source.

[0064] The temperature anomaly region identification module is used to acquire infrared temperature distribution data collected by the pyroelectric sensor and identify the center coordinates of the temperature anomaly region.

[0065] The planar position coordinate generation module is used to fuse the horizontal azimuth angle with the center coordinates of the temperature anomaly area to generate the planar position coordinates of the target device;

[0066] The drive control module is used to drive the linear motor array to generate azimuth-guided vibration according to the planar position coordinates, and to control the OLED projection module to display the target distance information.

[0067] Thirdly, this application provides a computer device, which adopts the following technical solution:

[0068] A computer device includes a memory, a processor, and a computer program stored in the memory, the processor executing the computer program to perform the steps of the method as described in the first aspect.

[0069] Fourthly, this application provides a computer-readable storage medium, which adopts the following technical solution:

[0070] A computer-readable storage medium storing a computer program that can be loaded by a processor and executed as in any of the methods in the first aspect.

[0071] In summary, this application includes at least one of the following beneficial technical effects: by combining an RF receiving module with adaptive noise reduction and electromagnetic feature analysis, it can accurately identify potential electronic device signal sources in the environment and calculate their azimuth angle; simultaneously, it works with a pyroelectric sensor to detect abnormal temperature areas, and accurately locates the planar coordinates of the target device through multi-source data fusion; then, based on this coordinate information, it drives a linear motor array to generate directional vibration feedback and an OLED projection module to indicate distance, providing users with intuitive tactile and visual guidance. The entire process does not require active signal transmission, achieving efficient passive detection of electronic devices, while improving detection accuracy and user experience, and achieving intelligent human-computer interaction while ensuring concealment. Attached Figure Description

[0072] Figure 1 This is a first flowchart of an electronic device detection method based on passive detection, according to one embodiment of this application.

[0073] Figure 2 This is a schematic diagram of the second process of an electronic device detection method based on passive detection, which is one embodiment of this application.

[0074] Figure 3 This is a schematic diagram of the third process of an electronic device detection method based on passive detection, according to one embodiment of this application.

[0075] Figure 4 This is a schematic diagram of the fourth process of an electronic device detection method based on passive detection, according to one embodiment of this application.

[0076] Figure 5 This is a schematic diagram of the fifth process of an electronic device detection method based on passive detection, according to one embodiment of this application.

[0077] Figure 6 This is a schematic diagram of the sixth process of an electronic device detection method based on passive detection, according to one embodiment of this application.

[0078] Figure 7 This is a schematic diagram of the seventh process of an electronic device detection method based on passive detection, according to one embodiment of this application. Detailed Implementation

[0079] To make the purpose, technical solution, and advantages of this application clearer, the following description is provided in conjunction with the appendix. Figure 1-7 The present application will be further described in detail below with reference to embodiments. It should be understood that the specific embodiments described herein are for illustrative purposes only and are not intended to limit the scope of the application.

[0080] This application discloses a method for detecting electronic devices based on passive detection, applicable to handheld detection devices integrating an RF receiving module, a pyroelectric sensor, a linear motor array, and an OLED projection module. This method aims to achieve efficient location and visual alerts for potentially operational electronic devices (such as mobile phones and wireless communication devices) in the environment, and is particularly suitable for use in covert or counter-surveillance scenarios, exhibiting strong practicality and security.

[0081] Reference Figure 1 A method for detecting electronic devices based on passive detection, the method comprising:

[0082] Step S101: Receive ambient electromagnetic signals through the radio frequency receiving module;

[0083] The system continuously collects electromagnetic signals from the environment over a wide frequency range (e.g., 10kHz to 6GHz) using a radio frequency receiving module, covering the operating frequency bands of common civilian and military electronic equipment. Because it collects signals passively, without actively emitting any detection beams, it does not reveal its location, thus improving the system's stealth and anti-interference capabilities. This design is particularly suitable for situations where it is necessary to avoid detection by the enemy, such as military reconnaissance and security patrols.

[0084] Step S102: Call the pre-stored noise feature library to perform adaptive noise reduction processing on the environmental electromagnetic signal and output the noise reduction spectrum data;

[0085] To improve the accuracy of subsequent analysis and reduce the impact of interference factors, the system uses a pre-established noise feature library to perform adaptive denoising on the original electromagnetic data, thereby obtaining cleaner denoised spectrum data. This process is not a simple filtering operation, but rather involves dynamically adjusting the denoising strategy based on real-time assessment of the current environmental background noise and historical statistical models.

[0086] Specifically, dynamic spectral subtraction can be used to reduce noise components. The system assesses the background noise level of the current environment in real time and uses an attenuation coefficient that adjusts with the signal-to-noise ratio to correct the signal energy, thereby effectively removing random noise components without destroying the useful target signal structure. This step significantly improves the quality of the spectral data, enabling subsequent processing to more accurately distinguish between real electron radiation sources and surrounding clutter interference.

[0087] Step S103: Scan the noise reduction spectrum data, extract the frequency bands whose signal strength exceeds the preset strength threshold, and generate a set of suspicious frequency bands;

[0088] The process involves a comprehensive scan of the denoised spectral data to identify frequency bands with signal strength exceeding a predetermined threshold (e.g., -65dBm), forming a set of suspected frequency bands. Setting this threshold filters out extremely weak signals that may be natural noise or non-functional radiation, concentrating resources on frequency ranges where active electronic activity is genuinely possible. This screening mechanism not only reduces unnecessary computational burden but also enhances the system's sensitivity to weak but critical signals.

[0089] Step S104: Perform a fast Fourier transform on each suspicious frequency band in the suspicious frequency band set to generate an electromagnetic feature vector, and calculate the normalized Euclidean distance between the electromagnetic feature vector and the preset electronic device feature database to generate an electromagnetic confidence weight factor.

[0090] The electromagnetic characteristic vector includes the fundamental frequency amplitude, the third harmonic amplitude ratio, and the instantaneous phase offset.

[0091] Specifically, for each frequency band marked as suspicious, the system further performs a Fast Fourier Transform (FFT) to transform it from the time domain to a frequency domain representation for analysis. Based on this, three key parameters are extracted to form an electromagnetic feature vector: the fundamental frequency amplitude, the ratio of the third harmonic amplitude to the fundamental frequency, and the instantaneous phase shift, forming a multi-dimensional electromagnetic feature vector. The fundamental frequency amplitude reflects the energy concentration at the dominant frequency, while the ratio of the third harmonic amplitude to it reflects the level of nonlinear distortion, which is particularly important for distinguishing different types of electronic devices. On the other hand, the standard deviation of the instantaneous phase characterizes dynamic properties such as the signal modulation method or the stability of the internal oscillator. For example, voice recorders typically use a fixed crystal oscillator, resulting in relatively small phase changes; in contrast, some wireless transmission modules may have significant jitter, causing this value to rise significantly. This multi-dimensional modeling helps enhance the model's generalization ability and robustness.

[0092] Furthermore, to determine whether the acquired electromagnetic features truly originate from a known type of electronic device, they need to be compared with a locally stored standardized database of electronic device features. For this purpose, normalized Euclidean distance is introduced as a quantitative standard for measuring similarity, and based on this, an electromagnetic confidence weighting factor reflecting the reliability of the match is derived. The smaller this value, the closer the signal under test is to a specific category of sample; conversely, a larger value indicates higher uncertainty or that it may belong to an unknown type of device.

[0093] Finally, the corresponding electromagnetic confidence weighting factor, i.e., the probabilistic score within the interval [0,1], can be calculated based on the normalized Euclidean distance. Essentially, it reflects the degree of credibility of the judgment result. A higher confidence level means that the signal is very likely to originate from an electronic device of a specific category; conversely, a lower confidence level may indicate a misjudgment or an unknown source.

[0094] Step S105: Based on the electromagnetic confidence weighting factor, the multi-frame environmental electromagnetic signals are weighted and fused to generate an enhanced electromagnetic feature vector and perform spatial spectrum estimation to calculate the horizontal azimuth angle of the signal source.

[0095] To further improve recognition accuracy, the system also incorporates data frame sequences obtained from multiple samplings. A weighted fusion operation is performed based on the electromagnetic confidence weight factors corresponding to each frame, ultimately generating a more powerful enhanced electromagnetic feature vector. This approach integrates and optimizes the results of multiple independent observations according to their respective reliability levels, preserving the information contribution of high-quality samples while suppressing the impact of random errors, thus significantly improving overall classification accuracy.

[0096] In this embodiment, spatial spectrum estimation can employ the MUSIC (Multiple Signal Classification) algorithm, a high-resolution direction estimation algorithm. This algorithm relies on constructing a covariance matrix from the complex signal received by the antenna array and separating two orthogonal parts—the signal subspace and the noise subspace—through eigenvalue decomposition of the matrix. Then, by using a steering vector model search, the direction angle at which the spatial spectrum function reaches its peak is obtained, which is the horizontal azimuth angle of the target signal source. This method has good resolution and robustness, and can accurately estimate the spatial direction of the transmitter even in the presence of strong reverberation or multipath propagation.

[0097] Step S106: Obtain infrared temperature distribution data collected by the pyroelectric sensor and identify the center coordinates of the temperature anomaly area;

[0098] Among these methods, pyroelectric sensors mounted on handheld detection devices periodically capture images of the surface temperature distribution of surrounding objects. These devices are extremely sensitive to changes in external heat and can reflect the location and distribution of potential heat sources in a relatively short time.

[0099] In one embodiment of this application, the temperature anomaly region identification meets the following criteria: region temperature difference threshold: ΔT ≥ 2℃, and region diameter threshold: D ≤ 10cm. When a local area experiences a temperature rise significantly higher than the surrounding environment (e.g., a surface temperature rise exceeding 2℃ due to a heating element), and its lateral dimension does not exceed a defined range (e.g., diameter less than or equal to 10 cm), it can be determined as a valid temperature anomaly region. The system then determines the two-dimensional coordinates of the center point of the anomaly region. While this type of auxiliary information is not as highly selective as electromagnetic signals, it plays an important supplementary role in confirming the physical location of the target.

[0100] Step S107: Combine the horizontal azimuth angle with the center coordinates of the temperature anomaly area to generate the planar position coordinates of the target device;

[0101] The horizontal azimuth angle is combined with the coordinates of the identified hotspot center, and after appropriate weighting correction, it is mapped to two-dimensional planar position coordinates in a unified coordinate system. A weighted average approach is used here, giving a larger weight to the data set with higher confidence levels to ensure that the overall estimation result is as close as possible to the actual situation.

[0102] In step S108, the linear motor array is driven to generate azimuth-guided vibration based on the planar position coordinates, and the OLED projection module is controlled to display the target distance information.

[0103] Specifically, a linear motor with an appropriate number is activated based on the relative angle of the target device, triggering a clear directional vibration felt by the user's hand, helping them intuitively perceive the approximate direction of the target. Simultaneously, the built-in OLED display projects a distance indicator arrow of suitable length based on the calculated relative coordinates of the target, allowing the visual observer to quickly grasp the actual distance between the two objects. This combined feedback mechanism greatly enhances the user experience, making it particularly suitable for use in noisy environments or when operating blindly.

[0104] The above embodiments fully utilize the advantages of dual sensing channels—electromagnetic induction and infrared sensing—combining advanced signal processing technology and intelligent fusion strategies to achieve accurate identification and three-dimensional spatial positioning of electronic devices operating in a concealed state. A key feature of this application is that it requires no artificial detection signals throughout the entire process, relying entirely on the physical field characteristics spontaneously generated by the external target for detection, thus possessing extremely high stealth capabilities and practical value. Simultaneously, through multi-level data cleaning, feature extraction, weight evaluation, and cross-modal information fusion, the limitations of single sensing methods are effectively overcome, significantly enhancing the stability and generalization ability of the entire system.

[0105] As one embodiment of the handheld detection device in this application, the core hardware of the device is a radio frequency receiving module located at the top of the detector, which passively receives ambient electromagnetic signals from 10kHz to 6GHz using a broadband micro patch antenna. A micro OLED projection module is added next to this module, projecting an indicator arrow onto the ground through a 45° refractive prism. The arrow length is proportional to the target distance, and the direction is synchronized with the azimuth angle θ. Furthermore, a pyroelectric sensor is embedded in a 4mm diameter window on the side of the device to detect infrared anomaly areas with a temperature difference ≥2℃ and a diameter ≤10cm, and outputs their center coordinates. The handle area contains a ring array composed of six micro linear motors, which activate corresponding zones and generate intensity gradient feedback based on the planar position angle θ. The main control module is located on the circuit board in the middle of the device, integrating a 9V power supply system and a processing chip. It is responsible for executing the MUSIC algorithm, adaptive noise reduction, feature extraction, and electromagnetic-infrared data fusion calculations to generate unified spatial positioning information.

[0106] Reference Figure 2 As one implementation of step S105, the steps of weighted fusion of multi-frame environmental electromagnetic signals based on electromagnetic confidence weighting factors to generate enhanced electromagnetic feature vectors and perform spatial spectrum estimation, and calculating the horizontal azimuth angle of the signal source include:

[0107] Step S201: Obtain the electromagnetic feature vector sequence and the corresponding electromagnetic confidence weight factor sequence of multiple consecutive frames of environmental electromagnetic signals.

[0108] The system collects electromagnetic signal samples from multiple time points within the target area and converts them into vector forms—electromagnetic feature vectors—that reflect the electromagnetic environment characteristics under the current receiving conditions. These vectors typically contain information in multiple dimensions, such as amplitude, phase, or frequency, and serve as the foundational input data for subsequent processing. Simultaneously, to measure the reliability of each frame of data in the overall assessment, an electromagnetic confidence weighting factor is assigned to each frame. This factor may originate from signal-to-noise ratio evaluation, sensor quality feedback, or other reliability indicators, reflecting the validity and stability of the measurement results at a specific moment.

[0109] Step S202: For each electromagnetic feature vector in the electromagnetic feature vector sequence, multiply it by the electromagnetic confidence weight factor of the corresponding frame to generate a weighted feature vector sequence.

[0110] In real-world environments, various interference factors (such as noise fluctuations and hardware drift) mean that not all frames of data have equal value. By introducing an electromagnetic confidence weighting factor as an adjustment parameter, more reliable data can occupy a larger proportion in the synthesis process, effectively suppressing the negative impact of low-quality information. This frame-by-frame weighting method not only enhances the expressiveness of the principal components but also improves the robustness and adaptability of the entire system.

[0111] Step S203: Perform time-domain averaging on the weighted feature vector sequence to generate enhanced electromagnetic feature vectors;

[0112] Specifically, a moving weighted average algorithm is used to accumulate and average the weighted feature vectors over a time window, thereby obtaining a more representative comprehensive feature representation. Compared to simple overlay or single-frame selection, this approach can significantly reduce the impact of random errors while preserving the main trends, resulting in a more stable and smoother result. It should be noted that "enhanced" here does not only mean changes at the numerical level, but more importantly, it mathematically improves the feature vectors' ability to represent and distinguish real physical phenomena.

[0113] Step S204: Construct the covariance matrix of the enhanced electromagnetic eigenvectors and perform eigenvalue decomposition to separate the signal subspace matrix and the noise subspace matrix;

[0114] The covariance matrix constructed from the enhanced feature vector set reveals the energy distribution in each direction. Eigenvalue decomposition of this matrix naturally distinguishes which directions carry useful target signal components (called signal subspace) and which are mainly dominated by background noise (called noise subspace), thus achieving the essential separation between signal and interference and providing theoretical support for further precise positioning.

[0115] Step S205: Construct a spatial spectrum function based on the noise subspace matrix, and calculate the spatial spectrum function values ​​corresponding to different azimuth angles;

[0116] In this model, assuming a known noise subspace, any direction not belonging to that subspace should be considered a location of the signal. The resulting spatial spectrum function is essentially an energy response model for angle variables. The goal is to find the angle direction that minimizes the energy projected onto the noise subspace, since theoretically, the steering vector will only be perfectly orthogonal to the noise subspace when the test direction is exactly equal to the true angle of incidence. Therefore, by traversing the entire search range and calculating the function values ​​for each candidate angle, a map describing the spatial energy distribution can be created, where the most prominent peak position represents the potential radiation source location.

[0117] Step S206: Output the azimuth angle corresponding to the maximum peak value of the spatial spectrum function as the horizontal azimuth angle of the signal source.

[0118] The noise subspace projection reaches its minimum only when the test angle is exactly equal to the location of the actual radiation source, at which point the corresponding spatial spectral function reaches its maximum. Therefore, by accurately finding the angle parameter where this extreme value is located, the specific orientation of the target under test can be deduced.

[0119] It should be noted that in practice, various auxiliary measures such as interpolation fitting and threshold determination are needed to ensure the accuracy of the judgment. However, for typical application scenarios under normal working conditions, this method is sufficient to meet the requirements of high accuracy.

[0120] The above implementation cleverly integrates the concepts of time-domain weighted averaging and time-frequency joint processing, fully leveraging the advantages of multi-level information collaboration guided by confidence. On the one hand, it improves the utilization and representativeness of the original data through a dynamic weighting mechanism; on the other hand, it achieves robust and efficient angle estimation functionality based on a mature subspace decomposition framework. Compared to traditional single-frame snapshots or simple multi-frame overlay modes, this solution not only adapts to the challenges of various uncertainties in complex electromagnetic environments but also maintains good real-time performance and generalization ability.

[0121] Reference Figure 3 As a further implementation of the electronic device detection method, after the step of identifying the center coordinates of the temperature anomaly region, the method further includes:

[0122] Step S301: Obtain the coordinate sequence of the center of the temperature anomaly region in N consecutive frames, and calculate the displacement vector between adjacent coordinates;

[0123] This step aims to establish a dynamic position tracking system in the time dimension. Since video or infrared image acquisition has a certain sampling frequency, the center of the temperature anomaly area in each frame can be considered a spatiotemporal node. By sequentially reading the spatial coordinates of these nodes and constructing a displacement vector using the relative offset between consecutive frames, the movement path of the hotspot in space and its local directional change trend can be effectively depicted. This step is essentially a discretized trajectory reconstruction method, suitable for tracking moving objects under low-speed, non-rigid deformation conditions.

[0124] For example, in an examination environment, if a candidate hides a mobile phone under their feet, the heat source will fluctuate periodically with the walking motion. It is difficult to distinguish this situation from the body's own heat dissipation parts (such as the armpits, waist, etc.) by relying solely on static thermal imaging.

[0125] Step S302: Generate a velocity distribution histogram and curvature change matrix of the motion trajectory based on the displacement vector sequence;

[0126] Among them, the velocity distribution histogram reflects the probability density of different velocity ranges within the entire observation window, revealing the changing patterns of the overall movement speed of the object; while the curvature variation matrix, obtained by differentiating the trajectory curve point by point, describes the temporal evolution of the path curvature. These two types of indicators quantify the stability and complexity of the motion state at the macroscopic and microscopic levels, respectively.

[0127] For electronic devices carried close to the body, because they are attached to the movement of the human limbs, they usually exhibit relatively stable small-amplitude oscillation characteristics, with a concentrated velocity distribution and obvious peaks, and the curvature tends to be constant or fluctuates slightly. In contrast, freely moving human joints or other active tissues may exhibit a wider range of acceleration and deceleration behaviors as well as frequent turning movements, resulting in high velocity dispersion and enhanced curvature diversity.

[0128] Step S303: Extract the peak standard deviation of the velocity distribution histogram and the entropy value of the curvature change matrix to construct the trajectory feature vector;

[0129] Among them, the standard deviation of the histogram peak reflects the dispersion of the main clustering intervals of the velocity distribution; the information entropy of the curvature matrix measures the overall uncertainty level of the trajectory morphological complexity. These two values, as abstracted high-level semantic representation parameters, together constitute a trajectory feature vector with discriminative capabilities.

[0130] Understandably, this combination of features was chosen because it can preserve the core behavioral pattern differences in the original trajectory without relying on a specific physical model, while also possessing good robustness and generalization potential. For example, even under the influence of slight jitter or viewpoint changes, as long as the overall motion rhythm remains consistent, the aforementioned statistical properties can still maintain their consistent expression.

[0131] Step S304: Simultaneously acquire the temperature change rate range and thermal gradient variance of the temperature anomaly area within the corresponding time period;

[0132] Among them, the temperature change rate range refers to the difference between the highest heating rate and the lowest cooling rate within a given observation period, reflecting the range of drastic changes in the intensity of work done by the heat source; while the thermal gradient variance measures the spatial heterogeneity of temperature differences between pixels in the region, indirectly reflecting whether the heat source structure is compact.

[0133] In this embodiment, these two indicators are mainly used to distinguish between temperature rise phenomena caused by artificial heating devices and those caused by natural physiological metabolism. Generally speaking, the Joule effect caused by current loading during the operation of electronic devices leads to a significant instantaneous temperature rise, and even after a brief shutdown, it cools down rapidly, resulting in large fluctuations in the rate of temperature change. In addition, components such as small integrated chips often form highly concentrated point-like thermal fields, resulting in steep temperature transition zones around them, exhibiting a high thermal gradient variance. In contrast, normal human body temperature regulation is a slow and gradual process, mainly driven by blood circulation for heat transfer, thus appearing more gentle and uniform in both amplitude and rhythm.

[0134] Step S305: Combine the trajectory feature vector, the temperature change rate range, and the thermal gradient variance into a multimodal feature vector;

[0135] Multimodal refers to data sets from different types of sensors or channels based on different physical principles. This encompasses two distinct but complementary information streams: visual kinematics (trajectory-related) and infrared thermodynamics (temperature-related). Unifying these streams into a single feature space helps overcome the problems of single-modality susceptibility to noise interference and blurred boundaries, thereby improving the learning efficiency and generalization ability of subsequent classifiers.

[0136] Step S306: Calculate the Mahalanobis distance between the multimodal feature vector and the preset device heat source mode library;

[0137] Mahalanobis distance fully considers the covariance structure among variables within each category cluster, automatically adapting to imbalances in feature scales and providing reasonable compensation for potential correlations. This means that even when faced with new sample instances not seen in the training set, a more reasonable attribution inference can be made based on their distance from existing templates.

[0138] In this embodiment, the preset device heat source pattern library is an empirical database extracted from a large number of historical real cases through manual annotation. It contains typical feature representations of various common prohibited electronic products (such as mobile phones, miniature earphones, wireless transmission modules, etc.) under typical usage conditions. With the help of efficient retrieval algorithms (such as hash indexing mechanisms), the ideal reference prototype closest to the current test sample can be located in a very short time.

[0139] Step S307: When the Mahalanobis distance is less than the preset judgment threshold, mark it as a heat source of the device and output the center coordinates of the abnormal temperature area; otherwise, mark it as a heat source of the human body and mask the coordinates.

[0140] Setting appropriate preset thresholds is crucial to ensuring the system can accurately detect suspicious targets while avoiding harming innocent ones. Ideally, the heat signals generated by legitimate users should be far removed from any known violation patterns, thus matching them at a distance exceeding the threshold limit; conversely, if they fall into the danger zone, it is necessary to activate an alarm mechanism for focused monitoring.

[0141] Understandably, the "masking coordinates" measure for parts identified as "human heat sources" is a result of considerations from both privacy protection and technical defense perspectives. On the one hand, the body heat distribution of ordinary people should not be the subject of continuous tracking; on the other hand, it also reduces the possibility of malicious evaders deceiving the detection system through spoofing.

[0142] In the above implementation, a highly reliable and sensitive intelligent identification scheme for close-carrying scenarios was constructed, which improved the ability of existing security inspection systems to combat concealed electronic cheating tools.

[0143] Reference Figure 4 As a further implementation of the electronic device detection method, after the step of marking the human body as a heat source and shielding the coordinates, the method further includes:

[0144] Step S401: Detect the electromagnetic confidence weighting factor of the temperature abnormal region where the human body heat source is located;

[0145] The electromagnetic confidence weight factor is a normalized value between 0 and 1, used to characterize the probability of the existence of real electronic devices in the region. The calculation process of this data has been mentioned in the previous steps. The electromagnetic confidence weight factor of the temperature anomaly region can be obtained by calculating the normalized Euclidean distance between the electromagnetic feature vector and the pre-set electronic device feature database.

[0146] Step S402: If the electromagnetic confidence weight factor continues to be greater than the preset confidence threshold, then the coordinate shielding is removed and the device is re-marked as a heat source.

[0147] The requirement of "consistently greater than" means that the previous judgment cannot be reversed by a single accidental signal trigger; rather, a high level of confidence output must be maintained within a set time window. This design effectively filters out the risk of false triggers caused by factors such as electrostatic discharge, industrial noise sources, or brief passage of handheld devices, improving the system's anti-disturbance capability and long-term tracking accuracy. When the conditions are met, the original human body heat source tag will be revoked, and the access permission for that location will be restored, and a new "device heat source" identity will be assigned, so that subsequent linkage with video surveillance, alarm push, and other related security subsystems can take appropriate measures.

[0148] The above implementation scheme constructs a composite identification method that integrates thermodynamic behavior modeling, macroscopic motion constraints, and microscopic electromagnetic radiation detection. This not only solves the technical bottleneck of traditional methods in accurately identifying contraband carried close to the body, but also enhances its adaptability to different camouflage strategies in complex usage scenarios. It demonstrates particularly promising application prospects in areas such as security checks to prevent cheating and the management of confidential locations.

[0149] Reference Figure 5 As one implementation of step S107, the step of fusing the horizontal azimuth angle and the center coordinates of the temperature anomaly area to generate the planar position coordinates of the target device includes:

[0150] Step S501: Receive the horizontal azimuth angle of the signal source and the center coordinates of the temperature anomaly area;

[0151] Step S502: Construct a unit direction vector based on the horizontal azimuth angle;

[0152] The unit direction vector is a vector with a magnitude of 1 that contains only direction information and does not involve distance factors. Its construction process is based on the principle of trigonometric functions and is achieved by converting the horizontal azimuth angle into component projections on the x-axis and y-axis.

[0153] Specifically, if a reference direction (such as due north or due east) is set as the starting reference, any given horizontal azimuth angle can be decomposed into corresponding cosine and sine components, forming a standard direction vector on a two-dimensional plane. The key role of this step is to establish a bridge between the polar coordinate system and the rectangular coordinate system, enabling angle parameters, which originally only have directional attributes, to be processed within the same mathematical framework as other positioning information based on Cartesian coordinates, thus laying the foundation for subsequent distance expansion and coordinate synthesis.

[0154] Step S503: Multiply the unit direction vector by the preset detection distance calibration value to generate electromagnetic positioning coordinates;

[0155] The preset detection distance calibration value represents an empirical setting for the effective detection range of electromagnetic signals by the system. For example, in this embodiment, 6cm can be used as the preset detection distance calibration value, which is an empirical parameter under typical application scenarios. By performing a scalar multiplication operation between the dimensionless unit direction vector and the distance calibration value with a length dimension, a complete position vector with both a clear direction and a fixed distance can be obtained. The coordinate point corresponding to the endpoint of this vector is the estimated position of the target device derived based on the electromagnetic signal.

[0156] Step S504: Calculate the weighted average of the center coordinates of the temperature anomaly area and the electromagnetic positioning coordinates, and output the weighted average result as the planar position coordinates of the target equipment.

[0157] Among them, the electromagnetic positioning coordinates are assigned an electromagnetic confidence weight factor to reflect their reliability under the current environmental conditions; while the temperature anomaly coordinates use 1 minus the electromagnetic confidence weight factor as their own weight to ensure that the sum of the two weights is always equal to 1.

[0158] Understandably, this dynamic weight adjustment strategy fully considers the performance differences of different sensing mechanisms under various operating conditions. For example, when strong electromagnetic interference in the environment leads to a decrease in the accuracy of electromagnetic positioning, its weight is reduced accordingly while the influence of temperature positioning is increased, and vice versa. The weighted average calculation linearly combines two candidate coordinate points according to their respective weights to finally obtain the optimal location estimation result with comprehensive evaluation.

[0159] The above embodiments organically combine electromagnetic induction direction finding technology and infrared thermal imaging positioning methods, and through data fusion strategies, solve the problems of susceptibility to interference and limited accuracy of single sensing modes, thereby improving the positioning stability and accuracy in complex electromagnetic environments.

[0160] Reference Figure 6 As one implementation of step S108, the step of driving a linear motor array to generate azimuth-guided vibration based on planar position coordinates and controlling the OLED projection module to display target distance information includes:

[0161] Step S601: Calculate the azimuth angle based on the horizontal component of the planar position coordinates of the target device;

[0162] This step essentially transforms the two-dimensional position information in the Cartesian coordinate system into a polar coordinate representation that is easier for humans to understand and perceive. In this process, the azimuth angle, a crucial navigation parameter, characterizes the angular position of the target object relative to the observer or reference origin in the horizontal plane. Its calculation is based on the classic arctangent function principle, derived through trigonometric operations on the proportional relationship between the x-axis and y-axis components of the position coordinates. The significance of this coordinate transformation lies in simplifying the complex description of two-dimensional spatial relationships, enabling subsequent actuators to respond based on a unified angular reference, avoiding the control complexity and synchronization problems that might arise from directly processing coordinate components.

[0163] Step S602: Map the azimuth angle to the preset motor partition index number and activate the linear motor corresponding to the motor partition index number in the linear motor array.

[0164] The motor partition index number Index is calculated using the following formula: In the above formula, θ is the azimuth angle. The motor is divided into zones with angular intervals.

[0165] Specifically, considering the physical limitations and cost constraints of actual hardware systems, it is impossible to equip every minute angle change with an independent execution unit. Therefore, it is necessary to divide the continuous angle range into several discrete sector regions, each corresponding to a specific motor partition. This partitioning mapping mechanism is essentially a quantitative approximation process. By converting continuous angle inputs into a finite number of index numbers, it achieves an efficient transition from theoretical calculation to physical execution. The specific index calculation involves angle normalization and translation processing, that is, converting the traditional -180° to +180° range into a 0° to 360° interval and then dividing it at equal intervals. This design not only meets the requirements of most control systems for non-negative integer indices but also ensures the integrity of full circumferential coverage, ensuring that a corresponding motor partition can be found for response at any azimuth angle.

[0166] Linear motors, as electromechanical conversion devices, can transform electrical signal excitation into mechanical vibration output. Their linear motion characteristics give them a unique advantage in generating directional vibration. By forming an array structure of multiple linear motors pre-arranged at different spatial orientations, specific motors can be selectively activated according to the target orientation, allowing the user to feel vibration cues from the correct direction. This spatial mapping-based tactile guidance method fully utilizes the sensitivity of human skin to changes in vibration direction and intensity, providing clear and unambiguous directional information even in visually limited or noisy environments, significantly improving the safety and convenience of the user experience.

[0167] Step S603: Calculate the length of the projected arrow based on the modulus of the planar position coordinates;

[0168] Wherein, the length L of the projected arrow arrow The specific calculation formula is as follows: ;

[0169] In the above formula, L max Where k is the maximum projection length and k is the scaling factor. The modulus of the planar position coordinates.

[0170] Specifically, the coordinate modulus, as a manifestation of the Euclidean norm, quantitatively reflects the spatial distance of a target object relative to a reference point. Converting this to the visual arrow length requires considering the physical size limitations of the display device and the optimal range of human visual perception. To address this, the system employs a linear scaling strategy with an upper limit constraint. This maintains the proportional relationship between distance changes and graphic length while preventing excessively long arrows from exceeding the display boundaries. This adaptive length adjustment mechanism ensures that regardless of whether the target is near or far, users can intuitively perceive the changing trend of relative distance through the arrow length, enhancing the expressiveness and readability of navigation information.

[0171] Step S604: Control the OLED projection module to project the indicator arrow;

[0172] The direction of the indicator arrow is aligned with the azimuth angle and its length is the same as the length of the projected arrow.

[0173] Specifically, by precisely controlling the pixel illumination sequence and brightness distribution, directional arrows with clear directionality and length characteristics can be generated and projected. The spatial orientation of these arrows strictly follows the calculated azimuth parameters, ensuring consistency between visual guidance and physical direction; while the length of the arrows carries distance information, allowing users to simultaneously obtain both directional and distance navigational elements. This information presentation method greatly reduces the user's cognitive load, enabling complex three-dimensional spatial relationships to be expressed on a two-dimensional plane.

[0174] In the above implementation, the abstract spatial coordinate information is parsed layer by layer and transformed into intuitive tactile and visual feedback, which not only improves the accuracy and real-time performance of positioning guidance, but also enhances the human-computer interaction experience.

[0175] Reference Figure 7 As a further implementation of the electronic device detection method, after the step of generating a set of suspicious frequency bands, the method further includes:

[0176] Step S701: Obtain the mean temperature distribution, electromagnetic noise floor intensity, and ambient light intensity of the current environment, and generate an environmental parameter vector;

[0177] After completing the initial spectrum scan and extraction of suspicious frequency bands, the system further collects key parameters of the current physical environment, including the average temperature distribution, electromagnetic noise floor intensity, and ambient light intensity. These three parameters represent the changes in the space's thermodynamic state, background electromagnetic activity level, and lighting conditions, respectively. For example, temperature distribution can reflect the temperature difference between indoors and outdoors or the presence of local heat sources, thus helping to determine whether the environment is a closed examination room or an open area; while the electromagnetic noise floor characterizes the intensity of persistent non-target signals in the environment, helping to assess the overall complexity of the electromagnetic environment; ambient light intensity is an important basis for identifying lighting conditions, for example, there are significant differences between outdoor scenes under strong sunlight and low-light monitoring environments. These measured data are integrated into a unified environmental parameter vector, which serves as the basic input for subsequent intelligent processing.

[0178] Step S702: Input the environmental parameter vector into the pre-trained scene classification model and output the scene identification code;

[0179] This step employs a machine learning-based pattern recognition method, training with a large amount of sample data to obtain a functional relationship that maps input features to specific application scenario labels. Typical application scenarios may include different categories such as "standardized examination rooms," "confidential office spaces," and "prison supervision areas," each corresponding to different electromagnetic behavior patterns and potential sources of interference. This classification mechanism endows the entire spectrum monitoring system with context-aware capabilities, allowing it to select the most suitable strategy path based on the current specific usage context, thereby improving the accuracy of identifying real threat signals.

[0180] Step S703: Call the corresponding electromagnetic interference feature library according to the scene identification code;

[0181] The electromagnetic interference feature library stores typical interference frequency bands and their probability distributions.

[0182] Specifically, the electromagnetic interference feature database contains frequency characteristic descriptions of common interference sources in specific environments, such as typical interference center frequencies, bandwidth ranges, and their probability distribution patterns. For example, metal detectors commonly used in examination rooms may generate periodic pulse signals concentrated in the 30-50MHz frequency band; while access control card readers widely used in some confidential buildings (such as RFID devices operating at 13.56MHz) may also generate continuous transmissions at fixed frequencies. Therefore, by establishing dedicated knowledge graphs for different application scenarios, the targeting and efficiency of subsequent comparative analysis can be effectively improved.

[0183] Step S704: Calculate the matching probability of each frequency band in the suspicious frequency band set with all frequency bands in the electromagnetic interference feature library;

[0184] The process involves calculating the similarity between each extracted suspicious frequency band and existing entries in the interference feature database, yielding a matching probability value for each frequency band relative to various known interference types. This matching probability is not a simple binary judgment but rather a fusion of multiple dimensions, including frequency domain position deviation and power density curve shape consistency. It reflects the likelihood that a frequency band belongs to a certain known interference type and allows for a certain fluctuation range to account for slight frequency shifts caused by factors such as hardware aging and temperature drift. This soft-decision method is more robust and reliable than traditional hard-threshold classification and is closer to practical application needs.

[0185] Step S705: If there is a frequency band with the highest matching probability exceeding the preset dynamic threshold, then the frequency band is identified as an electromagnetic interference frequency band and is eliminated.

[0186] Furthermore, to determine which frequency bands should be considered invalid interference and excluded, their respective maximum match probability values ​​must be compared with dynamically set thresholds. Only when the maximum match probability of a frequency band exceeds this threshold will it be identified as a genuine interference signal from the external environment and removed from the original suspicious list.

[0187] Among them, the preset dynamic threshold T d The calculation formula is: In the above formula, S env Given the current ambient light intensity, S max P is the maximum illumination calibration value. hist α represents the false alarm rate in the same historical scenario, and β represents the weighting coefficients.

[0188] It should be noted that this threshold is not a fixed numerical constant, but rather a result of dynamic adjustment based on the current environmental conditions and historical performance. This means that the system can flexibly change the screening criteria when facing electromagnetic environments of varying complexity, avoiding both excessive leniency leading to increased risk of missed detections and excessive stringency causing excessive loss of useful signals.

[0189] Step S706: Construct an optimized set of suspicious frequency bands based on the remaining frequency bands in the set of suspicious frequency bands after elimination, and calculate the interference confidence factor for each frequency band according to the matching probability.

[0190] The formula for the interference confidence factor is:

[0191] ;

[0192] In the above formula, maxP(f,db) i C represents the highest matching probability between the current frequency band f and the i-th interference frequency band in the electromagnetic interference feature library. int A quantitative index characterizing the purity of a frequency band (with a value of 0 to 1).

[0193] Understandably, the interference confidence factors for each unremoved frequency band are derived from the matching probabilities mentioned above, and these are then reorganized into a new, optimized set of suspicious frequency bands. The interference confidence factor essentially reflects the degree to which a frequency band still retains signal components with unknown characteristics. If it does not closely match any known interference template, it means that it is more likely to carry illegal communication content or other noteworthy behavioral characteristics.

[0194] The above implementation achieves effective purification and refined management of the initially screened suspicious frequency band set. Compared with traditional static rule filtering methods, this technical solution can not only significantly improve the ability to suppress false alarms, but also ensure that truly valuable abnormal signals are not missed, thereby greatly enhancing the applicability and stability of the system in diverse and complex scenarios while ensuring security.

[0195] In this embodiment, by combining a dynamic threshold mechanism with an interference confidence factor, accurate identification and elimination of interference frequency bands in specific scenarios are achieved. The system first adaptively adjusts the decision threshold based on environmental parameters (such as illumination and historical false alarm rates) to filter out suspicious interference frequency bands. Then, through dual matching of frequency similarity and temporal waveform correlation, the maximum matching probability between each frequency band and known interference signals is calculated, and an interference confidence factor is generated accordingly. This factor quantifies the purity of the frequency band and guides subsequent signal feature processing. This technical solution effectively improves the detection accuracy of target signals in complex electromagnetic environments and significantly reduces the false alarm rate.

[0196] This application also discloses an electronic device detection system based on passive detection.

[0197] An electronic device detection system based on passive detection includes:

[0198] A signal receiving module, used to receive ambient electromagnetic signals via a radio frequency receiving module;

[0199] The noise reduction module is used to call the pre-stored noise feature library to perform adaptive noise reduction processing on environmental electromagnetic signals and output noise-reduced spectrum data.

[0200] The suspicious frequency band extraction module is used to scan the noise-reduced spectrum data, extract frequency bands whose signal strength exceeds a preset strength threshold, and generate a suspicious frequency band set.

[0201] The electromagnetic feature extraction module is used to perform a fast Fourier transform on each suspicious frequency band in the suspicious frequency band set to generate an electromagnetic feature vector;

[0202] The confidence weight calculation module is used to calculate the normalized Euclidean distance between the electromagnetic feature vector and the preset electronic device feature database, and generate the electromagnetic confidence weight factor.

[0203] The horizontal azimuth angle calculation module is used to perform weighted fusion of multiple frames of environmental electromagnetic signals based on electromagnetic confidence weighting factors, generate enhanced electromagnetic feature vectors and perform spatial spectrum estimation to calculate the horizontal azimuth angle of the signal source.

[0204] The temperature anomaly area identification module is used to acquire infrared temperature distribution data collected by the pyroelectric sensor and identify the center coordinates of the temperature anomaly area.

[0205] The planar position coordinate generation module is used to fuse the horizontal azimuth angle and the center coordinates of the temperature anomaly area to generate the planar position coordinates of the target device;

[0206] The drive control module is used to drive the linear motor array to generate azimuth-guided vibration based on the planar position coordinates, and to control the OLED projection module to display the target distance information.

[0207] An electronic device detection system based on passive detection according to an embodiment of this application can implement any of the above-described electronic device detection methods, and the specific working process of each module in the electronic device detection system can refer to the corresponding process in the above-described method embodiments.

[0208] In the several embodiments provided in this application, it should be understood that the provided methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for example, the division of a certain module is merely a logical functional division, and in actual implementation there may be other division methods, such as multiple modules can be combined or integrated into another system, or some features can be ignored or not executed.

[0209] This application also discloses a computer device.

[0210] A computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement an electronic device detection method based on passive detection as described above.

[0211] This application also discloses a computer-readable storage medium.

[0212] A computer-readable storage medium storing a computer program that can be loaded by a processor and executed as described above in any of the methods for detecting electronic devices based on passive detection.

[0213] The computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus, or device; the program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0214] In this application, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0215] Although this application has been described herein in conjunction with various embodiments, those skilled in the art, by reviewing the accompanying drawings, disclosure, and appended claims, will understand and implement other variations of the disclosed embodiments in carrying out the claimed application. In the claims, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude a plurality. A single processor or other unit can implement several functions listed in the claims. While different dependent claims may recite certain measures, this does not mean that these measures cannot be combined to produce a good effect.

[0216] The above are all preferred embodiments of this application and are not intended to limit the scope of protection of this application. Any feature disclosed in this specification (including the abstract and drawings) may be replaced by other equivalent or similar features unless specifically stated otherwise. That is, unless specifically stated otherwise, each feature is only one example of a series of equivalent or similar features.

Claims

1. A method for detecting electronic devices based on passive detection, applied to a handheld detection device integrating an RF receiving module, a pyroelectric sensor, a linear motor array, and an OLED projection module, characterized in that, The detection method includes: The radio frequency receiving module receives environmental electromagnetic signals. The pre-stored noise feature library is used to perform adaptive noise reduction processing on the environmental electromagnetic signal, and the noise-reduced spectrum data is output. Scan the noise-reduced spectrum data, extract frequency bands whose signal strength exceeds a preset strength threshold, and generate a set of suspicious frequency bands; For each suspicious frequency band in the suspicious frequency band set, perform a fast Fourier transform to generate an electromagnetic feature vector, and calculate the normalized Euclidean distance between the electromagnetic feature vector and the preset electronic device feature database to generate an electromagnetic confidence weight factor. Based on the electromagnetic confidence weighting factor, the multi-frame environmental electromagnetic signals are weighted and fused to generate an enhanced electromagnetic feature vector and perform spatial spectrum estimation to calculate the horizontal azimuth angle of the signal source. Acquire the infrared temperature distribution data collected by the pyroelectric sensor and identify the center coordinates of the temperature anomaly area; By combining the horizontal azimuth angle with the center coordinates of the temperature anomaly area, the planar position coordinates of the target device are generated; The linear motor array is driven to generate azimuth-guided vibration based on the planar position coordinates, and the OLED projection module is controlled to display the target distance information.

2. The method for detecting electronic devices based on passive detection according to claim 1, characterized in that, The steps of weighted fusion of multi-frame environmental electromagnetic signals based on the electromagnetic confidence weighting factor, generating enhanced electromagnetic feature vectors and performing spatial spectrum estimation, and calculating the horizontal azimuth angle of the signal source include: Obtain the electromagnetic feature vector sequence and the corresponding electromagnetic confidence weight factor sequence of environmental electromagnetic signals in multiple consecutive frames; For each electromagnetic feature vector in the electromagnetic feature vector sequence, multiply it by the electromagnetic confidence weight factor of the corresponding frame to generate a weighted feature vector sequence; Perform time-domain averaging on the weighted feature vector sequence to generate enhanced electromagnetic feature vectors; Construct the covariance matrix of the enhanced electromagnetic eigenvectors and perform eigenvalue decomposition to separate the signal subspace matrix and the noise subspace matrix; A spatial spectrum function is constructed based on the noise subspace matrix, and the spatial spectrum function values ​​corresponding to different azimuth angles are calculated. The azimuth angle corresponding to the maximum peak value of the spatial spectrum function is output as the horizontal azimuth angle of the signal source.

3. The method for detecting electronic devices based on passive detection according to claim 1, characterized in that, Following the step of identifying the center coordinates of the temperature anomaly region, the following steps are also included: Obtain the coordinate sequence of the center of the temperature anomaly region in N consecutive frames, and calculate the displacement vector between adjacent coordinates; Generate a velocity distribution histogram and curvature variation matrix of the motion trajectory based on the displacement vector sequence; Extract the peak standard deviation and the entropy value of the curvature change matrix from the velocity distribution histogram to construct the trajectory feature vector; Simultaneously acquire the temperature change rate range and thermal gradient variance of the temperature anomaly region within the corresponding time period; The trajectory feature vector, the temperature change rate range, and the thermal gradient variance are combined into a multimodal feature vector; Calculate the Mahalanobis distance between the multimodal feature vector and the preset device heat source pattern library; When the Mahalanobis distance is less than the preset threshold, it is marked as a heat source of the device and the center coordinates of the abnormal temperature area are output; otherwise, it is marked as a human heat source and the coordinates are masked.

4. The method for detecting electronic devices based on passive detection according to claim 3, characterized in that, Following the steps of marking the human body as a heat source and masking the coordinates, the following is also included: Detect the electromagnetic confidence weighting factor of the temperature abnormality region where the human body heat source is located; If the electromagnetic confidence weight factor continues to be greater than the preset confidence threshold, then the coordinate shielding is removed and the device is re-marked as a heat source.

5. The method for detecting electronic devices based on passive detection according to claim 1, characterized in that, The steps for generating the planar position coordinates of the target device by fusing the horizontal azimuth angle with the center coordinates of the temperature anomaly area include: The horizontal azimuth of the receiving signal source and the center coordinates of the temperature anomaly area; Construct a unit direction vector based on the horizontal azimuth angle; Multiply the unit direction vector by the preset detection distance calibration value to generate electromagnetic positioning coordinates; Calculate the weighted average of the center coordinates of the temperature anomaly area and the electromagnetic positioning coordinates, and output the weighted average result as the planar position coordinates of the target equipment.

6. The method for detecting electronic devices based on passive detection according to claim 5, characterized in that, The steps of driving the linear motor array to generate azimuth-guided vibration based on the planar position coordinates and controlling the OLED projection module to display target distance information include: Calculate the azimuth angle based on the horizontal component of the planar position coordinates of the target device; The azimuth angle is mapped to a preset motor partition index number, and the linear motor corresponding to the motor partition index number in the linear motor array is activated; Calculate the length of the projected arrow based on the modulus of the plane position coordinates; The OLED projection module is controlled to project an indicator arrow; wherein the direction of the indicator arrow is aligned with the azimuth angle and its length is the length of the projected arrow.

7. A method for detecting electronic devices based on passive detection according to any one of claims 1 to 6, characterized in that, Following the step of generating a set of suspicious frequency bands, the following steps are also included: Obtain the mean temperature distribution, electromagnetic noise floor intensity, and ambient light intensity of the current environment, and generate an environmental parameter vector; The environmental parameter vector is input into a pre-trained scene classification model, which outputs a scene identifier code. The corresponding electromagnetic interference feature library is invoked according to the scene identification code; wherein, the electromagnetic interference feature library stores typical interference frequency bands and probability distributions; Calculate the matching probability of each frequency band in the suspected frequency band set with all frequency bands in the electromagnetic interference feature library; If there is a frequency band with the highest matching probability exceeding a preset dynamic threshold, then the frequency band is identified as an electromagnetic interference frequency band and is eliminated. An optimized set of suspicious frequency bands is constructed based on the remaining frequency bands in the set of suspicious frequency bands after elimination, and the interference confidence factor of each frequency band is calculated according to the matching probability.

8. An electronic device detection system based on passive detection, applied to a handheld detection device integrating a radio frequency receiving module, a pyroelectric sensor, a linear motor array, and an OLED projection module, characterized in that, The detection system includes: The signal receiving module is used to receive ambient electromagnetic signals through the radio frequency receiving module; The noise reduction processing module is used to call a pre-stored noise feature library to perform adaptive noise reduction processing on the environmental electromagnetic signal and output noise reduction spectrum data; The suspicious frequency band extraction module is used to scan the noise-reduced spectrum data, extract frequency bands whose signal strength exceeds a preset strength threshold, and generate a suspicious frequency band set. The electromagnetic feature extraction module is used to perform a fast Fourier transform on each suspicious frequency band in the suspicious frequency band set to generate an electromagnetic feature vector; The confidence weight calculation module is used to calculate the normalized Euclidean distance between the electromagnetic feature vector and the preset electronic device feature database, and generate the electromagnetic confidence weight factor. The horizontal azimuth angle calculation module is used to perform weighted fusion of multiple frames of environmental electromagnetic signals based on the electromagnetic confidence weighting factor, generate an enhanced electromagnetic feature vector and perform spatial spectrum estimation to calculate the horizontal azimuth angle of the signal source. The temperature anomaly region identification module is used to acquire infrared temperature distribution data collected by the pyroelectric sensor and identify the center coordinates of the temperature anomaly region. The planar position coordinate generation module is used to fuse the horizontal azimuth angle with the center coordinates of the temperature anomaly area to generate the planar position coordinates of the target device; The drive control module is used to drive the linear motor array to generate azimuth-guided vibration according to the planar position coordinates, and to control the OLED projection module to display the target distance information.

9. A computer device, characterized in that: The method includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer program is stored that can be loaded by a processor and executed as described in any one of claims 1 to 7.

Images