Item inspection methods, devices and security gates

Abstract

This application discloses an item detection method, device, and security gate, belonging to the field of security inspection technology. The item detection method includes: acquiring electromagnetic signals of each sub-region to be detected through scanning; correcting the electromagnetic signals of each sub-region to obtain corrected electromagnetic signals for each sub-region; and determining whether a target item exists in the detection area based on the corrected electromagnetic signals of each sub-region, wherein the detection area is composed of the aforementioned sub-regions. This application performs item detection based on the corrected electromagnetic signals, improving the accuracy and anti-interference capability of item detection.

Description

Technical Field

[0001] This application belongs to the field of security inspection technology, and in particular relates to a method, device and security gate for detecting items. Background Technology

[0002] Most security gates on the market currently rely on magnetic field scanning technology to acquire signal characteristics, scanning the items being inspected by generating a strong magnetic field. This process involves magnetic field induction technology, which uses changes in the magnetic field to detect items. When an item passes through the security gate, its metallic composition alters the distribution of the magnetic field, and this change is detected by sensors installed inside the gate. During the magnetic field scanning process, the security gate collects a large amount of raw electromagnetic signals. However, due to the influence of the surrounding environment, the raw electromagnetic signals are prone to problems such as signal deviation or missing signals, requiring calibration before they can be used to identify and distinguish different items and subsequently determine the presence of a target item. It is difficult to solve the problems in the electromagnetic signals from a hardware perspective alone. A method is needed to calibrate the electromagnetic signals and apply it to item detection scenarios to improve the accuracy of item detection. Summary of the Invention

[0003] This application aims to address at least one of the technical problems existing in the related art. To this end, this application proposes an article detection method, apparatus, and security gate that detects articles based on calibrated electromagnetic signals, thereby improving the accuracy and anti-interference capability of article detection.

[0004] Firstly, this application provides a method for detecting articles, the method comprising:

[0005] Electromagnetic signals of each sub-region to be detected are obtained by scanning.

[0006] The electromagnetic signals of each sub-region to be detected are corrected to obtain the corrected electromagnetic signals of each sub-region to be detected.

[0007] Based on the corrected electromagnetic signals of each sub-region to be detected, it is determined whether there is a target item in the region to be detected, wherein the region to be detected is composed of each sub-region to be detected.

[0008] In the above technical solution, the electromagnetic signals of each sub-region to be detected are obtained by scanning, and the electromagnetic signals of each sub-region to be detected are corrected to obtain the corrected electromagnetic signals of each sub-region to be detected. Then, based on the corrected electromagnetic signals of each sub-region to be detected, it is determined whether there is a target item in the area to be detected composed of each sub-region to be detected. By dividing the sub-region to be detected and correcting the electromagnetic signals, the item detection is performed based on the corrected electromagnetic signals, which improves the accuracy and anti-interference of item detection.

[0009] According to one embodiment of this application, determining whether a target item exists in a detection area based on the corrected electromagnetic signals of each detection sub-region includes:

[0010] The correction electromagnetic signals of each sub-region to be detected are filtered according to the allowable value, and the stable electromagnetic signals of each sub-region to be detected are retained.

[0011] Based on the characteristic values ​​of the stable electromagnetic signals of each sub-region to be detected and its adjacent sub-regions to be detected, and the threshold range corresponding to the target item, it is determined whether the target item exists in the region to be detected.

[0012] In the above technical solution, the correction electromagnetic signals of each sub-region to be detected are filtered according to the allowable value, and the stable electromagnetic signals of each sub-region to be detected are retained. Based on the characteristic values ​​of the stable electromagnetic signals of each sub-region to be detected and its adjacent sub-regions to be detected and the threshold range corresponding to the target item, it is determined whether the target item exists in the area to be detected, thereby improving the accuracy and anti-interference of item detection.

[0013] According to one embodiment of this application, the step of correcting the electromagnetic signals of each sub-region to be detected to obtain corrected electromagnetic signals for each sub-region to be detected includes:

[0014] The electromagnetic signals of each sub-region to be detected are input into a multivariate time series interpolation diffusion model to obtain the corrected electromagnetic signals of each sub-region to be detected.

[0015] The multivariate time series interpolation diffusion model is obtained by training on electromagnetic signal samples.

[0016] In the above technical solution, the electromagnetic signals of each sub-region to be detected are input into a multivariate time series interpolation diffusion model trained based on electromagnetic signal samples, thereby obtaining the corrected electromagnetic signals of each sub-region to be detected. This solves the problem of electromagnetic signal defects, realizes the correction of electromagnetic signals of each sub-region to be detected, and is applied to the detection of objects to improve the accuracy and anti-interference of object detection.

[0017] According to one embodiment of this application, the multivariate time series interpolation diffusion model is trained through the following steps:

[0018] Feature extraction is performed on the electromagnetic signal samples to obtain electromagnetic signal sample features;

[0019] The electromagnetic signal sample features are subjected to frequency division processing to determine the dominant frequency signal features and the non-dominant frequency signal features of the electromagnetic signal sample.

[0020] Based on the non-dominant frequency signal characteristics of the electromagnetic signal samples, a joint Gaussian distribution is determined;

[0021] Based on the joint Gaussian distribution, the electromagnetic signal sample features are forward diffused to obtain forward diffused data of the electromagnetic signal sample features, which includes noisy electromagnetic signal sample features.

[0022] Based on the forward diffusion data of the electromagnetic signal sample features, the noisy electromagnetic signal sample features are subjected to reverse diffusion processing. In the reverse diffusion processing, the loss of the multivariate time series interpolation diffusion model is determined, and the parameters of the multivariate time series interpolation diffusion model are optimized based on the loss.

[0023] In the above technical solution, feature extraction is performed on electromagnetic signal samples to obtain electromagnetic signal sample features. Frequency division processing is then applied to these features to determine the dominant frequency signal features and non-dominant frequency signal features. Based on the non-dominant frequency signal features, a joint Gaussian distribution is determined. Based on this distribution, the electromagnetic signal sample features are forward diffused to obtain forward diffused data. Based on this forward diffused data, noisy electromagnetic signal sample features are reverse diffused. In the reverse diffused processing, the loss of the multivariate time series interpolation diffusion model is determined. Based on this loss, the model parameters are optimized, thus enabling the training of the multivariate time series interpolation diffusion model. This improves the accuracy and stability of the model's prediction of missing electromagnetic signal values, thereby enhancing the accuracy and stability of electromagnetic signal correction. Using the corrected electromagnetic signals for object detection improves the accuracy and anti-interference capabilities of object detection.

[0024] According to one embodiment of this application, the step of performing forward diffusion processing on the electromagnetic signal sample features based on the joint Gaussian distribution to obtain forward diffusion data of the electromagnetic signal sample features includes:

[0025] Based on the current time step and the preset time step, the joint Gaussian distribution is sampled to determine the positive noise of the current step;

[0026] Add the current step positive noise to the positive features of the previous step electromagnetic signal sample to obtain the current step electromagnetic signal sample positive features;

[0027] Gradually increase the current time step and repeat the above steps until the current time step is equal to the preset time step, and determine the positive feature of the electromagnetic signal sample in the current step as the feature of the noisy electromagnetic signal sample;

[0028] The forward diffusion data of the electromagnetic signal sample features also includes the forward noise of each current step and the forward features of each current step electromagnetic signal sample.

[0029] In the above technical solution, the joint Gaussian distribution is sampled according to the current time step and the preset time step to determine the positive noise of the current step. The positive noise of the current step is added to the positive features of the electromagnetic signal sample of the previous step to obtain the positive features of the electromagnetic signal sample of the current step. The current time step is gradually increased and the above steps are repeated until the current time step is equal to the preset time step. The positive features of the electromagnetic signal sample of the current step are determined to be noisy electromagnetic signal sample features. This realizes the positive diffusion processing of electromagnetic signal sample features and obtains the positive diffusion data of electromagnetic signal sample features. This data can be used to perform reverse diffusion processing on noisy electromagnetic signal sample features and optimize the multivariate time series interpolation diffusion model in the reverse diffusion processing.

[0030] According to one embodiment of this application, when the current time step is 1, the electromagnetic signal sample feature is determined to be the positive feature of the previous electromagnetic signal sample.

[0031] In the above technical solution, when the current time step is 1, the electromagnetic signal sample features are determined to be the positive features of the previous electromagnetic signal sample, thus realizing the first noise addition to the electromagnetic signal sample features during the forward diffusion process.

[0032] According to one embodiment of this application, the step of performing reverse diffusion processing on the noisy electromagnetic signal sample features based on the forward diffusion data of the electromagnetic signal sample features, wherein the loss of the multivariate time series interpolation diffusion model is determined in the reverse diffusion processing, and the parameters of the multivariate time series interpolation diffusion model are optimized based on the loss, includes:

[0033] Based on the current time step and the inverse characteristics of the current step electromagnetic signal sample, the inverse noise of the current step is determined;

[0034] Based on the current step reverse noise, noise removal is performed on the reverse features of the current step electromagnetic signal sample to determine the reverse features of the previous step electromagnetic signal sample;

[0035] The current step loss of the multivariate time series interpolation diffusion model is determined by using an error estimation model.

[0036] The parameters of the multivariate time series interpolation diffusion model are optimized based on the current step loss.

[0037] Gradually reduce the current time step and repeat the above steps until the current time step is 1.

[0038] In the above technical solution, the reverse noise of the current step is determined based on the current time step and the reverse characteristics of the electromagnetic signal sample of the current step. Based on the reverse characteristics of the electromagnetic signal sample of the current step and the reverse noise of the current step, the reverse characteristics of the electromagnetic signal sample of the previous step are determined. Through the error estimation model, the current step loss of the multivariate time series interpolation diffusion model is determined. Based on the current step loss, the parameters of the multivariate time series interpolation diffusion model are optimized. The current time step is gradually reduced and the above steps are repeated until the current time step is 1. This realizes the reverse diffusion processing of the noisy electromagnetic signal sample characteristics. In the reverse diffusion processing, the multivariate time series interpolation diffusion model is optimized, which improves the accuracy and stability of the model in predicting the missing values ​​of the electromagnetic signal. This, in turn, improves the accuracy and stability of the electromagnetic signal correction. Based on the corrected electromagnetic signal, the object detection can be performed, which can improve the accuracy and anti-interference of the object detection.

[0039] According to one embodiment of this application, when the current time step is equal to the preset time step, the noisy electromagnetic signal sample feature is determined to be the inverse feature of the current step electromagnetic signal sample.

[0040] In the above technical solution, when the current time step is equal to the preset time step, the characteristics of the noisy electromagnetic signal sample are determined as the reverse characteristics of the electromagnetic signal sample in the current step, thereby realizing the first noise removal of the characteristics of the noisy electromagnetic signal sample during the reverse diffusion process.

[0041] According to one embodiment of this application, determining the current step loss of the multivariate time series interpolation diffusion model using an error estimation model includes:

[0042] The first current step loss is determined based on the current step inverse noise and the current step forward noise;

[0043] The second current step loss is determined based on the inverse features and forward features of the previous electromagnetic signal sample.

[0044] The current step loss of the multivariate time series interpolation diffusion model is determined based on the first current step loss and the second current step loss.

[0045] In the above technical solution, a first current step loss is determined based on the current step inverse noise and the current step forward noise. A second current step loss is determined based on the previous step electromagnetic signal sample inverse features and the previous step electromagnetic signal sample forward features. Based on the first and second current step losses, the current step loss of the multivariate time series interpolation diffusion model is determined. The parameters of the multivariate time series interpolation diffusion model are optimized based on the current step loss, which improves the accuracy and stability of the model's prediction of missing values ​​of electromagnetic signals and improves the accuracy and stability of electromagnetic signal correction. Item detection is then performed based on the corrected electromagnetic signals, thereby improving the accuracy and anti-interference capability of item detection.

[0046] According to one embodiment of this application, the error estimation model includes:

[0047] The step embedding module is used to determine the embedding vector based on the current time step, so that the multivariate time series interpolation diffusion model can determine the current time step;

[0048] A time embedding module is used to add time dimension information to the features of the electromagnetic signal samples;

[0049] An attention mechanism estimator is used to perform high-dimensional computation processing on the outputs of the step embedding module and the time embedding module based on the attention mechanism.

[0050] In the above technical solution, the error estimation model includes a step embedding module, a time embedding module, and an attention mechanism estimator. The step embedding module is used to determine the embedding vector according to the current time step, so that the multivariate time series interpolation diffusion model can determine the current time step. The time embedding module is used to add time dimension information to the electromagnetic signal sample features. The attention mechanism estimator is used to perform high-dimensional calculation processing on the outputs of the step embedding module and the time embedding module according to the attention mechanism. By maintaining a learnable embedding vector and adding time dimension information, the error estimation model can more accurately capture the feature differences of the electromagnetic signal time series at different time steps, improve the accuracy of error estimation, thereby improving the training effect of the multivariate time series interpolation diffusion model, improving the accuracy and stability of electromagnetic signal correction, and performing object detection based on the corrected electromagnetic signal, thereby improving the accuracy and anti-interference of object detection.

[0051] Secondly, this application provides an electromagnetic signal discrimination device, which includes:

[0052] The acquisition unit is used to acquire the electromagnetic signals of each sub-region to be detected.

[0053] The correction unit is used to correct the electromagnetic signals of each sub-region to be detected, so as to obtain the corrected electromagnetic signals of each sub-region to be detected.

[0054] The discrimination unit is used to determine whether a target item exists in the detection area based on the correction electromagnetic signal of each detection sub-region, wherein the detection area is composed of each detection sub-region.

[0055] In the above technical solution, the item detection device is used to acquire the electromagnetic signals of each sub-region to be detected by scanning, correct the electromagnetic signals of each sub-region to be detected to obtain the corrected electromagnetic signals of each sub-region to be detected, and then determine whether there is a target item in the area to be detected composed of each sub-region to be detected based on the corrected electromagnetic signals of each sub-region to be detected. By dividing the sub-regions to be detected and correcting the electromagnetic signals, and performing item detection based on the corrected electromagnetic signals, the accuracy and anti-interference ability of item detection are improved.

[0056] Thirdly, this application provides a security gate, 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 the item detection method as described in the first aspect above.

[0057] Fourthly, this application provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the article detection method as described in the first aspect above.

[0058] Fifthly, this application provides a chip including a processor and a communication interface, the communication interface being coupled to the processor, the processor being used to run programs or instructions to implement the article detection method as described in the first aspect above.

[0059] Sixthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the article detection method as described in the first aspect above.

[0060] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0061] The above and / or additional aspects and advantages of this application will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which:

[0062] Figure 1 This is a schematic flowchart of an article detection method provided in some embodiments of this application;

[0063] Figure 2 This is a schematic diagram of electromagnetic signal offset provided in some embodiments of this application;

[0064] Figure 3 This is a schematic diagram illustrating the absence of electromagnetic signals provided in some embodiments of this application;

[0065] Figure 4 This is a schematic diagram of the correction electromagnetic signal provided in some embodiments of this application;

[0066] Figure 5 This is one of the flowcharts illustrating the training process of a multivariate time series interpolation diffusion model provided in some embodiments of this application;

[0067] Figure 6 This is the second schematic diagram of the training process of a multivariate time series interpolation diffusion model provided in some embodiments of this application;

[0068] Figure 7 These are schematic diagrams of error estimation models provided in some embodiments of this application;

[0069] Figure 8 This is a schematic diagram of the structure of an article detection device provided in some embodiments of this application;

[0070] Figure 9 This is a schematic diagram of the structure of a security gate provided in some embodiments of this application.

[0071] Explanation of reference numerals in the attached figures:

[0072] 800: Item detection device; 801: Acquisition unit; 802: Calibration unit; 803: Discrimination unit;

[0073] 900: Security gate; 901: Processor; 902: Memory. Detailed Implementation

[0074] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0075] The terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that embodiments of this application can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0076] The article below, in conjunction with the accompanying drawings, provides a detailed description of the article detection method, device, and security gate provided in this application through specific embodiments and application scenarios.

[0077] The article detection method provided in this application embodiment can be executed by an electronic device or a functional module or entity in an electronic device that can implement the article detection method. The electronic devices mentioned in this application embodiment include, but are not limited to, security gates or security inspection machines. The article detection method provided in this application embodiment is described below using an electronic device as the execution subject.

[0078] Figure 1 This is a flowchart illustrating an article detection method provided in some embodiments of this application. For example... Figure 1 As shown, the method for detecting the item includes steps 110, 120, and 130.

[0079] Step 110: Obtain the electromagnetic signals of each sub-region to be detected by scanning.

[0080] It is understandable that by scanning each sub-area to be detected by detection equipment (such as security gate), the electromagnetic signal obtained can be converted into a relatively stable sine wave through quadrature demodulation of the signal with a fixed frequency alternating voltage. The signal stability of this sine wave varies at different frequencies and can be converted into (x, y) coordinates in the XY two-dimensional rectangular coordinate system. However, since electromagnetic signals are easily affected by the surrounding environment, there may be numerical offsets or missing values. Figure 2 This is a schematic diagram of electromagnetic signal offset provided in some embodiments of this application. For example... Figure 2 As shown, the electromagnetic signal has undergone significant shift and interference, as can be seen from the fourth quadrant of the coordinate axis. Figure 3 This is a schematic diagram illustrating the absence of electromagnetic signals provided in some embodiments of this application. For example... Figure 3 As shown, the second quadrant of the coordinate axis indicates that the electromagnetic signal has data loss and cannot form a complete and continuous curve.

[0081] Step 120: Correct the electromagnetic signals of each sub-region to be detected to obtain the corrected electromagnetic signals of each sub-region to be detected.

[0082] It is understandable that the purpose of correcting the electromagnetic signals of each sub-region to be detected is to perform data completion and fitting of the electromagnetic signals, in order to resolve potential issues in the electromagnetic signals of each sub-region to be detected, such as those mentioned above. Figure 2 and Figure 3 Problems such as numerical offset or missing values ​​are identified by obtaining calibrated and recovered electromagnetic signals, which are the corrected electromagnetic signals for each sub-region to be detected. Detecting items based on these corrected electromagnetic signals helps improve the accuracy and anti-interference capabilities of item detection.

[0083] Step 130: Determine whether a target item exists in the detection area based on the corrected electromagnetic signals of each detection sub-region, wherein the detection area is composed of each detection sub-region.

[0084] Understandably, by dividing the area to be detected into sub-areas, calibrating and detecting each sub-area separately, and combining the detection results of each sub-area to determine the detection result of the entire area to be detected, the risk of false detection and missed detection is reduced, which helps to improve the accuracy of the item detection method.

[0085] In some embodiments, the target item includes at least one of the following: security-restricted items; electronic devices; metal articles; and controlled items. Security-restricted items refer to items that are restricted from being carried through security checks; electronic devices include mobile phones, tablets, laptops, smart wearable devices, power banks, Bluetooth headsets, and wireless equipment; metal articles refer to materials, components, or products made of metal, such as tools and equipment made of metal, metal jewelry and accessories, metal everyday items, metal components in electronic devices such as mobile phones, computers, and headsets, or metal materials; and controlled items include knives and weapons.

[0086] It should be noted that this application does not limit the specific type of target item, which can be determined according to the needs of the specific application scenario. For example, in an examination monitoring scenario, the presence of electronic devices in the area to be detected can be determined based on the calibration electromagnetic signals of each sub-area to be detected, thereby achieving security checks.

[0087] In the above technical solution, the electromagnetic signals of each sub-region to be detected are obtained by scanning, and the electromagnetic signals of each sub-region to be detected are corrected to obtain the corrected electromagnetic signals of each sub-region to be detected. Then, based on the corrected electromagnetic signals of each sub-region to be detected, it is determined whether there is a target item in the area to be detected composed of each sub-region to be detected. By dividing the sub-regions to be detected and correcting the electromagnetic signals, the item detection is performed based on the corrected electromagnetic signals, which improves the accuracy and anti-interference of item detection.

[0088] In some embodiments of this application, determining whether a target item exists in the detection area based on the corrected electromagnetic signals of each detection sub-region includes:

[0089] The correction electromagnetic signals of each sub-region to be detected are filtered according to the allowable value, and the stable electromagnetic signals of each sub-region to be detected are retained.

[0090] Based on the characteristic values ​​of the stable electromagnetic signals of each sub-region to be detected and its adjacent sub-regions to be detected, and the threshold range corresponding to the target item, it is determined whether the target item exists in the region to be detected.

[0091] Understandably, the allowable values ​​are set based on the actual application scenario and testing requirements, representing an acceptable range of electromagnetic signal fluctuations. Even though the electromagnetic signals of each sub-region to be tested should be stable after calibration, there may still be cases that exceed the allowable values. By screening and eliminating these calibrated electromagnetic signals that exceed the allowable values, the electromagnetic signals used for item detection become more stable, further improving the accuracy and anti-interference capability of item detection.

[0092] The characteristic values ​​of the stable electromagnetic signals in each sub-region to be detected can be phase characteristic values ​​or frequency characteristic values, etc. The threshold range corresponding to the target item is a pre-set threshold range determined based on the characteristic values ​​of the electromagnetic signals obtained by scanning the target item. If the characteristic values ​​of the stable electromagnetic signals in the sub-region to be detected are within the threshold range corresponding to the target item, then the target item can be considered to exist in the current sub-region to be detected.

[0093] It is understandable that different components of a target item may be distributed across different sub-regions to be detected. Therefore, item detection based on the electromagnetic signals of these sub-regions is not limited to a single sub-region but may extend to adjacent sub-regions. Thus, combining the characteristic values ​​of the stable electromagnetic signals of each sub-region and its adjacent sub-regions helps identify situations that are not obvious in a single sub-region but, when combined, meet the threshold range corresponding to the target item. Furthermore, by combining the characteristic values ​​of the stable electromagnetic signals of each sub-region and its adjacent sub-regions, false positives caused by abnormal electromagnetic signals or interference from a single sub-region can be reduced, thus improving the accuracy and reliability of item detection.

[0094] In the above technical solution, the correction electromagnetic signals of each sub-region to be detected are filtered according to the allowable value, and the stable electromagnetic signals of each sub-region to be detected are retained. Based on the characteristic values ​​of the stable electromagnetic signals of each sub-region to be detected and its adjacent sub-regions to be detected and the threshold range corresponding to the target item, it is determined whether the target item exists in the area to be detected, thereby improving the accuracy and anti-interference of item detection.

[0095] In some embodiments of this application, the step of correcting the electromagnetic signals of each sub-region to be detected to obtain corrected electromagnetic signals for each sub-region to be detected includes:

[0096] The electromagnetic signals of each sub-region to be detected are input into a multivariate time series interpolation diffusion model to obtain the corrected electromagnetic signals of each sub-region to be detected.

[0097] The multivariate time series interpolation diffusion model is obtained by training on electromagnetic signal samples.

[0098] Figure 4This is a schematic diagram of the correction electromagnetic signal provided in some embodiments of this application. For example... Figure 4 As shown, the corrected electromagnetic signal is the electromagnetic signal obtained by solving the problems of electromagnetic signal offset or missing electromagnetic signals. The electromagnetic signal sample refers to the set of electromagnetic signals used to train the multivariate time series interpolation diffusion model.

[0099] Time series imputation refers to the process of filling missing values ​​in time series data. Multivariate time series imputation focuses on time series data containing multiple dimensions or variables. Diffusion models restore the data by gradually injecting noise into the data and applying a learnable inverse process until the original clear data is obtained.

[0100] In the above technical solution, the electromagnetic signals of each sub-region to be detected are input into a multivariate time series interpolation diffusion model trained based on electromagnetic signal samples, thereby obtaining the corrected electromagnetic signals of each sub-region to be detected. This solves the problem of electromagnetic signal defects, realizes the correction of electromagnetic signals of each sub-region to be detected, and is applied to the detection of objects to improve the accuracy and anti-interference of object detection.

[0101] Figure 5 This is one of the schematic diagrams illustrating the training process of a multivariate time series interpolation diffusion model provided in some embodiments of this application. For example... Figure 5 As shown, the multivariate time series interpolation diffusion model is trained through the following steps, including: steps 510, 520, 530, 540 and 550.

[0102] Step 510: Extract features from the electromagnetic signal sample to obtain electromagnetic signal sample features.

[0103] Figure 6 This is the second schematic diagram of the training process for a multivariate time series interpolation diffusion model provided in some embodiments of this application. For example... Figure 6 As shown, during model training, feature extraction is performed on electromagnetic signal samples to obtain electromagnetic signal sample features. These electromagnetic signal sample features include the observed features of the electromagnetic signal samples. Features of missing values ​​in electromagnetic signal samples t = 0 indicates that the diffusion process has not yet started. and These correspond to the observed value features and missing value features of the electromagnetic signal sample in its initial state, respectively.

[0104] Step 520: Perform frequency division processing on the electromagnetic signal sample features to determine the main frequency signal features and non-main frequency signal features of the electromagnetic signal sample;

[0105] Step 530: Determine the joint Gaussian distribution based on the non-dominant frequency signal characteristics of the electromagnetic signal samples;

[0106] In this embodiment, a conditional distribution is generated from the latent representation of the observations, and noise sampling is performed according to this distribution to explicitly preserve the intrinsic correlation between the observed data and the missing data. In each step of noise addition, a joint Gaussian distribution with different covariance matrices is used for noise sampling. The noise sampling source is the characteristic values ​​of electromagnetic signal samples at other interference frequencies that are not the dominant frequency, so that the added noise can be closer to the noise characteristics in the actual application scenario. Based on the non-dominant frequency signal characteristics of the electromagnetic signal samples, the parameters of the joint Gaussian distribution, including the mean and covariance matrix, are determined so that the statistical characteristics of the noise match the actual interference frequency characteristics.

[0107] Understandably, a joint Gaussian distribution model based on observed and missing values ​​is constructed in the model, assuming the electromagnetic signal observation value x c and missing value x m They all follow a joint Gaussian distribution P(x), that is, (x... c ,x m By accurately estimating the joint distribution parameters, including calculating the infimum of P(x), we can infer missing values ​​using known observations, thereby achieving effective supplementation and recovery of electromagnetic signals in multiple dimensions.

[0108] Step 540: According to the joint Gaussian distribution, perform forward diffusion processing on the electromagnetic signal sample features to obtain forward diffusion data of the electromagnetic signal sample features, wherein the forward diffusion data of the electromagnetic signal sample features includes noisy electromagnetic signal sample features.

[0109] Understandably, according to the joint Gaussian distribution, the forward diffusion process of electromagnetic signal sample features refers to adding noise sampled from the joint Gaussian distribution at each time step. As the diffusion steps increase, the noise level of the electromagnetic signal sample features gradually increases, resulting in noisy electromagnetic signal sample features. The local features of the noisy electromagnetic signal sample features are covered by the added noise, which can be further used to gradually remove noise during the reverse diffusion process.

[0110] Forward diffusion data of electromagnetic signal sample features refers to a series of data generated during the forward diffusion process of electromagnetic signal sample features. This includes noise sampled from the joint Gaussian distribution added at each time step (called forward noise), as well as the intermediate state of the electromagnetic signal sample features after adding noise at each time step (called forward features of electromagnetic signal samples).

[0111] In some embodiments, adding a standard Gaussian-based noise value to the observation at step T can be represented as a Markov chain:

[0112]

[0113] Where q(X) 1:T |X0) represents the entire time series X given the initial state X0. 1:T The joint probability distribution of q(X); t |X t-1 ) represents the state X given the previous state. t-1 Under the given conditions, the current state is X. t The conditional probability; ∏ represents the product, i.e., the product of all terms;

[0114] Establish a conditional distribution of the approximate error based on given observations, based on (x c ,x m P(x) can be approximately expressed as:

[0115]

[0116] in, Indicates that at a given observation value missing values The probability of occurrence, t=0 represents the initial value. Let represent an expected value, where q is an approximate distribution used to estimate the true distribution p. The expected value is calculated under the q-distribution and consists of the sum of two log-likelihood ratios: Represents the entire time series X 0:T The probability and conditional probability q(X) 1:T The logarithm of the ratio of |X0), Indicates under given conditions Lower sequence The probability and The logarithm of the ratio;

[0117] Through derivation, the conditional distribution of the approximation error can finally be obtained:

[0118]

[0119] Where D KL Expressed as KL divergence calculation, E q This is expressed as the expected value calculation, where p θ It is a probabilistic imputation model designed to approximate the true conditional distribution P(X). m |X c ).

[0120] Step 550: Based on the forward diffusion data of the electromagnetic signal sample features, perform reverse diffusion processing on the noisy electromagnetic signal sample features. In the reverse diffusion processing, determine the loss of the multivariate time series interpolation diffusion model, and optimize the parameters of the multivariate time series interpolation diffusion model based on the loss.

[0121] Understandably, the reverse diffusion process starts with the noisy electromagnetic signal sample features at the end of the forward diffusion process. At each time step, the noise added during the forward diffusion process is predicted, and the predicted noise (called reverse noise) is removed from the current noisy data to obtain the intermediate state of the electromagnetic signal sample features at each time step after noise removal (called the reverse feature of the electromagnetic signal sample).

[0122] During the back diffusion process, after removing noise at each time step, the difference between the back noise and the forward noise can be calculated using the loss function to quantify the accuracy of the model's prediction. Based on the evaluation results of the loss function, the parameters of the multivariate time series interpolation diffusion model can be optimized using gradient descent or other optimization algorithms to minimize the loss function, thereby improving the model's prediction performance.

[0123] In the above technical solution, feature extraction is performed on electromagnetic signal samples to obtain electromagnetic signal sample features. Frequency division processing is then applied to these features to determine the dominant frequency signal features and non-dominant frequency signal features. Based on the non-dominant frequency signal features, a joint Gaussian distribution is determined. Based on this distribution, the electromagnetic signal sample features are forward diffused to obtain forward diffused data. Based on this forward diffused data, noisy electromagnetic signal sample features are reverse diffused. In the reverse diffused processing, the loss of the multivariate time series interpolation diffusion model is determined. Based on this loss, the model parameters are optimized, thus enabling the training of the multivariate time series interpolation diffusion model. This improves the accuracy and stability of the model's prediction of missing electromagnetic signal values, thereby enhancing the accuracy and stability of electromagnetic signal correction. Using the corrected electromagnetic signals for object detection improves the accuracy and anti-interference capabilities of object detection.

[0124] In some embodiments of this application, the step of performing forward diffusion processing on the electromagnetic signal sample features according to the joint Gaussian distribution to obtain forward diffusion data of the electromagnetic signal sample features includes:

[0125] Based on the current time step and the preset time step, the joint Gaussian distribution is sampled to determine the positive noise of the current step;

[0126] Add the current step positive noise to the positive features of the previous step electromagnetic signal sample to obtain the current step electromagnetic signal sample positive features;

[0127] Gradually increase the current time step and repeat the above steps until the current time step is equal to the preset time step, and determine the positive feature of the electromagnetic signal sample in the current step as the feature of the noisy electromagnetic signal sample;

[0128] The forward diffusion data of the electromagnetic signal sample features also includes the forward noise of each current step and the forward features of each current step electromagnetic signal sample.

[0129] It is understandable that the current time step t refers to the sequence number of the current forward diffusion step. Each time forward noise is added, the current time step increases by 1. This process continues until the current time step equals the preset time step t = T. The preset time step T refers to the total number of times forward noise needs to be added during the entire forward noise addition process. After T times of forward noise addition, the noisy electromagnetic signal sample features are obtained. It is understandable that the local features of the noisy electromagnetic signal sample features are covered by the added forward noise, which can be further used to gradually remove noise during the reverse diffusion process to complete and fit the electromagnetic signal samples, thereby solving problems such as numerical offsets or missing values ​​that may exist in the electromagnetic signal samples.

[0130] The positive feature of an electromagnetic signal sample refers to the intermediate state feature of the electromagnetic signal sample after noise is gradually added during the forward diffusion process. At the current time step, the positive noise of the current step is added to the positive feature of the electromagnetic signal sample of the previous step to obtain the positive feature of the electromagnetic signal sample of the current step.

[0131] The forward diffusion data of electromagnetic signal sample features includes the forward noise added at each time step and the resulting forward features of the electromagnetic signal samples.

[0132] like Figure 6 As shown, taking the current time step as T as an example, the current time step is equal to the preset time step t = T. That is, the current step is the last time positive noise is added. The joint Gaussian distribution is sampled to determine the current step positive noise (not shown in the figure). The positive features (observation features) of the electromagnetic signal sample from the previous step are also considered. and missing value features Adding positive noise to the current step yields the positive features (observation features) of the electromagnetic signal sample at the current step. and missing value features ), as a characteristic of noisy electromagnetic signal samples.

[0133] In some embodiments, when the current time step is 1, the electromagnetic signal sample feature is determined to be the positive feature of the previous electromagnetic signal sample.

[0134] It can be understood that the current time step is 1, that is, the current step is the first time to add positive noise. Since there is no positive feature of the electromagnetic signal sample from the previous step, the feature of the electromagnetic signal sample is determined to be the positive feature of the electromagnetic signal sample from the previous step, thus realizing the first noise addition to the feature of the electromagnetic signal sample during the forward diffusion process.

[0135] like Figure 6 As shown, the current time step is 1. A joint Gaussian distribution is sampled to determine the positive noise of the current step (not shown in the figure), and the electromagnetic signal sample characteristics (observation characteristics) are analyzed. and missing value features Add positive noise for the current step to obtain the positive features (observation features) of the electromagnetic signal sample for the current step. and missing value features ).

[0136] In the above technical solution, when the current time step is 1, the electromagnetic signal sample features are determined to be the positive features of the previous electromagnetic signal sample, thus realizing the first noise addition to the electromagnetic signal sample features during the forward diffusion process.

[0137] In the above technical solution, the joint Gaussian distribution is sampled according to the current time step and the preset time step to determine the positive noise of the current step. The positive noise of the current step is added to the positive features of the electromagnetic signal sample of the previous step to obtain the positive features of the electromagnetic signal sample of the current step. The current time step is gradually increased and the above steps are repeated until the current time step is equal to the preset time step. The positive features of the electromagnetic signal sample of the current step are determined to be noisy electromagnetic signal sample features. This realizes the positive diffusion processing of electromagnetic signal sample features and obtains the positive diffusion data of electromagnetic signal sample features. This data can be used to perform reverse diffusion processing on noisy electromagnetic signal sample features and optimize the multivariate time series interpolation diffusion model in the reverse diffusion processing.

[0138] In some embodiments of this application, the step of performing reverse diffusion processing on the noisy electromagnetic signal sample features based on the forward diffusion data of the electromagnetic signal sample features, wherein the loss of the multivariate time series interpolation diffusion model is determined in the reverse diffusion processing, and the parameters of the multivariate time series interpolation diffusion model are optimized based on the loss, includes:

[0139] Based on the current time step and the inverse characteristics of the current step electromagnetic signal sample, the inverse noise of the current step is determined;

[0140] Based on the current step reverse noise, noise removal is performed on the reverse features of the current step electromagnetic signal sample to determine the reverse features of the previous step electromagnetic signal sample;

[0141] The current step loss of the multivariate time series interpolation diffusion model is determined by using an error estimation model.

[0142] The parameters of the multivariate time series interpolation diffusion model are optimized based on the current step loss.

[0143] Gradually reduce the current time step and repeat the above steps until the current time step is 1.

[0144] During the reverse diffusion process, the current time step gradually decreases from the preset time step T, where the current time step t refers to the step number of the current reverse diffusion operation. Each time reverse noise removal is performed, the current time step decreases by 1; this process continues until the current time step is 1. The preset time step T refers to the total number of reverse noise removal operations required throughout the entire process. It can be assumed that after T reverse noise removal operations, the noise in the noisy electromagnetic signal sample features has been almost completely removed.

[0145] The reverse feature of an electromagnetic signal sample refers to the intermediate state feature of the noisy electromagnetic signal sample after gradually removing noise during the reverse diffusion process. At the current time step, the reverse noise of the current step electromagnetic signal sample is removed from the reverse feature of the current step electromagnetic signal sample to obtain the reverse feature of the previous step electromagnetic signal sample.

[0146] Understandably, each time inverse noise removal is performed, the loss of the current step is confirmed through the error estimation model, and the parameters of the multivariate time series interpolation diffusion model are optimized based on the loss. The optimized parameters are then applied to the next inverse noise removal, so that the inverse features of the electromagnetic signal samples in each step are as close as possible to the corresponding forward features, thereby achieving fine-tuning of the model.

[0147] like Figure 6 As shown, taking a current time step of 1 as an example, that is, the current step is the last time to remove inverse noise (Denoising), based on the inverse features (missing value feature estimation) of the electromagnetic signal sample in the current step. Given the current time step t=1, determine the inverse noise of the current step (not shown in the figure), and based on the inverse noise of the current step, perform inverse feature analysis on the electromagnetic signal sample of the current step. Noise removal is performed to determine the inverse features of the previous electromagnetic signal samples (missing value feature estimation). This method, used in a multivariate time-series interpolation diffusion model, estimates electromagnetic signal sample features obtained through forward and reverse diffusion processing. Forward diffusion refers to the process of gradually adding noise to the time-series data (i.e., the electromagnetic signal sample features), while reverse diffusion refers to the process of gradually removing noise from the noisy electromagnetic signal sample features obtained through forward diffusion, ultimately yielding the estimated electromagnetic signal sample features. It is understood that, for a trained model, this electromagnetic signal sample feature estimation can achieve data completion and fitting of electromagnetic signal samples, resolving issues such as potential numerical offsets or missing values ​​in the electromagnetic signal samples.

[0148] In some embodiments, when the current time step is equal to the preset time step, the noisy electromagnetic signal sample feature is determined to be the inverse feature of the current step electromagnetic signal sample.

[0149] It can be understood that the current time step is equal to the preset time step t=T, that is, the current step is the first removal of reverse noise. Since there is no reverse feature of the electromagnetic signal sample in the current step, the feature of the noisy electromagnetic signal sample is determined as the reverse feature of the electromagnetic signal sample in the current step, thus realizing the first noise removal of the feature of the noisy electromagnetic signal sample in the reverse diffusion process.

[0150] like Figure 6 As shown, the current time step is T. Based on the inverse features of the electromagnetic signal samples at the current step (missing value feature estimation) Given the current time step t = T, determine the inverse noise of the current step (not shown in the figure). Based on the inverse noise of the current step, perform inverse feature analysis on the electromagnetic signal sample of the current step. Noise removal is performed to determine the inverse features of the previous electromagnetic signal samples (missing value feature estimation). ).

[0151] In the above technical solution, when the current time step is equal to the preset time step, the noisy electromagnetic signal sample features are determined as the reverse features of the electromagnetic signal sample in the current step, thereby realizing the first noise removal of the noisy electromagnetic signal sample features during the reverse diffusion process.

[0152] In some embodiments, noise removal is performed on the inverse features of the electromagnetic signal sample of the current step based on the inverse noise of the current step to determine the inverse features of the electromagnetic signal sample of the previous step, including:

[0153] Based on the inverse noise of the current step, determine the mean value of the inverse features of the electromagnetic signal sample from the previous step;

[0154] Randomly sample the positive features (observed value features and missing value features) of the previous electromagnetic signal sample to obtain the sampling noise of the current step;

[0155] Based on the mean of the inverse features of the previous electromagnetic signal sample and the sampling noise of the current step, the inverse features of the previous electromagnetic signal sample (missing value feature estimation) are determined.

[0156] In the above technical solution, the reverse noise of the current step is determined based on the current time step and the reverse characteristics of the electromagnetic signal sample of the current step. Based on the reverse characteristics of the electromagnetic signal sample of the current step and the reverse noise of the current step, the reverse characteristics of the electromagnetic signal sample of the previous step are determined. Through the error estimation model, the current step loss of the multivariate time series interpolation diffusion model is determined. Based on the current step loss, the parameters of the multivariate time series interpolation diffusion model are optimized. The current time step is gradually reduced and the above steps are repeated until the current time step is 1. This realizes the reverse diffusion processing of the noisy electromagnetic signal sample characteristics. In the reverse diffusion processing, the multivariate time series interpolation diffusion model is optimized, which improves the accuracy and stability of the model in predicting the missing values ​​of the electromagnetic signal. This, in turn, improves the accuracy and stability of the electromagnetic signal correction. Based on the corrected electromagnetic signal, the object detection can be performed, which can improve the accuracy and anti-interference of the object detection.

[0157] In some embodiments of this application, determining the current step loss of the multivariate time series interpolation diffusion model using an error estimation model includes:

[0158] The first current step loss is determined based on the current step inverse noise and the current step forward noise;

[0159] The second current step loss is determined based on the inverse features and forward features of the previous electromagnetic signal sample.

[0160] The current step loss of the multivariate time series interpolation diffusion model is determined based on the first current step loss and the second current step loss.

[0161] Based on the inverse noise and the forward noise of the current step, a first current step loss is determined. The first current step loss can measure the accuracy of the model's prediction of noise during the inverse process, that is, the difference between the inverse noise predicted by the model and the actual added forward noise. Based on the inverse features and forward features of the electromagnetic signal sample of the previous step, a second current step loss is determined. The second current step loss can measure the accuracy of the model's prediction of features during the inverse process, that is, the difference between the inverse features predicted by the model and the actual forward features.

[0162] In the reverse diffusion process, after each reverse noise removal step, the first and second current-step losses are determined using an error estimation model. The parameters of the multivariate time-series interpolation diffusion model are then optimized based on these losses. These optimized parameters are applied to the next reverse noise removal step to ensure that the reverse features of the electromagnetic signal samples at each step are as close as possible to their corresponding forward features, thus achieving fine-tuning of the model. It is understandable that the error estimation model primarily ensures the correct convergence direction during the reverse diffusion process, preventing ineffective convergence due to excessive errors.

[0163] In the above technical solution, a first current step loss is determined based on the current step inverse noise and the current step forward noise. A second current step loss is determined based on the previous step electromagnetic signal sample inverse features and the previous step electromagnetic signal sample forward features. Based on the first and second current step losses, the current step loss of the multivariate time series interpolation diffusion model is determined. The parameters of the multivariate time series interpolation diffusion model are optimized based on the current step loss, which improves the accuracy and stability of the model's prediction of missing values ​​of electromagnetic signals and improves the accuracy and stability of electromagnetic signal correction. Item detection is then performed based on the corrected electromagnetic signals, thereby improving the accuracy and anti-interference capability of item detection.

[0164] In one embodiment of this application, the error estimation model includes:

[0165] The step embedding module is used to determine the embedding vector based on the current time step, so that the multivariate time series interpolation diffusion model can determine the current time step;

[0166] A time embedding module is used to add time dimension information to the features of the electromagnetic signal samples;

[0167] An attention mechanism estimator is used to perform high-dimensional computation processing on the outputs of the step embedding module and the time embedding module based on the attention mechanism.

[0168] Figure 7 These are schematic diagrams of error estimation models provided in some embodiments of this application. For example... Figure 7 As shown, the error estimation model includes: an input layer, a linear transformation layer, a step embedding module, a temporal embedding module, a self-attention mechanism estimator, and an addition and normalization layer.

[0169] Understandably, since there are significant differences between the inverse and forward features of electromagnetic signal samples at different time steps, the error estimation model needs to obtain the current time step t for accurate error estimation. The step embedding module maintains a learnable N-dimensional embedding vector of T for all steps; that is, given the current time step t, the step embedding module outputs an N-dimensional vector.

[0170] It is understandable that the electromagnetic signal time series input can be viewed as sequential information. The time embedding module is used to add a time dimension to the sequential information, so that similar magnetic field signals at different points in the sequence can learn from each other and effectively compensate for missing data. Optionally, a fixed position embedding is compiled using a one-hot encoding method, where the fixed encoding is calculated as follows:

[0171] TE(t,2i)=sin(t / 10000 2i / d )

[0172] TE(t,2i+1)=cos(t / 10000 2i / d )

[0173] Where t is the time step, i is the dimension, and TE(T,2I) represents the position encoding.

[0174] Understandably, the attention mechanism estimator utilizes the attention mechanism to model and perform high-dimensional computation on the output data of the step embedding module and the temporal embedding module. Through deep learning, the error estimation model can autonomously learn to fit the data. In some embodiments, Z represents the vector layer of the attention layer, where SE represents Step Embedding and TE represents Temporal Embedding, as shown below:

[0175] Z0 = X T +SE+TE

[0176] After mapping Z0 to a length of 512, a self-attention mechanism is introduced for computation, represented as:

[0177]

[0178] Where, d model This indicates the size of the dimension, currently 512. Z0 is the input mapping vector, and W... Q To generate the weight matrix of the query vector, W V The softmax function is used to normalize the attention scores to generate the weight matrix of the value vector, and Att(Z0) refers to the output of the self-attention mechanism.

[0179] In the above technical solution, the error estimation model includes a step embedding module, a time embedding module, and an attention mechanism estimator. The step embedding module is used to determine the embedding vector according to the current time step, so that the multivariate time series interpolation diffusion model can determine the current time step. The time embedding module is used to add time dimension information to the electromagnetic signal sample features. The attention mechanism estimator is used to perform high-dimensional calculation processing on the outputs of the step embedding module and the time embedding module according to the attention mechanism. By maintaining a learnable embedding vector and adding time dimension information, the error estimation model can more accurately capture the feature differences of the electromagnetic signal time series at different time steps, improve the accuracy of error estimation, thereby improving the training effect of the multivariate time series interpolation diffusion model, improving the accuracy and stability of electromagnetic signal correction, and performing object detection based on the corrected electromagnetic signal, thereby improving the accuracy and anti-interference of object detection.

[0180] The article detection method provided in this application can be executed by an article detection device. This application uses an article detection device executing the article detection method as an example to illustrate the article detection device provided in this application.

[0181] Figure 8 This is a schematic diagram of the structure of an article detection device provided in some embodiments of this application. For example... Figure 8 As shown, the item detection device 800 includes:

[0182] Acquisition unit 801 is used to acquire the electromagnetic signals of each sub-region to be detected;

[0183] The correction unit 802 is used to correct the electromagnetic signals of each sub-region to be detected, so as to obtain the corrected electromagnetic signals of each sub-region to be detected.

[0184] The discrimination unit 803 is used to determine whether a target item exists in the detection area based on the correction electromagnetic signals of each detection sub-region, wherein the detection area is composed of each detection sub-region.

[0185] Optionally, the discrimination unit 803 is used for:

[0186] The correction electromagnetic signals of each sub-region to be detected are filtered according to the allowable value, and the stable electromagnetic signals of each sub-region to be detected are retained.

[0187] Based on the characteristic values ​​of the stable electromagnetic signals of each sub-region to be detected and its adjacent sub-regions to be detected, and the threshold range corresponding to the target item, it is determined whether the target item exists in the region to be detected.

[0188] Optionally, the correction unit 802 is used for:

[0189] The electromagnetic signals of each sub-region to be detected are input into a multivariate time series interpolation diffusion model to obtain the corrected electromagnetic signals of each sub-region to be detected.

[0190] The multivariate time series interpolation diffusion model is obtained by training on electromagnetic signal samples.

[0191] Optionally, the multivariate time series interpolation diffusion model is trained through the following steps:

[0192] Feature extraction is performed on the electromagnetic signal samples to obtain electromagnetic signal sample features;

[0193] The electromagnetic signal sample features are subjected to frequency division processing to determine the dominant frequency signal features and the non-dominant frequency signal features of the electromagnetic signal sample.

[0194] Based on the non-dominant frequency signal characteristics of the electromagnetic signal samples, a joint Gaussian distribution is determined;

[0195] Based on the joint Gaussian distribution, the electromagnetic signal sample features are forward diffused to obtain forward diffused data of the electromagnetic signal sample features, which includes noisy electromagnetic signal sample features.

[0196] Based on the forward diffusion data of the electromagnetic signal sample features, the noisy electromagnetic signal sample features are subjected to reverse diffusion processing. In the reverse diffusion processing, the loss of the multivariate time series interpolation diffusion model is determined, and the parameters of the multivariate time series interpolation diffusion model are optimized based on the loss.

[0197] Optionally, the step of performing forward diffusion processing on the electromagnetic signal sample features according to the joint Gaussian distribution to obtain forward diffusion data of the electromagnetic signal sample features includes:

[0198] Based on the current time step and the preset time step, the joint Gaussian distribution is sampled to determine the positive noise of the current step;

[0199] Add the current step positive noise to the positive features of the previous step electromagnetic signal sample to obtain the current step electromagnetic signal sample positive features;

[0200] Gradually increase the current time step and repeat the above steps until the current time step is equal to the preset time step, and determine the positive feature of the electromagnetic signal sample in the current step as the feature of the noisy electromagnetic signal sample;

[0201] The forward diffusion data of the electromagnetic signal sample features also includes the forward noise of each current step and the forward features of each current step electromagnetic signal sample.

[0202] Optionally, when the current time step is 1, the electromagnetic signal sample feature is determined to be the positive feature of the previous electromagnetic signal sample.

[0203] Optionally, the step of performing reverse diffusion processing on the noisy electromagnetic signal sample features based on the forward diffusion data of the electromagnetic signal sample features, wherein the loss of the multivariate time series interpolation diffusion model is determined in the reverse diffusion processing, and the parameters of the multivariate time series interpolation diffusion model are optimized based on the loss, includes:

[0204] Based on the current time step and the inverse characteristics of the current step electromagnetic signal sample, the inverse noise of the current step is determined;

[0205] Based on the current step reverse noise, noise removal is performed on the reverse features of the current step electromagnetic signal sample to determine the reverse features of the previous step electromagnetic signal sample;

[0206] The current step loss of the multivariate time series interpolation diffusion model is determined by using an error estimation model.

[0207] The parameters of the multivariate time series interpolation diffusion model are optimized based on the current step loss.

[0208] Gradually reduce the current time step and repeat the above steps until the current time step is 1.

[0209] Optionally, if the current time step is equal to the preset time step, the noisy electromagnetic signal sample feature is determined to be the inverse feature of the electromagnetic signal sample of the current step.

[0210] Optionally, determining the current step loss of the multivariate time series interpolation diffusion model using the error estimation model includes:

[0211] The first current step loss is determined based on the current step inverse noise and the current step forward noise;

[0212] The second current step loss is determined based on the inverse features and forward features of the previous electromagnetic signal sample.

[0213] The current step loss of the multivariate time series interpolation diffusion model is determined based on the first current step loss and the second current step loss.

[0214] Optionally, the error estimation model includes:

[0215] The step embedding module is used to determine the embedding vector based on the current time step, so that the multivariate time series interpolation diffusion model can determine the current time step;

[0216] A time embedding module is used to add time dimension information to the features of the electromagnetic signal samples;

[0217] An attention mechanism estimator is used to perform high-dimensional computation processing on the outputs of the step embedding module and the time embedding module based on the attention mechanism.

[0218] In the above technical solution, the item detection device is used to acquire the electromagnetic signals of each sub-region to be detected by scanning, correct the electromagnetic signals of each sub-region to be detected to obtain the corrected electromagnetic signals of each sub-region to be detected, and then determine whether there is a target item in the area to be detected composed of each sub-region to be detected based on the corrected electromagnetic signals of each sub-region to be detected. By dividing the sub-regions to be detected and correcting the electromagnetic signals, and performing item detection based on the corrected electromagnetic signals, the accuracy and anti-interference ability of item detection are improved.

[0219] The item detection device in this application embodiment can be an electronic device or a component of an electronic device, such as an integrated circuit or a chip. The electronic device can be a terminal or other devices besides a terminal. For example, the electronic device can be a mobile phone, tablet computer, laptop computer, PDA, in-vehicle electronic device, mobile internet device (MID), augmented reality (AR) / virtual reality (VR) device, robot, wearable device, ultra-mobile personal computer (UMPC), netbook, or personal digital assistant (PDA), etc. It can also be a server, network attached storage (NAS), personal computer (PC), television (TV), ATM, or self-service machine, etc. This application embodiment does not specifically limit the device.

[0220] The item detection device in this application embodiment can be a device with an operating system. This operating system can be a Microsoft (Windows) operating system, an Android operating system, an iOS operating system, or other possible operating systems; this application embodiment does not specifically limit it.

[0221] The article detection device provided in this application embodiment can realize all the processes implemented in the article detection method embodiment, and will not be described again here to avoid repetition.

[0222] In some embodiments, such as Figure 9 As shown, this application embodiment also provides a security gate 900, including a processor 901, a memory 902, and a computer program stored in the memory 902 and executable on the processor 901. When the program is executed by the processor 901, it implements the various processes of the above-described item detection method embodiment and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0223] This application also provides a non-transitory computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the various processes of the above-described article detection method embodiments and achieves the same technical effects. To avoid repetition, it will not be described again here.

[0224] The processor mentioned above is the processor in the security gate described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0225] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the various processes of the above-described article detection method embodiments.

[0226] The processor mentioned above is the processor in the security gate described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0227] This application embodiment also provides a chip, which includes a processor and a communication interface. The communication interface is coupled to the processor. The processor is used to run programs or instructions to implement the various processes of the above-described item detection method embodiments and can achieve the same technical effects. To avoid repetition, it will not be described again here.

[0228] It should be understood that the chip mentioned in the embodiments of this application may also be referred to as a system-on-a-chip, system chip, chip system, or system-on-a-chip, etc.

[0229] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other equivalent elements in the process, method, article, or apparatus that includes that element. Furthermore, it should be noted that the scope of the methods and apparatuses in the embodiments of this application is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. Additionally, features described with reference to certain examples may be combined in other examples.

[0230] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the related technology, can be embodied in the form of a computer software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions to cause a terminal (which may be a mobile phone, computer, server, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0231] The embodiments of this application have been described above with reference to the accompanying drawings. However, this application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of this application.

[0232] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to equivalent embodiments or examples. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0233] Although embodiments of this application have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of this application, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for detecting an item, characterized in that, include: Electromagnetic signals of each sub-region to be detected are obtained by scanning. Correcting the electromagnetic signals of each sub-region to be detected to obtain the corrected electromagnetic signals of each sub-region to be detected includes: inputting the electromagnetic signals of each sub-region to be detected into a multivariate time series interpolation diffusion model to obtain the corrected electromagnetic signals of each sub-region to be detected; the multivariate time series interpolation diffusion model is trained based on electromagnetic signal samples. Based on the corrected electromagnetic signals of each sub-region to be detected, it is determined whether there is a target item in the region to be detected. The region to be detected is composed of each sub-region to be detected, including: filtering the corrected electromagnetic signals of each sub-region to be detected according to an allowable value, and retaining the stable electromagnetic signals of each sub-region to be detected. Based on the characteristic values ​​of the stable electromagnetic signals of each sub-region to be detected and its adjacent sub-regions to be detected, and the threshold range corresponding to the target item, it is determined whether the target item exists in the region to be detected.

2. The article detection method according to claim 1, characterized in that, The characteristic values ​​of the stable electromagnetic signals of each sub-region to be detected are phase characteristic values ​​and / or frequency characteristic values.

3. The article detection method according to claim 1, characterized in that, The multivariate time series interpolation diffusion model is trained through the following steps: Feature extraction is performed on the electromagnetic signal samples to obtain electromagnetic signal sample features; The electromagnetic signal sample features are subjected to frequency division processing to determine the dominant frequency signal features and the non-dominant frequency signal features of the electromagnetic signal sample. Based on the non-dominant frequency signal characteristics of the electromagnetic signal samples, a joint Gaussian distribution is determined; Based on the joint Gaussian distribution, the electromagnetic signal sample features are forward diffused to obtain forward diffused data of the electromagnetic signal sample features, which includes noisy electromagnetic signal sample features. Based on the forward diffusion data of the electromagnetic signal sample features, the noisy electromagnetic signal sample features are subjected to reverse diffusion processing. In the reverse diffusion processing, the loss of the multivariate time series interpolation diffusion model is determined, and the parameters of the multivariate time series interpolation diffusion model are optimized based on the loss.

4. The article detection method according to claim 3, characterized in that, The step of performing forward diffusion processing on the electromagnetic signal sample features according to the joint Gaussian distribution to obtain forward diffusion data of the electromagnetic signal sample features includes: Based on the current time step and the preset time step, the joint Gaussian distribution is sampled to determine the positive noise of the current step; Add the current step positive noise to the positive features of the previous step electromagnetic signal sample to obtain the current step electromagnetic signal sample positive features; Gradually increase the current time step and repeat the above steps until the current time step is equal to the preset time step, and determine the positive feature of the electromagnetic signal sample in the current step as the feature of the noisy electromagnetic signal sample; The forward diffusion data of the electromagnetic signal sample features also includes the forward noise of each current step and the forward features of each current step electromagnetic signal sample.

5. The article detection method according to claim 4, characterized in that, The step involves performing reverse diffusion processing on the noisy electromagnetic signal sample features based on the forward diffusion data of the electromagnetic signal sample features. In this reverse diffusion processing, the loss of the multivariate time-series interpolation diffusion model is determined, and the parameters of the multivariate time-series interpolation diffusion model are optimized based on the loss. This includes: Based on the current time step and the inverse characteristics of the current step electromagnetic signal sample, the inverse noise of the current step is determined; Based on the current step reverse noise, noise removal is performed on the reverse features of the current step electromagnetic signal sample to determine the reverse features of the previous step electromagnetic signal sample; The current step loss of the multivariate time series interpolation diffusion model is determined by using an error estimation model. The parameters of the multivariate time series interpolation diffusion model are optimized based on the current step loss. Gradually reduce the current time step and repeat the above steps until the current time step is 1.

6. The article detection method according to claim 5, characterized in that, The step of determining the current step loss of the multivariate time series interpolation diffusion model through the error estimation model includes: The first current step loss is determined based on the current step inverse noise and the current step forward noise; The second current step loss is determined based on the inverse features and forward features of the previous electromagnetic signal sample. The current step loss of the multivariate time series interpolation diffusion model is determined based on the first current step loss and the second current step loss.

7. The article detection method according to any one of claims 5-6, characterized in that, The error estimation model includes: The step embedding module is used to determine the embedding vector based on the current time step, so that the multivariate time series interpolation diffusion model can determine the current time step; A time embedding module is used to add time dimension information to the features of the electromagnetic signal samples; An attention mechanism estimator is used to perform high-dimensional computation processing on the outputs of the step embedding module and the time embedding module based on the attention mechanism.

8. The article detection method according to claim 5, characterized in that, Based on the current step reverse noise, noise removal is performed on the reverse features of the current step electromagnetic signal sample to determine the reverse features of the previous step electromagnetic signal sample, including: Based on the current step reverse noise, determine the mean value of the reverse features of the previous step electromagnetic signal sample; Randomly sample the positive features of the electromagnetic signal sample from the previous step to obtain the sampling noise for the current step; The reverse characteristics of the electromagnetic signal sample in the previous step are determined based on the mean of the reverse characteristics of the previous step electromagnetic signal sample and the sampling noise in the current step.

9. An item detection device, characterized in that, include: The acquisition unit is used to acquire the electromagnetic signals of each sub-region to be detected. The correction unit is used to correct the electromagnetic signals of each sub-region to be detected, so as to obtain the corrected electromagnetic signals of each sub-region to be detected. The discrimination unit is used to determine whether a target item exists in the detection area based on the correction electromagnetic signal of each detection sub-region, wherein the detection area is composed of each detection sub-region; Correcting the electromagnetic signals of each sub-region to be detected to obtain the corrected electromagnetic signals of each sub-region to be detected includes: inputting the electromagnetic signals of each sub-region to be detected into a multivariate time series interpolation diffusion model to obtain the corrected electromagnetic signals of each sub-region to be detected; the multivariate time series interpolation diffusion model is trained based on electromagnetic signal samples. Based on the corrected electromagnetic signals of each sub-region to be detected, it is determined whether there is a target item in the region to be detected. The region to be detected is composed of each sub-region to be detected, including: filtering the corrected electromagnetic signals of each sub-region to be detected according to an allowable value, and retaining the stable electromagnetic signals of each sub-region to be detected. Based on the characteristic values ​​of the stable electromagnetic signals of each sub-region to be detected and its adjacent sub-regions to be detected, and the threshold range corresponding to the target item, it is determined whether the target item exists in the region to be detected.

10. A security gate, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the article detection method as described in any one of claims 1-8.

Images