CN113537343A - Metal classification method, device, equipment and storage medium - Google Patents
Metal classification method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN113537343A CN113537343A CN202110797634.4A CN202110797634A CN113537343A CN 113537343 A CN113537343 A CN 113537343A CN 202110797634 A CN202110797634 A CN 202110797634A CN 113537343 A CN113537343 A CN 113537343A
- Authority
- CN
- China
- Prior art keywords
- metal
- training
- scanning data
- scanning
- classification model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V3/00—Electric or magnetic prospecting or detecting; Measuring magnetic field characteristics of the earth, e.g. declination, deviation
- G01V3/08—Electric or magnetic prospecting or detecting; Measuring magnetic field characteristics of the earth, e.g. declination, deviation operating with magnetic or electric fields produced or modified by objects or geological structures or by detecting devices
- G01V3/10—Electric or magnetic prospecting or detecting; Measuring magnetic field characteristics of the earth, e.g. declination, deviation operating with magnetic or electric fields produced or modified by objects or geological structures or by detecting devices using induction coils
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Abstract
The application discloses a metal classification method, a device, equipment and a storage medium, wherein scanning data of each channel are obtained by respectively scanning an object to be detected through a plurality of different coil induction areas, and further, the scanning data of each channel are integrated to obtain comprehensive scanning data, so that the scanning data obtained by scanning the object to be detected from multiple angles can be combined together, and the characteristic information of the object to be detected can be more comprehensively obtained; finally, inputting the comprehensive scanning data into the set metal classification model to obtain the metal type contained in the object to be detected; the obtained type of the metal contained in the object to be detected is used as a basis for judging whether the detected metal or metal product belongs to dangerous articles or not and whether the detected metal or metal product belongs to forbidden metal articles, so that the problem of manual detection is solved, and the efficiency of safety detection or anti-theft detection is improved.
Description
Technical Field
The present application relates to the field of metal classification, and more particularly, to a metal classification method, apparatus, device, and storage medium.
Background
In many occasions of life or work, metal detection needs to be performed on passing people or luggage, for example, for safety, safety detection needs to be performed on personnel and carry-on luggage thereof entering public places such as high-speed rail stations, airports, hospitals and the like, and whether dangerous metal objects are carried or not is detected; in addition, in some raw material smelting plants or large-scale enterprises, the passing pedestrians and their carry-on luggage also need to be subjected to metal detection to prevent theft.
However, the existing metal detection equipment can only detect whether passers-by and carry-on luggage contain metal objects, and cannot identify the type of metal or the type of metal product to classify, so that the detected metal can not be judged to belong to metal products which are not dangerous in daily life or metal products which are dangerous or metal raw materials which are forbidden to be brought out by factories or enterprises. In this case, a human search by a detection person is required to determine whether the detected metal object is dangerous or belongs to an illegal portable object, which may reduce the inspection efficiency.
Therefore, it is highly desirable to provide a metal classification scheme for detecting the types of metal objects in pedestrians and their carry-on luggage, and without manually searching to confirm the types of metal objects, so as to improve the efficiency of security detection or anti-theft detection.
Disclosure of Invention
In view of the foregoing, the present application provides a method, an apparatus, a device and a storage medium for metal classification, so as to improve the efficiency of security detection and anti-theft detection. The specific scheme is as follows:
a metal sorting method comprising:
scanning data of each channel of an object to be detected is obtained, wherein the scanning data of each channel are obtained by scanning the object to be detected in a plurality of different coil induction areas respectively;
integrating the scanning data of each channel to obtain comprehensive scanning data;
inputting the comprehensive scanning data into a set metal classification model to obtain the type of metal contained in the object to be detected; the metal classification model is obtained by taking comprehensive scanning data obtained by integrating scanning data of all channels obtained by scanning a training object by a plurality of different coil induction areas as a training sample and taking the type of metal contained in the training object as a training label for training.
Optionally, the training process of the metal classification model includes:
scanning data of each channel of a training object are obtained, wherein the scanning data of each channel are obtained by respectively scanning the training object in a plurality of different coil induction areas;
Integrating the scanning data of each channel of the training object to obtain comprehensive training scanning data;
taking the type of metal contained in the training object as a sample label of the comprehensive training scanning data;
inputting the comprehensive training scanning data into the metal classification model to obtain the type of metal contained in the training object output by the metal classification model;
and updating the parameters of the metal classification model by taking the type of the metal contained in the output training object approaching to the type of the metal contained in the training object as a training target.
Optionally, the metal classification model includes an input layer, an intermediate layer, and a fully-connected output layer;
inputting the comprehensive scanning data into a set metal classification model to obtain the type of the metal contained in the object to be detected, wherein the method comprises the following steps:
performing band-pass filtering on the comprehensive scanning data by using the input layer to obtain effective waveform data;
performing wavelet denoising on the effective waveform data by using the intermediate layer to obtain effective characteristic information;
and obtaining the type of the metal contained in the object to be detected by utilizing the full-connection output layer based on the effective characteristic information.
Optionally, the input layer is a filter layer in a parameterized sinnet function network model sincenet; the middle layer is a residual error shrinkage module RSBU-CS in the depth residual error shrinkage network.
Optionally, the acquiring scan data of each channel of the training object includes:
and acquiring scanning data of each channel, which is obtained by scanning the training object by each different coil induction area when the training object carries different types of metals and passes through the plurality of different coil induction areas at different speeds.
A metal sorting apparatus comprising:
the scanning data acquisition unit is used for acquiring scanning data of each channel of the object to be detected, wherein the scanning data of each channel are obtained by scanning the object to be detected respectively in a plurality of different coil induction areas;
the data integration unit is used for integrating the scanning data of each channel to obtain comprehensive scanning data;
the model prediction unit is used for inputting the comprehensive scanning data into a set metal classification model to obtain the type of metal contained in the object to be detected; the metal classification model is obtained by taking comprehensive scanning data obtained by integrating scanning data of all channels obtained by scanning a training object by a plurality of different coil induction areas as a training sample and taking the type of metal contained in the training object as a training label for training.
Optionally, the metal classification device further includes a metal classification model training unit, the metal classification model training unit is configured to train the metal classification model, and a training process of the metal classification model includes:
scanning data of each channel of a training object are obtained, wherein the scanning data of each channel are obtained by respectively scanning the training object in a plurality of different coil induction areas;
integrating the scanning data of each channel of the training object to obtain comprehensive training scanning data;
taking the type of metal contained in the training object as a sample label of the comprehensive training scanning data;
inputting the comprehensive training scanning data into the metal classification model to obtain the type of metal contained in the training object output by the metal classification model;
and updating the parameters of the metal classification model by taking the type of the metal contained in the output training object approaching to the type of the metal contained in the training object as a training target.
Optionally, the metal classification model includes an input layer, an intermediate layer, and a fully-connected output layer; the model prediction unit includes:
the band-pass filtering unit is used for performing band-pass filtering on the comprehensive scanning data by utilizing the input layer to obtain effective waveform data;
The wavelet denoising unit is used for performing wavelet denoising on the effective waveform data by utilizing the intermediate layer to obtain effective characteristic information;
and the result acquisition unit is used for acquiring the type of the metal contained in the object to be detected based on the effective characteristic information by utilizing the full-connection output layer.
A metal sorting apparatus comprising: a memory and a processor;
the memory is used for storing programs;
the processor is used for executing the program and realizing the steps of the metal classification method.
A storage medium having stored thereon a computer program which, when executed by a processor, carries out the steps of the above-mentioned metal classification method.
By means of the technical scheme, the metal classification model can be trained in advance for metal classification. Specifically, the comprehensive scan data integrated with the scan data of the training object may be used as a training sample, and the type of the metal included in the training object may be used as a training label for training. After the metal classification model is obtained through training, the method and the device can respectively scan the object to be detected through a plurality of different coil induction areas to obtain the scanning data of each channel, further integrate the scanning data of each channel to obtain comprehensive scanning data, and thus can combine the scanning data obtained by scanning the object to be detected from a plurality of angles together, thereby more comprehensively obtaining the characteristic information of the object to be detected; and finally, inputting the comprehensive scanning data into the trained metal classification model to obtain the metal type contained in the object to be detected. The obtained metal type contained in the object to be detected can judge whether the detected metal or metal product belongs to dangerous articles or whether the detected metal or metal product belongs to prohibited carrying metal articles, so that the problem of manual detection is solved, and the efficiency of safety detection or anti-theft detection is improved.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly introduced below, it is obvious that the drawings in the following description are only embodiments of the present application, and for those skilled in the art, other drawings can be obtained according to the provided drawings without creative efforts.
Fig. 1 is a schematic flow chart of a metal classification method according to an embodiment of the present disclosure;
fig. 2 is a schematic structural diagram of a metal classification model according to an embodiment of the present disclosure;
FIG. 3 is a schematic diagram of the distribution and collection space of coils of a metal detection door according to an example of the present application;
fig. 4 is a schematic structural diagram of a metal sorting apparatus disclosed in an embodiment of the present application;
fig. 5 is a schematic structural diagram of a metal sorting apparatus according to an embodiment of the present application.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
In some specific public places, safety inspection is generally required to be carried out on personnel entering a site to check whether the personnel entering the site carry dangerous goods, particularly metal dangerous goods; on the other hand, at the exit of some metal smelting plants and enterprises, it is generally necessary to check whether the persons away carry the metal raw materials prohibited from being carried out, so as to prevent the metal smelting raw materials from being stolen.
At present, a metal detector or a metal detection door only detects whether detected people carry metal, but the detected people may carry non-dangerous metal living goods, and if only detecting whether the detected people carry metal, whether the detected metal belongs to forbidden metal goods cannot be confirmed. In this case, it is necessary to manually confirm whether the detected metal belongs to a dangerous metal article or a metal material which is prohibited from being taken out, which reduces the detection efficiency. Therefore, it is desirable to provide a scheme for classifying metals carried by a detection object, which can distinguish the types of metal articles. The application provides a metal classification scheme, which can be suitable for solving the task that various types need metal classification.
Next, a detailed description is given of a metal classification method provided in the present application, please refer to fig. 1, and fig. 1 is a schematic flow chart of the metal classification method provided in the present application. The metal classification method provided by the embodiment of the application can comprise the following steps:
step S100, scanning data of each channel of the object to be detected is obtained.
Specifically, the object to be detected can be scanned through a plurality of different coil induction areas, so that scanning data of each channel of the object to be detected can be obtained. The object to be detected can be a person carrying personal belongings, and when the person carrying the personal belongings passes through the metal detection device, such as a metal detection door and a metal detector, a plurality of coils arranged on the metal detection device can scan the passing person to acquire scanning data of each channel.
In the embodiment of the present application, the scan data obtained by scanning the object to be detected by one coil may be represented as the scan data of one channel, and so on, the scan data of multiple channels may be obtained by scanning the object to be detected by multiple different coils. The number of the coil induction areas can be 6 or 10, and the specific number of the coils can be determined according to the type and the size of the metal detection equipment. The purpose of setting a plurality of coils is to scan the same object to be detected from a plurality of angles and acquire richer scanning data which can represent characteristic information of the object to be detected more comprehensively.
And step S110, integrating the scanning data of each channel to obtain comprehensive scanning data.
Specifically, the scanning data of each channel of the object to be detected, which is obtained in step S100, is integrated to obtain comprehensive scanning data. The scanning data of each channel can be obtained by scanning the object to be detected from multiple angles through a plurality of coil induction areas, the scanning data of each channel obtained from multiple angles can be integrated, and richer comprehensive scanning data which can comprehensively represent the object to be detected are obtained. In this step, the scanning data of each channel may be integrated by superimposing waveform data obtained by scanning the object to be detected in each coil induction region, and integrating the waveform data into comprehensive scanning data, which is used in the subsequent data processing step.
And step S120, inputting the comprehensive scanning data into a set metal classification model to obtain the type of the metal contained in the object to be detected.
Specifically, the comprehensive scan data obtained in step S110 is used in this step, and the type of the metal contained in the object to be detected can be obtained by inputting the comprehensive scan data into a preset metal classification model. In this step, the preset metal classification model may be obtained by training with the integrated scan data integrated from the scan data of each channel as a training sample and the type of metal included in the training object as a training label. The scanning data of each channel may be obtained by scanning the training object through a plurality of different coil induction areas respectively.
By means of the technical scheme, the metal classification model can be trained in advance for metal classification. Specifically, the comprehensive scan data integrated with the scan data of the training object may be used as a training sample, and the type of the metal included in the training object may be used as a training label for training. After the metal classification model is obtained through training, the method and the device can respectively scan the object to be detected through a plurality of different coil induction areas to obtain the scanning data of each channel, further integrate the scanning data of each channel to obtain comprehensive scanning data, and thus can combine the scanning data obtained by scanning the object to be detected from a plurality of angles together, thereby more comprehensively obtaining the characteristic information of the object to be detected; and finally, inputting the comprehensive scanning data into the trained metal classification model to obtain the metal type contained in the object to be detected. The obtained metal type contained in the object to be detected can judge whether the detected metal or metal product belongs to dangerous articles or whether the detected metal or metal product belongs to prohibited carrying metal articles, so that the problem of manual detection is solved, and the efficiency of safety detection or anti-theft detection is improved.
In some embodiments of the present application, a training process of the preset metal classification model is described, in which the training process of the metal classification model may include the following steps:
and S1, acquiring the scanning data of each channel of the training object.
Specifically, the training object is scanned through a plurality of different coil induction areas, so that the scanning data of each channel can be obtained, and the scanning data of each channel obtained in the step can represent the training object from a plurality of different angles.
And S2, integrating the scanning data of each channel of the training object to obtain comprehensive training scanning data.
Specifically, the scanning data of each channel obtained in step S1 are subjected to waveform superposition to obtain integrated comprehensive training scanning data. Dividing the integrated training data into two sets of scan data, respectively trainingExercise and collection phitrainAnd test set Φtest。ΦtrainFor training metal classification models, phitestAnd the method is used for verifying whether the obtained metal classification model can output an accurate detection result.
And S3, using the metal type contained in the training object as a sample label of the comprehensive training scanning data.
Specifically, the obtained comprehensive training scan data obtained by integrating the scan data of each channel of the training object may use the type of the metal contained in the training object as a sample label of the comprehensive training scan data to the training set Φ trainAnd test set ΦtestAnd carrying out manual marking, and further training the metal classification model.
And S4, inputting the comprehensive training scanning data into the metal classification model to obtain the type of the metal contained in the training object output by the metal classification model.
And S5, taking the type of the metal contained in the output training object approaching to the type of the metal contained in the training object as a training target, and updating the parameters of the metal classification model.
Specifically, when the metal classification model is trained, the training set Φ obtained in step S3 is usedtrainAnd inputting the type of the metal contained in the training object output by the metal classification model into the metal classification model. The type of the metal contained in the training object output by the metal classification model is a prediction result, and the parameter of the metal classification model is updated by taking the prediction result, that is, the type of the metal contained in the training object output by the metal classification model, approaching to the sample label, that is, the type of the metal contained in the training object as a training target, so as to finally obtain the trained metal classification model.
Further, the test set Φ of the above step S3 may betestInputting the obtained trained metal classification model to verify whether the obtained metal classification model can output an accurate detection result, and if so, using the metal classification model; if not, continuing to expand the training sample data volume to the metal score And training the class model, and further updating the parameters of the model.
In some embodiments of the present application, the metal classification model may include an input layer, an intermediate layer, and a fully connected output layer. Referring to fig. 2, fig. 2 is a schematic structural diagram of a metal classification model according to an embodiment of the present disclosure. The process of processing the acquired data by using the input layer, the intermediate layer and the full connection output layer may include:
and S1, performing band-pass filtering on the comprehensive scanning data by using the input layer to obtain effective waveform data.
Specifically, the input layer may be a first convolution layer of the metal classification model, and the input layer is used to allow the comprehensive scanning data in the specific frequency band to pass through, and shield other waveform data that do not conform to the specific frequency band, so as to obtain effective waveform data. Specifically, the input layer allows frequency components in a certain frequency range to pass through, attenuating frequency components in other ranges to extremely low levels. In this embodiment, the integrated scanbook data may be represented as 1 × nch × nSamples, where nch may be represented as the number of channels, and nSamples may be represented as the number of sampling points. The sampling points may be spatial positions of uniformly distributed collection points within a scanning range of the metal detection device. After the comprehensive scanning data 1 × nch × nSamples passes through the input layer, waveform data which does not conform to the specific frequency band in the data 1 × nch × nSamples can be filtered out, so that part of interference information is removed.
And S2, performing wavelet denoising on the effective waveform data by using the intermediate layer to obtain effective characteristic information.
Specifically, the intermediate layer may further remove noise from the effective waveform data, where the noise may be understood as "feature information unrelated to the current metal classification task", and the intermediate layer performs wavelet denoising on the effective waveform data, so as to remove redundant feature information in the effective waveform data and obtain effective feature information.
And S3, obtaining the type of the metal contained in the object to be detected by using the full-connection output layer based on the effective characteristic information.
Specifically, on the basis of the effective feature information obtained in step S2, the fully-connected output layer may output the type of the metal contained in the object to be detected, and complete the classification task of the metal classification model.
In this embodiment of the present application, the input layer of the metal classification model may be a filter layer in a sinnet based on a parameterized sine function network model, and the filter layer in the sinnet may be denoted as sinnet Filters. The sincent Filters can perform convolution operation on the waveform of the comprehensive scanning data and a set of parameterized sinc functions for realizing band-pass filtering, so that the band-pass filtering of the comprehensive scanning data is realized.
In addition, the middle layer of the metal classification model can be composed of a plurality of Residual Shrinkage modules RSBU-CS, and the Residual Shrinkage modules RSBU-CS can be Residual Shrinkage layers in a deep Residual Shrinkage network, are Residual modules sharing a threshold value among channels, and are called Residual shock Building units with Channel-Shared thresholds for short. Compared with the traditional residual shrinking module, the RSBU-CS residual module has a small sub-network, and the sub-network can adaptively set the threshold value.
In the embodiment of the application, the effective characteristic information in the comprehensive scanning data can be better extracted through the combination of the two network layers SincNet Filters and the RSBU-CS, so that the classification accuracy of the metal classification model is improved, the model parameters of the two network layers SincNet Filters and the RSBU-CS are few, and the complexity of model training can be reduced.
In some embodiments of the present application, the obtaining of the scanning data of each channel of the training object of the metal classification model may be obtaining of the scanning data of each channel obtained by scanning the training object by each different coil induction area when the training object carries different types of metals and passes through a plurality of different coil induction areas at different speeds.
Specifically, the training object may be a passer who carries different types of metal objects, for example, the passer carries a mobile phone, a key, a pen box, a lunch box, scissors, coins, MP3, a cup, a pen, a magnet toy, a silver bracelet, etc. with him, and the metal detection device scans the passer who carries different types of metal objects. Meanwhile, the application of the scheme considers that different types of metal objects can have different states, and the characteristic information of the scanning data of each channel of the training object can be different. Therefore, the training subject also considers the person carrying different types of metal objects placed in different postures, such as horizontal, inclined and vertical.
On the other hand, because the passing speed of the passing person passing through the metal detection device is different, different types of metal objects carried by the passing person can also pass through a plurality of different coil induction areas at different speeds, and in such a case, characteristic information of scanning data of each channel obtained by scanning the training object is also different. Therefore, the present application also considers the passing speed of the training object, and scans the training object passing through the coil induction area at different passing speeds, wherein the passing speed of the training object can be 0.2m/s, 0.4m/s, 0.6m/s, 0.8m/s, 1m/s, 1.2m/s, 1.4m/s, 1.8m/s, 2 m/s.
In the embodiment of the present application, please refer to fig. 3, and fig. 3 is a schematic diagram of distribution and acquisition space of each coil of a metal detection door according to an example of the present application. The training object under the multiple conditions is scanned through the multiple coil induction areas, the scanning data of each channel is obtained, and meanwhile, the scanning data of each channel are sampled through multiple spatial positions. Therefore, diverse training data can be obtained, the trained metal classification model can have better robustness under the condition of diversity of the training objects, and the object to be detected can be detected under various different conditions to obtain an accurate detection result.
The following describes a metal sorting device provided in an embodiment of the present application, and the metal sorting device described below and the metal sorting method described above may be referred to correspondingly.
Referring to fig. 4, fig. 4 is a schematic structural diagram of a metal sorting apparatus disclosed in the embodiment of the present application.
As shown in fig. 4, the apparatus may include:
the scanning data acquiring unit 11 is configured to acquire scanning data of each channel of an object to be detected, where the scanning data of each channel is obtained by scanning a plurality of different coil induction areas respectively on the object to be detected;
A data integration unit 12, configured to integrate the scanning data of each channel to obtain comprehensive scanning data;
the model prediction unit 13 is configured to input the comprehensive scanning data into a set metal classification model to obtain a type of metal contained in the object to be detected; the metal classification model is obtained by taking comprehensive scanning data obtained by integrating scanning data of all channels obtained by scanning a training object by a plurality of different coil induction areas as a training sample and taking the type of metal contained in the training object as a training label for training.
Optionally, the metal classification device further includes a metal classification model training unit, where the metal classification model training unit is configured to train the metal classification model, and a training process of the metal classification model includes:
scanning data of each channel of a training object are obtained, wherein the scanning data of each channel are obtained by respectively scanning the training object in a plurality of different coil induction areas;
integrating the scanning data of each channel of the training object to obtain comprehensive training scanning data;
taking the type of metal contained in the training object as a sample label of the comprehensive training scanning data;
Inputting the comprehensive training scanning data into the metal classification model to obtain the type of metal contained in the training object output by the metal classification model;
and updating the parameters of the metal classification model by taking the type of the metal contained in the output training object approaching to the type of the metal contained in the training object as a training target.
Optionally, the metal classification model includes an input layer, an intermediate layer, and a fully-connected output layer; the model prediction unit 13 includes:
the band-pass filtering unit is used for performing band-pass filtering on the comprehensive scanning data by utilizing the input layer to obtain effective waveform data;
the wavelet denoising unit is used for performing wavelet denoising on the effective waveform data by utilizing the intermediate layer to obtain effective characteristic information;
and the result acquisition unit is used for acquiring the type of the metal contained in the object to be detected based on the effective characteristic information by utilizing the full-connection output layer.
Optionally, the model prediction unit 13 includes:
and the training object scanning unit is used for acquiring scanning data of each channel, which is obtained by scanning the training object by each different coil induction area when the training object carries different types of metals and passes through the plurality of different coil induction areas at different speeds.
The metal classification device provided by the embodiment of the application can be applied to metal classification equipment, such as a metal detector, a metal detection door, a luggage security inspection machine and the like. Optionally, fig. 5 is a schematic structural diagram of a metal classification device provided in an embodiment of the present application, and referring to fig. 5, a hardware structure of the metal classification device may include: at least one processor 1, at least one communication interface 2, at least one memory 3 and at least one communication bus 4;
in the embodiment of the application, the number of the processor 1, the communication interface 2, the memory 3 and the communication bus 4 is at least one, and the processor 1, the communication interface 2 and the memory 3 complete mutual communication through the communication bus 4;
the processor 1 may be a central processing unit CPU, or an application Specific Integrated circuit asic, or one or more Integrated circuits configured to implement embodiments of the present invention, etc.;
the memory 3 may include a high-speed RAM memory, and may further include a non-volatile memory (non-volatile memory) or the like, such as at least one disk memory;
wherein the memory stores a program and the processor can call the program stored in the memory, the program for:
Scanning data of each channel of an object to be detected is obtained, wherein the scanning data of each channel are obtained by scanning the object to be detected in a plurality of different coil induction areas respectively;
integrating the scanning data of each channel to obtain comprehensive scanning data;
inputting the comprehensive scanning data into a set metal classification model to obtain the type of metal contained in the object to be detected; the metal classification model is obtained by taking comprehensive scanning data obtained by integrating scanning data of all channels obtained by scanning a training object by a plurality of different coil induction areas as a training sample and taking the type of metal contained in the training object as a training label for training.
Alternatively, the detailed function and the extended function of the program may be as described above.
Embodiments of the present application further provide a storage medium, where a program suitable for execution by a processor may be stored, where the program is configured to:
scanning data of each channel of an object to be detected is obtained, wherein the scanning data of each channel are obtained by scanning the object to be detected in a plurality of different coil induction areas respectively;
integrating the scanning data of each channel to obtain comprehensive scanning data;
Inputting the comprehensive scanning data into a set metal classification model to obtain the type of metal contained in the object to be detected; the metal classification model is obtained by taking comprehensive scanning data obtained by integrating scanning data of all channels obtained by scanning a training object by a plurality of different coil induction areas as a training sample and taking the type of metal contained in the training object as a training label for training.
Alternatively, the detailed function and the extended function of the program may be as described above.
Finally, it should also be noted that, herein, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions. Also, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
The embodiments in the present description are described in a progressive manner, each embodiment focuses on differences from other embodiments, the embodiments may be combined as needed, and the same and similar parts may be referred to each other.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the application. Thus, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
Claims (10)
1. A method of classifying a metal, comprising:
scanning data of each channel of an object to be detected is obtained, wherein the scanning data of each channel are obtained by scanning the object to be detected in a plurality of different coil induction areas respectively;
integrating the scanning data of each channel to obtain comprehensive scanning data;
inputting the comprehensive scanning data into a set metal classification model to obtain the type of metal contained in the object to be detected; the metal classification model is obtained by taking comprehensive scanning data obtained by integrating scanning data of all channels obtained by scanning a training object by a plurality of different coil induction areas as a training sample and taking the type of metal contained in the training object as a training label for training.
2. The method of claim 1, wherein the training process of the metal classification model comprises:
scanning data of each channel of a training object are obtained, wherein the scanning data of each channel are obtained by respectively scanning the training object in a plurality of different coil induction areas;
integrating the scanning data of each channel of the training object to obtain comprehensive training scanning data;
taking the type of metal contained in the training object as a sample label of the comprehensive training scanning data;
inputting the comprehensive training scanning data into the metal classification model to obtain the type of metal contained in the training object output by the metal classification model;
and updating the parameters of the metal classification model by taking the type of the metal contained in the output training object approaching to the type of the metal contained in the training object as a training target.
3. The method of claim 1, wherein the metal classification model comprises an input layer, an intermediate layer, a fully connected output layer;
inputting the comprehensive scanning data into a set metal classification model to obtain the type of the metal contained in the object to be detected, wherein the method comprises the following steps:
Performing band-pass filtering on the comprehensive scanning data by using the input layer to obtain effective waveform data;
performing wavelet denoising on the effective waveform data by using the intermediate layer to obtain effective characteristic information;
and obtaining the type of the metal contained in the object to be detected by utilizing the full-connection output layer based on the effective characteristic information.
4. The method of claim 3, wherein the input layer is a filter layer in a parameterized-based SincNet network model; the middle layer is a residual error shrinkage module RSBU-CS in the depth residual error shrinkage network.
5. The method of claim 2, wherein the acquiring scan data for each channel of the training subject comprises:
and acquiring scanning data of each channel, which is obtained by scanning the training object by each different coil induction area when the training object carries different types of metals and passes through the plurality of different coil induction areas at different speeds.
6. A metal sorting device, comprising:
the scanning data acquisition unit is used for acquiring scanning data of each channel of the object to be detected, wherein the scanning data of each channel are obtained by scanning the object to be detected respectively in a plurality of different coil induction areas;
The data integration unit is used for integrating the scanning data of each channel to obtain comprehensive scanning data;
the model prediction unit is used for inputting the comprehensive scanning data into a set metal classification model to obtain the type of metal contained in the object to be detected; the metal classification model is obtained by taking comprehensive scanning data obtained by integrating scanning data of all channels obtained by scanning a training object by a plurality of different coil induction areas as a training sample and taking the type of metal contained in the training object as a training label for training.
7. The method according to claim 6, wherein the metal classification device further comprises a metal classification model training unit, the metal classification model training unit is used for training the metal classification model, and the training process of the metal classification model comprises:
scanning data of each channel of a training object are obtained, wherein the scanning data of each channel are obtained by respectively scanning the training object in a plurality of different coil induction areas;
integrating the scanning data of each channel of the training object to obtain comprehensive training scanning data;
taking the type of metal contained in the training object as a sample label of the comprehensive training scanning data;
Inputting the comprehensive training scanning data into the metal classification model to obtain the type of metal contained in the training object output by the metal classification model;
and updating the parameters of the metal classification model by taking the type of the metal contained in the output training object approaching to the type of the metal contained in the training object as a training target.
8. The metal classification device of claim 6, wherein the metal classification model comprises an input layer, an intermediate layer, a fully connected output layer; the model prediction unit includes:
the band-pass filtering unit is used for performing band-pass filtering on the comprehensive scanning data by utilizing the input layer to obtain effective waveform data;
the wavelet denoising unit is used for performing wavelet denoising on the effective waveform data by utilizing the intermediate layer to obtain effective characteristic information;
and the result acquisition unit is used for acquiring the type of the metal contained in the object to be detected based on the effective characteristic information by utilizing the full-connection output layer.
9. A metal sorting apparatus, comprising: a memory and a processor;
the memory is used for storing programs;
the processor is used for executing the program and realizing the steps of the metal classification method according to any one of claims 1 to 5.
10. A storage medium having stored thereon a computer program, wherein the computer program, when executed by a processor, carries out the steps of the metal classification method according to any one of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110797634.4A CN113537343A (en) | 2021-07-14 | 2021-07-14 | Metal classification method, device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110797634.4A CN113537343A (en) | 2021-07-14 | 2021-07-14 | Metal classification method, device, equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113537343A true CN113537343A (en) | 2021-10-22 |
Family
ID=78099190
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110797634.4A Pending CN113537343A (en) | 2021-07-14 | 2021-07-14 | Metal classification method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113537343A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114662550A (en) * | 2022-05-24 | 2022-06-24 | 安徽启新明智科技有限公司 | Neural network training data acquisition method based on metal detection gate and storage medium |
| CN115795353A (en) * | 2023-02-06 | 2023-03-14 | 北京科技大学 | Underground metal target classification method and system based on unbalanced data set |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102759753A (en) * | 2011-04-29 | 2012-10-31 | 同方威视技术股份有限公司 | Method and device for detecting hidden dangerous article |
| CN103442815A (en) * | 2011-01-07 | 2013-12-11 | 休伦瓦雷钢铁公司 | Scrap metal sorting system |
| CN105427614A (en) * | 2015-08-28 | 2016-03-23 | 北京动视元科技有限公司 | Model classification system and method |
| WO2017162107A1 (en) * | 2016-03-25 | 2017-09-28 | 东莞市华盾电子科技有限公司 | Discrimination method and test method for detection of metal classification and system thereof |
| CN109655928A (en) * | 2019-01-30 | 2019-04-19 | 成都智元汇信息技术股份有限公司 | A kind of detector gate metal analysis system and method based on cloud processing |
| WO2020134848A1 (en) * | 2018-12-28 | 2020-07-02 | 深圳市华讯方舟太赫兹科技有限公司 | Intelligent detection method and device applied to millimeter wave security check instrument, and storage device |
| CN111399069A (en) * | 2020-03-20 | 2020-07-10 | 惠州市迈测科技有限公司 | Metal detection device |
| CN111667464A (en) * | 2020-05-21 | 2020-09-15 | 平安科技（深圳）有限公司 | Dangerous goods three-dimensional image detection method and device, computer equipment and storage medium |
| CN111751896A (en) * | 2019-03-27 | 2020-10-09 | 北京快安科技有限公司 | Three-dimensional display method for security inspection result |
-
2021
- 2021-07-14 CN CN202110797634.4A patent/CN113537343A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103442815A (en) * | 2011-01-07 | 2013-12-11 | 休伦瓦雷钢铁公司 | Scrap metal sorting system |
| CN102759753A (en) * | 2011-04-29 | 2012-10-31 | 同方威视技术股份有限公司 | Method and device for detecting hidden dangerous article |
| CN105427614A (en) * | 2015-08-28 | 2016-03-23 | 北京动视元科技有限公司 | Model classification system and method |
| WO2017162107A1 (en) * | 2016-03-25 | 2017-09-28 | 东莞市华盾电子科技有限公司 | Discrimination method and test method for detection of metal classification and system thereof |
| WO2020134848A1 (en) * | 2018-12-28 | 2020-07-02 | 深圳市华讯方舟太赫兹科技有限公司 | Intelligent detection method and device applied to millimeter wave security check instrument, and storage device |
| CN109655928A (en) * | 2019-01-30 | 2019-04-19 | 成都智元汇信息技术股份有限公司 | A kind of detector gate metal analysis system and method based on cloud processing |
| CN111751896A (en) * | 2019-03-27 | 2020-10-09 | 北京快安科技有限公司 | Three-dimensional display method for security inspection result |
| CN111399069A (en) * | 2020-03-20 | 2020-07-10 | 惠州市迈测科技有限公司 | Metal detection device |
| CN111667464A (en) * | 2020-05-21 | 2020-09-15 | 平安科技（深圳）有限公司 | Dangerous goods three-dimensional image detection method and device, computer equipment and storage medium |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114662550A (en) * | 2022-05-24 | 2022-06-24 | 安徽启新明智科技有限公司 | Neural network training data acquisition method based on metal detection gate and storage medium |
| CN115795353A (en) * | 2023-02-06 | 2023-03-14 | 北京科技大学 | Underground metal target classification method and system based on unbalanced data set |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113537343A (en) | Metal classification method, device, equipment and storage medium | |
| CN109214274B (en) | Airport security management system | |
| CN107358146A (en) | Method for processing video frequency, device and storage medium | |
| CN104616494B (en) | Method and device for recording and determining target object based on base station and block port | |
| JP2005521975A5 (en) | ||
| CN107172590A (en) | Moving state information processing method, device and mobile terminal based on mobile terminal | |
| CN104036177A (en) | Intelligent terminal fingerprint unlocking device and method | |
| CN107451455A (en) | Solve lock control method and Related product | |
| CN108009690A (en) | A kind of routine bus system optimized based on modularity steals group's automatic testing method | |
| CN103279744A (en) | Multi-scale tri-mode texture feature-based method and system for detecting counterfeit fingerprints | |
| CN112101234B (en) | Detection code matching processing method and graph code joint detection system | |
| Marivate et al. | Catching crime: Detection of public safety incidents using social media | |
| CN105138886B (en) | Robot biometric identification system | |
| CN105843570A (en) | Display method and apparatus for privacy protection | |
| US10528950B2 (en) | System and a method for detecting fraudulent transactions at a transaction site | |
| CN107844737A (en) | Iris image detection method and device | |
| CN110046245A (en) | A kind of data monitoring method and device, a kind of calculating equipment and storage medium | |
| CN110301892A (en) | A kind of detection method and Related product based on hand vein recognition | |
| CN202404694U (en) | Adaptive disturbance signal identification module of distributing type optical fiber sensing application system | |
| CN114139021B (en) | Index information management method and system | |
| CN110428402A (en) | Distorted image recognition methods, device, computer equipment and storage medium | |
| CN114998837A (en) | Method, device, system and medium for identifying intrusion events in real time | |
| CN109308782A (en) | A kind of behavioral value method and system of target object | |
| CN107730717A (en) | A kind of suspicious card identification method of public transport of feature based extraction | |
| Pandey | An amalgamated strategy for iris recognition employing neural network and hamming distance |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |