CN111046877A - Millimeter wave image suspicious article detection method and system - Google Patents
Millimeter wave image suspicious article detection method and system Download PDFInfo
- Publication number
- CN111046877A CN111046877A CN201911324885.XA CN201911324885A CN111046877A CN 111046877 A CN111046877 A CN 111046877A CN 201911324885 A CN201911324885 A CN 201911324885A CN 111046877 A CN111046877 A CN 111046877A
- Authority
- CN
- China
- Prior art keywords
- image
- target
- millimeter wave
- suspicious
- standard image
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 98
- 238000013507 mapping Methods 0.000 claims abstract description 48
- 238000012545 processing Methods 0.000 claims abstract description 22
- 238000002372 labelling Methods 0.000 claims description 12
- 238000000605 extraction Methods 0.000 claims description 11
- 238000012549 training Methods 0.000 claims description 11
- 238000013527 convolutional neural network Methods 0.000 claims description 7
- 238000011176 pooling Methods 0.000 claims description 6
- 238000006243 chemical reaction Methods 0.000 claims description 4
- 238000004458 analytical method Methods 0.000 claims description 3
- 238000000034 method Methods 0.000 abstract description 13
- 238000003384 imaging method Methods 0.000 abstract description 11
- 230000000007 visual effect Effects 0.000 abstract description 4
- 238000010586 diagram Methods 0.000 description 12
- 239000010410 layer Substances 0.000 description 9
- 238000004590 computer program Methods 0.000 description 7
- 238000004422 calculation algorithm Methods 0.000 description 5
- 238000013528 artificial neural network Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000007689 inspection Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000013135 deep learning Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 239000002360 explosive Substances 0.000 description 1
- 239000007788 liquid Substances 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000002344 surface layer Substances 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30196—Human being; Person
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/05—Recognition of patterns representing particular kinds of hidden objects, e.g. weapons, explosives, drugs
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/07—Target detection
Abstract
The scheme provides a millimeter wave image suspicious article detection method and a millimeter wave image suspicious article detection system, wherein the method comprises the following steps: processing the millimeter wave image of the target to obtain a multi-view target mapping standard image; and identifying the target mapping standard image based on a pre-constructed target detection network model, and determining the position and the category of the suspicious object. The technical scheme of the application can carry out joint estimation on the image detection results under each visual angle, can effectively improve the detection accuracy, and can inhibit false alarms caused by noise interference under a certain angle; by utilizing the millimeter wave imaging results under all angles of the human body for detection, the detection range coverage is wider, and suspicious object images hidden in all parts of the human body can be obtained.
Description
Technical Field
The application relates to the field of millimeter wave detection, in particular to a millimeter wave image suspicious article detection system method and system based on multi-view images.
Background
The millimeter wave imaging technology is a research hotspot in the security inspection field in recent years because the millimeter wave imaging technology can penetrate through clothes to image the surface of a human body, is harmless to the human body, and can effectively detect suspicious objects hidden on the surface layer of the human body, such as pistols, explosives, liquid and the like. The active human body security inspection system is widely applied due to the advantages of small influence of environmental factors, high image signal to noise ratio and the like.
The deep neural network plays an important role in the field of image target detection, has an end-to-end characteristic, and can learn the deep level characteristics of suspicious objects and give the positions and attributes of the suspicious objects in an input image through proper training. The target detection mainly comprises an One-stage algorithm and a Two-stage algorithm, the detection speed of the One-stage algorithm is high, but the detection precision is low, and the detection time of the Two-stage algorithm is relatively long, but the detection precision is higher. Rapid target detection (FasterRCNN) is representative of Two-stage methods and has excellent detection performance.
In addition, suspicious item detection based on millimeter wave images remains problematic. Millimeter wave imaging is easily interfered by noise, so that pollution such as texture and clutter which cannot be eliminated appears in an image, difficulty is increased for suspicious article detection, and high requirements on detection rate and false alarm rate are difficult to meet.
Disclosure of Invention
The application provides a millimeter wave image suspicious object detection method and system based on multi-view images, so that millimeter wave detection precision is improved.
According to a first aspect of the embodiments of the present application, there is provided a millimeter wave image suspicious item detection method, including the steps of:
processing the millimeter wave image of the target to obtain a multi-view target mapping standard image;
and identifying the target mapping standard image based on a pre-constructed target detection network model, and determining the position and the category of the suspicious object.
In a preferred embodiment, the step of processing the millimeter-wave image of the target to obtain the multi-view target mapping standard image further includes:
and acquiring a millimeter wave scanning image of the target.
In a preferred embodiment, the processing the millimeter wave image of the target to obtain the multi-view target mapping standard image includes:
taking each image of the front and back of the target human body as a standard image;
and performing coordinate conversion on the millimeter wave images, and mapping image pixels corresponding to the front side and image pixels corresponding to the back side in the millimeter wave images of the target to the front side standard image and the back side standard image of the target human body respectively according to the rotation geometric relationship between the millimeter wave images to obtain the mapping standard images of the front side and the back side of the target human body.
In a preferred embodiment, the step of constructing the object detection network model includes:
marking the position and the category of the suspicious object in the target mapping standard image;
based on the convolutional neural network, a front standard image and a back standard image of a target human body are used as input, the standards are used as training labels, and the target detection network is trained to obtain a target detection network model.
In a preferred embodiment, the step of identifying the target mapping standard image based on the pre-constructed target detection network model and determining the position and the category of the suspicious item includes:
performing feature extraction on the target mapping standard image to obtain a feature image;
generating a plurality of candidate regions according to the characteristic image;
converting the candidate regions into feature images with the same resolution;
processing the characteristic images of the candidate regions based on the full connection layers, labeling the position and the category of the suspicious object in the target human body image, and generating confidence corresponding to the labeled position.
In a preferred embodiment, the step of identifying the target mapping standard image based on the pre-constructed target detection network model, and determining the location and the category of the suspicious item further includes:
comparing the confidence corresponding to the labeling position with a preset threshold;
if the confidence coefficient is greater than or equal to a preset threshold value, determining that suspicious articles exist at the marked position;
and if the confidence coefficient is smaller than a preset threshold value, determining that no suspicious articles exist at the marked position.
According to a second aspect of the embodiments of the present application, there is provided a millimeter wave image suspicious item detection system, including:
the mapping unit is used for processing the millimeter wave image of the target to obtain a multi-view target mapping standard image;
and the identification unit is used for identifying the target mapping standard image based on a pre-constructed target detection network model and determining the position and the category of the suspicious object.
In a preferred embodiment, the system further comprises: and the millimeter wave scanning device is used for performing millimeter wave scanning on the target human body to obtain a millimeter wave scanning image of the target.
In a preferred embodiment, the mapping unit specifically performs the following steps:
marking the position and the category of the suspicious object in the target mapping standard image;
based on the convolutional neural network, a front standard image and a back standard image of a target human body are used as input, the standards are used as training labels, and the target detection network is trained to obtain a target detection network model.
In a preferred embodiment, the identification unit comprises:
the characteristic extraction module is used for extracting the characteristics of the target mapping standard image to obtain a characteristic image;
a region generation module for generating a plurality of candidate regions according to the feature image;
the interest area pooling layer is used for converting the candidate areas into characteristic images with the same resolution;
the classification positioning module is used for processing the characteristic images of the candidate regions based on the full connection layers, marking the position and the category of the suspicious object in the target human body image and marking the confidence coefficient corresponding to the position;
the analysis module is used for comparing the confidence coefficient with a preset threshold value; if the confidence coefficient is greater than or equal to a preset threshold value, determining that suspicious articles exist at the marked position; and if the confidence coefficient is smaller than a preset threshold value, determining that no suspicious articles exist at the marked position.
Advantageous effects
The technical scheme of the application can carry out joint estimation on the image detection results under each visual angle, can effectively improve the detection accuracy, and can inhibit false alarms caused by noise interference under a certain angle; by utilizing the millimeter wave imaging results under all angles of the human body for detection, the detection range coverage is wider, and suspicious object images hidden in all parts of the human body can be obtained.
The technical scheme is simple in calculation and high in data rate, greatly reduces the workload of labeling, and can meet the requirement of target detection on high data rate.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the application and together with the description serve to explain the application and not to limit the application. In the drawings:
fig. 1 is a schematic diagram illustrating a millimeter wave image suspicious item detection method according to the present embodiment;
fig. 2 shows a schematic diagram of the millimeter wave scanning device according to the present embodiment;
FIG. 3 is a schematic diagram of a target detection network model according to the present solution;
fig. 4 shows a schematic diagram of the recognition unit according to the present solution.
Detailed Description
In order to make the technical solutions and advantages of the embodiments of the present application more apparent, the following further detailed description of the exemplary embodiments of the present application with reference to the accompanying drawings makes it clear that the described embodiments are only a part of the embodiments of the present application, and are not exhaustive of all embodiments. It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict.
The existing millimeter wave image detection method for hiding suspicious articles generally detects based on a single millimeter wave image, and then fuses detection results under one-time scanning and multiple angles, so that judgment of whether a human body carries dangerous articles is given. The judgment method has the advantages of low detection speed and low detection precision, and cannot meet the requirements of modern rapid and accurate detection. Therefore, the deep neural network detection model based on the multi-view millimeter wave image is designed, images at all angles are scanned at one time and used as neural network input, feature extraction and recognition are carried out, and finally a judgment result is given. The network detects multi-angle continuous images, meets the requirement of rapid detection and has higher detection precision.
As shown in fig. 1, the present disclosure discloses a millimeter wave image suspicious object detection method, which needs to process a millimeter wave image of a target to obtain a multi-view target mapping standard image; then, the target mapping standard image is identified based on a pre-constructed target detection network model, and the position and the category of the suspicious object are determined.
In the scheme, in order to more accurately detect suspicious articles, the cylindrical scanning array is adopted to carry out millimeter wave scanning imaging on a human body in an all-dimensional mode, and human body imaging results under multiple angles are obtained. Therefore, suspicious articles hidden in various positions of the human body can be effectively detected. Moreover, through observation of continuous angles, textures or noise can be further distinguished, and false alarms are effectively suppressed.
In the scheme, in the step of processing the millimeter wave image of the target to obtain the target mapping standard image with multiple visual angles, each image on the front side and the back side of the target human body can be used as the standard image; then, the millimeter wave images are subjected to coordinate conversion, image pixels corresponding to the front side and image pixels corresponding to the back side in the millimeter wave images of the target are respectively mapped into the front side standard image and the back side standard image of the target human body according to the rotation geometric relationship between the millimeter wave images, and the mapping standard images of the front side and the back side of the target human body are obtained
In one embodiment, the pixels of all images obtained in one scan are subjected to coordinate transformation by taking each image of the front and back of the human body as a standard map according to the rotational geometric relationship between the images. Namely, all the front images are mapped into the front standard graph, and all the back images are mapped into the back standard graph, so that a plurality of images under the same viewing angle of the front or the back are obtained.
In the scheme, the millimeter wave image of the target human body needs to be processed to obtain a multi-view target mapping standard image. In one embodiment, all front images are mapped into a front standard map and all back images are mapped into a back standard map, resulting in several images at the same viewing angle for either the front or the back.
In the scheme, a target detection network model needs to be constructed first, and then the constructed target detection network model is used for identifying a target mapping standard image and determining the position and the type of suspicious articles hidden in a target human body image. In one embodiment, the location and category of the suspicious object is marked in the target mapping standard image by using a rectangular frame; and training by adopting a typical network faster region convolutional neural network (FasterRCNN) to obtain a target detection network model. When a target detection network model is utilized, firstly, feature extraction is carried out on an image to be detected to obtain a feature image (feature map); generating a plurality of candidate regions according to the characteristic image; converting the candidate regions into feature images with the same resolution; and processing the characteristic images of the candidate regions to preliminarily determine the suspicious object type and the position information in the two-dimensional image of the human body.
In this scheme, in order to ensure the accuracy of detection, the initial detection result needs to be further rechecked. In one embodiment, the preliminarily determined confidence corresponding to the rectangular frame of the suspicious object label is compared with a preset identification condition, and if the identification condition is met, the suspicious object is determined to exist in the position of the rectangular frame.
This scheme further discloses a suspicious article detecting system of millimeter wave image, and this system includes: the device comprises millimeter wave scanning equipment, a mapping unit and an identification unit; millimeter wave scanning is carried out on a target human body through millimeter wave scanning equipment, and a millimeter wave scanning image of the target is obtained; processing the millimeter wave image of the target by using a mapping unit to obtain a multi-view target mapping standard image; the identification unit identifies the target mapping standard image based on a pre-constructed target detection network model, and determines the position and the category of the suspicious object. The identification unit can identify the target mapping standard image according to the target detection network model, preliminarily predict the position of the suspicious object in the image and the corresponding confidence value; and finally, determining whether the suspicious object exists in the image or not by comparing the preset threshold with the confidence value. In one embodiment, the identification unit comprises: the characteristic extraction module is used for extracting the characteristics of the target mapping standard image to obtain a characteristic image; a region generation module for generating a plurality of candidate regions according to the feature image; the interest area pooling layer is used for converting the candidate areas into characteristic images with the same resolution; the classification positioning module is used for processing the characteristic images of the candidate regions based on the full connection layers, labeling suspicious target categories and position information in the target human body image and labeling confidence degrees corresponding to the positions; the analysis module is used for comparing the confidence corresponding to the marking position with a preset threshold; if the confidence coefficient is greater than or equal to a preset threshold value, determining that suspicious articles exist at the marked position; and if the confidence coefficient is smaller than a preset threshold value, determining that no suspicious articles exist at the marked position.
The present solution is further illustrated by the following examples.
The embodiment provides a millimeter wave image suspicious object detection method, which combines millimeter wave human body two-dimensional images under multiple viewing angles and performs joint detection on the human body images under the multiple viewing angles based on a deep learning method, so that the detection capability of hidden suspicious objects is improved, and the human body security inspection accuracy is further improved. Compared with the prior art, the human body security check equipment and the human body security check method can improve the passing efficiency and the automation degree of the security check channel. The method for detecting the hidden articles in the human body can effectively improve the accuracy of an automatic suspicious article identification algorithm, obviously reduce the burden of security personnel and enhance the robustness of human body security equipment. The specific implementation process of the detection method is as follows:
step one, obtaining a two-dimensional imaging result of the millimeter wave cylindrical scanning system. As shown in fig. 2, a person stands in the system and keeps still, transmits broadband signals through the rotating radar scanning array, receives signals reflected by the human body, performs a series of signal processing operations, and can synthesize a human body three-dimensional imaging result. And projecting the three-dimensional imaging result to a two-dimensional plane at certain fixed angles to obtain a plurality of two-dimensional images. Due to the shielding effect, half of the images can obtain the front information of the human body, and the other half of the images can obtain the back information of the human body.
And secondly, according to the rotation geometric relationship between the images, taking each image of the front and the back of the human body as a standard graph, and performing coordinate conversion on pixels of all the images obtained in one scanning. Specifically, all the front images are mapped into the front standard graph, and all the back images are mapped into the back standard graph, so that a plurality of images under the same viewing angle of the front or the back are obtained. For example, if the scan angle of a certain front image differs by 30 ° from the scan angle of a front standard image, the pixel coordinates (x, y) in the front image are horizontally rotated by 30 °, and the coordinates (x ', y') after the rotation are obtained. The new coordinates (x', y) are the corresponding coordinates of the front image in the front standard image. In the same way, all the front/back side images in one scan are mapped into the front/back side standard images.
And step three, marking pictures of suspicious articles. Through a plurality of tests, a large number of human body two-dimensional images carrying various suspicious articles are obtained, and the positions and the types of the suspicious articles are marked in the front standard image or the back standard image by adopting a rectangular frame.
And step four, training a model. And inputting the two-dimensional image and the corresponding marking box into a convolutional neural network for training. Here, a typical network faster region convolutional neural network (fasterncnn) is used for training. As shown in fig. 3, the object detection network model may include four basic modules: a feature extraction network 41, a Region selection network (RPN) 42, a Region of interest Pooling layer 43(ROI Pooling), and a classification and location detection network 44.
Step 401, dividing all images in one scanning into two groups according to the front and the back, wherein each group of images to be trained consists of front/back standard images and all images mapped to the front/back standard images.
And step 402, inputting each group of images as a multichannel of the neural network, marking a rectangular box as a training label, and training the target detection network.
The target detection network model is composed of four sub-networks, wherein the feature extraction network 41 performs feature extraction on the image to be detected through a plurality of convolution layers of the feature extraction network to obtain a feature image (feature map) of the image to be detected. The area generation network 42 of the target detection network model generates a plurality of candidate areas from the feature image. The region of interest pooling layer 43 converts a plurality of candidate regions into feature images of the same resolution. The classification and location detection network 44 processes the feature images of the candidate regions through a plurality of Full Connectivity (FC) layers thereof, and determines suspicious object types and location information in the two-dimensional image of the human body.
And step five, the joint detection of the suspicious articles under multiple viewing angles can be realized based on the identification unit shown in fig. 4.
Step 501, mapping each image coordinate to a front/back standard graph according to a rotation geometric relationship between front/back images in a group of scans.
Step 502, inputting the mapped front/back image group into a trained FasterRCNN detection network, and automatically outputting the predicted suspicious object labeling rectangular frame position and confidence c by the network.
Step 503, comparing the weighted confidence c of each detected suspicious object with a preset threshold T, and if c ≧ T, determining that there is a suspicious object at the position, and outputting a detection frame corresponding to the suspicious object. If c < T, the suspicious item is not considered to exist at the position.
In conclusion, compared with the traditional millimeter wave image detection method, the millimeter wave imaging result under each angle of the human body can be used for detection, the detection range coverage is wider, and the suspicious object image hidden in each part of the human body can be obtained.
The scheme can carry out joint estimation on the image detection results under various visual angles, can effectively improve the detection accuracy and inhibit false alarms caused by noise interference under a certain angle;
the scheme is simple in calculation and high in data rate, and can meet the requirement of target detection on high data rate.
The scheme greatly reduces the workload of labeling.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The present invention is not limited to the above embodiments, and any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention are included in the scope of the claims of the present invention which are filed as the application.
Claims (10)
1. A millimeter wave image suspicious object detection method is characterized by comprising the following steps:
processing the millimeter wave image of the target to obtain a multi-view target mapping standard image;
and identifying the target mapping standard image based on a pre-constructed target detection network model, and determining the position and the category of the suspicious object.
2. The millimeter wave image detection method for suspicious objects according to claim 1, wherein the step of processing the millimeter wave image of the target to obtain the multi-view target mapping standard image comprises:
and acquiring a millimeter wave scanning image of the target.
3. The millimeter wave image suspicious object detection method according to claim 1 or 2, wherein the step of processing the millimeter wave image of the target to obtain a multi-view target mapping standard image comprises:
taking each image of the front and back of the target human body as a standard image;
and performing coordinate conversion on the millimeter wave images, and mapping image pixels corresponding to the front side and image pixels corresponding to the back side in the millimeter wave images of the target to the front side standard image and the back side standard image of the target human body respectively according to the rotation geometric relationship between the millimeter wave images to obtain the mapping standard images of the front side and the back side of the target human body.
4. The millimeter wave image suspicious item detection method according to claim 1, wherein the step of constructing the target detection network model comprises:
marking the position and the category of the suspicious object in the target mapping standard image;
based on the convolutional neural network, a front standard image and a back standard image of a target human body are used as input, the standards are used as training labels, and the target detection network is trained to obtain a target detection network model.
5. The millimeter wave image suspicious object detection method according to claim 1, wherein the step of identifying the target mapping standard image based on the pre-constructed target detection network model and determining the location and the category of the suspicious object comprises:
performing feature extraction on the target mapping standard image to obtain a feature image;
generating a plurality of candidate regions according to the characteristic image;
converting the candidate regions into feature images with the same resolution;
and processing the characteristic images of the candidate regions based on the full connection layers, labeling the position and the category of the suspicious object in the target human body image, and generating a confidence coefficient corresponding to the labeling position.
6. The millimeter wave image suspicious object detection method according to claim 5, wherein the step of identifying the target mapping standard image based on the pre-constructed target detection network model and determining the location and the category of the suspicious object further comprises:
comparing the confidence corresponding to the labeling position with a preset threshold;
if the confidence coefficient is greater than or equal to a preset threshold value, determining that suspicious articles exist at the marked position;
and if the confidence coefficient is smaller than a preset threshold value, determining that no suspicious articles exist at the marked position.
7. A millimeter wave image suspicious item detection system, the system comprising:
the mapping unit is used for processing the millimeter wave image of the target to obtain a multi-view target mapping standard image;
and the identification unit is used for identifying the target mapping standard image based on a pre-constructed target detection network model and determining the position and the category of the suspicious object.
8. The millimeter wave image suspicious object detection system according to claim 7, characterized in that the system further comprises: and the millimeter wave scanning device is used for performing millimeter wave scanning on the target human body to obtain a millimeter wave scanning image of the target.
9. The millimeter wave image suspicious item detection system according to claim 7, wherein said mapping unit specifically executes the following steps:
marking the position and the category of the suspicious object in the target mapping standard image;
based on the convolutional neural network, a front standard image and a back standard image of a target human body are used as input, the standards are used as training labels, and the target detection network is trained to obtain a target detection network model.
10. The millimeter wave image suspicious object detection system according to claim 7, characterized in that said identification unit comprises:
the characteristic extraction module is used for extracting the characteristics of the target mapping standard image to obtain a characteristic image;
a region generation module for generating a plurality of candidate regions according to the feature image;
the interest area pooling layer is used for converting the candidate areas into characteristic images with the same resolution;
the classification positioning module is used for processing the characteristic images of the candidate regions based on the full connection layers, labeling the category and position information of the suspicious object in the target human body image and labeling the confidence corresponding to the position;
the analysis module is used for comparing the confidence corresponding to the marking position with a preset threshold; if the confidence coefficient is greater than or equal to a preset threshold value, determining that suspicious articles exist at the marked position; and if the confidence coefficient is smaller than a preset threshold value, determining that no suspicious articles exist at the marked position.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911324885.XA CN111046877A (en) | 2019-12-20 | 2019-12-20 | Millimeter wave image suspicious article detection method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911324885.XA CN111046877A (en) | 2019-12-20 | 2019-12-20 | Millimeter wave image suspicious article detection method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111046877A true CN111046877A (en) | 2020-04-21 |
Family
ID=70237379
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911324885.XA Pending CN111046877A (en) | 2019-12-20 | 2019-12-20 | Millimeter wave image suspicious article detection method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111046877A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112597989A (en) * | 2020-12-18 | 2021-04-02 | 中国科学院上海微系统与信息技术研究所 | Millimeter wave three-dimensional holographic image concealed article detection method and system |
| CN112633057A (en) * | 2020-11-04 | 2021-04-09 | 北方工业大学 | Intelligent monitoring method for abnormal behaviors in bus |
| CN115311684A (en) * | 2022-08-05 | 2022-11-08 | 杭州电子科技大学 | Method for integrating millimeter wave image multi-angle detection results |
| CN116125466A (en) * | 2023-03-02 | 2023-05-16 | 武汉理工大学 | Ship personnel hidden threat object carrying detection method and device and electronic equipment |
| CN117111164A (en) * | 2023-10-17 | 2023-11-24 | 杭州海康威视数字技术股份有限公司 | Millimeter wave-based foreign matter detection method and device and electronic equipment |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090041293A1 (en) * | 2007-08-08 | 2009-02-12 | Brian Andrew | Multiple camera imaging method and system for detecting concealed objects |
| CN108182663A (en) * | 2017-12-26 | 2018-06-19 | 北京无线电计量测试研究所 | A kind of millimeter-wave image effect reinforcing method, equipment and readable storage medium storing program for executing |
| CN109086679A (en) * | 2018-07-10 | 2018-12-25 | 西安恒帆电子科技有限公司 | A kind of millimetre-wave radar safety check instrument foreign matter detecting method |
| CN110443158A (en) * | 2019-07-18 | 2019-11-12 | 华中科技大学 | A kind of human body concealment object detecting method based on millimeter-wave image |
-
2019
- 2019-12-20 CN CN201911324885.XA patent/CN111046877A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090041293A1 (en) * | 2007-08-08 | 2009-02-12 | Brian Andrew | Multiple camera imaging method and system for detecting concealed objects |
| CN108182663A (en) * | 2017-12-26 | 2018-06-19 | 北京无线电计量测试研究所 | A kind of millimeter-wave image effect reinforcing method, equipment and readable storage medium storing program for executing |
| CN109086679A (en) * | 2018-07-10 | 2018-12-25 | 西安恒帆电子科技有限公司 | A kind of millimetre-wave radar safety check instrument foreign matter detecting method |
| CN110443158A (en) * | 2019-07-18 | 2019-11-12 | 华中科技大学 | A kind of human body concealment object detecting method based on millimeter-wave image |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112633057A (en) * | 2020-11-04 | 2021-04-09 | 北方工业大学 | Intelligent monitoring method for abnormal behaviors in bus |
| CN112633057B (en) * | 2020-11-04 | 2024-01-30 | 北方工业大学 | Intelligent monitoring method for abnormal behavior in bus |
| CN112597989A (en) * | 2020-12-18 | 2021-04-02 | 中国科学院上海微系统与信息技术研究所 | Millimeter wave three-dimensional holographic image concealed article detection method and system |
| CN115311684A (en) * | 2022-08-05 | 2022-11-08 | 杭州电子科技大学 | Method for integrating millimeter wave image multi-angle detection results |
| CN115311684B (en) * | 2022-08-05 | 2023-03-28 | 杭州电子科技大学 | Method for integrating multi-angle detection results of millimeter wave images |
| CN116125466A (en) * | 2023-03-02 | 2023-05-16 | 武汉理工大学 | Ship personnel hidden threat object carrying detection method and device and electronic equipment |
| CN117111164A (en) * | 2023-10-17 | 2023-11-24 | 杭州海康威视数字技术股份有限公司 | Millimeter wave-based foreign matter detection method and device and electronic equipment |
| CN117111164B (en) * | 2023-10-17 | 2024-01-26 | 杭州海康威视数字技术股份有限公司 | Millimeter wave-based foreign matter detection method and device and electronic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111046877A (en) | Millimeter wave image suspicious article detection method and system | |
| CN106651752B (en) | Three-dimensional point cloud data registration method and splicing method | |
| Nissimov et al. | Obstacle detection in a greenhouse environment using the Kinect sensor | |
| WO2022012158A1 (en) | Target determination method and target determination device | |
| CN110826450A (en) | Automatic suspicious article detection method based on millimeter wave image | |
| CN107657644B (en) | Sparse scene flows detection method and device under a kind of mobile environment | |
| CN107527368B (en) | Three-dimensional space attitude positioning method and device based on two-dimensional code | |
| CN111781608A (en) | Moving target detection method and system based on FMCW laser radar | |
| CN113743385A (en) | Unmanned ship water surface target detection method and device and unmanned ship | |
| Yin et al. | Spherical coordinates based methods of ground extraction and objects segmentation using 3-D LiDAR sensor | |
| Hansard et al. | Automatic detection of calibration grids in time-of-flight images | |
| CN111913177A (en) | Method and device for detecting target object and storage medium | |
| Ye et al. | Exploiting depth camera for 3d spatial relationship interpretation | |
| Romero-Ramirez et al. | Tracking fiducial markers with discriminative correlation filters | |
| Farahnakian et al. | Deep convolutional neural network-based fusion of rgb and ir images in marine environment | |
| CN111260607B (en) | Automatic suspicious article detection method, terminal equipment, computer equipment and medium | |
| Christen et al. | Target marker: A visual marker for long distances and detection in realtime on mobile devices | |
| Jiao et al. | A novel and fast corner detection method for sar imagery | |
| CN108805218B (en) | Optical target association method based on deviation mapping clustering | |
| Yao et al. | Real-time multiple moving targets detection from airborne IR imagery by dynamic Gabor filter and dynamic Gaussian detector | |
| Yao et al. | A novel method for real-time multiple moving targets detection from moving IR camera | |
| CN112598738A (en) | Figure positioning method based on deep learning | |
| CN106097362A (en) | Automatically detection and the localization method of artificial circular labelling in a kind of x-ray image | |
| Zhang et al. | A glass detection method based on multi-sensor data fusion in simultaneous localization and mapping | |
| Lühr et al. | Feature detection with a constant FAR in sparse 3-D point cloud data |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200421 |
|
| RJ01 | Rejection of invention patent application after publication |