CN110133739B - X-ray security check equipment and automatic image recognizing method thereof - Google Patents
X-ray security check equipment and automatic image recognizing method thereof Download PDFInfo
- Publication number
- CN110133739B CN110133739B CN201910269269.2A CN201910269269A CN110133739B CN 110133739 B CN110133739 B CN 110133739B CN 201910269269 A CN201910269269 A CN 201910269269A CN 110133739 B CN110133739 B CN 110133739B
- Authority
- CN
- China
- Prior art keywords
- image
- ray
- contraband
- module
- frame
- 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.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 49
- 238000001514 detection method Methods 0.000 claims abstract description 85
- 238000012545 processing Methods 0.000 claims abstract description 28
- 238000013135 deep learning Methods 0.000 claims abstract description 13
- 238000013500 data storage Methods 0.000 claims abstract description 12
- 239000000463 material Substances 0.000 claims abstract description 12
- 238000000605 extraction Methods 0.000 claims description 23
- 238000012549 training Methods 0.000 claims description 20
- 238000013527 convolutional neural network Methods 0.000 claims description 14
- 238000000638 solvent extraction Methods 0.000 claims description 8
- 239000002390 adhesive tape Substances 0.000 claims description 7
- 238000013528 artificial neural network Methods 0.000 claims description 6
- 238000012937 correction Methods 0.000 claims description 6
- 230000000149 penetrating effect Effects 0.000 claims description 6
- 238000011160 research Methods 0.000 claims description 5
- RYGMFSIKBFXOCR-UHFFFAOYSA-N Copper Chemical compound [Cu] RYGMFSIKBFXOCR-UHFFFAOYSA-N 0.000 claims description 4
- 230000008859 change Effects 0.000 claims description 4
- 229910052802 copper Inorganic materials 0.000 claims description 4
- 239000010949 copper Substances 0.000 claims description 4
- 238000004458 analytical method Methods 0.000 claims description 3
- 230000000903 blocking effect Effects 0.000 claims description 3
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 230000006835 compression Effects 0.000 claims description 3
- 238000007906 compression Methods 0.000 claims description 3
- 230000004927 fusion Effects 0.000 claims description 3
- 238000007477 logistic regression Methods 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 3
- 238000003672 processing method Methods 0.000 claims description 3
- 238000013138 pruning Methods 0.000 claims description 3
- 238000004846 x-ray emission Methods 0.000 claims description 3
- 238000001914 filtration Methods 0.000 claims description 2
- 238000009432 framing Methods 0.000 claims description 2
- 238000009434 installation Methods 0.000 claims description 2
- 238000007689 inspection Methods 0.000 abstract description 30
- 238000005516 engineering process Methods 0.000 abstract description 4
- 230000005855 radiation Effects 0.000 abstract description 4
- 238000003384 imaging method Methods 0.000 abstract description 2
- 230000008569 process Effects 0.000 description 8
- 241001270131 Agaricus moelleri Species 0.000 description 5
- 230000001965 increasing effect Effects 0.000 description 5
- 230000000875 corresponding effect Effects 0.000 description 3
- 238000012856 packing Methods 0.000 description 3
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical compound [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 description 2
- 230000005540 biological transmission Effects 0.000 description 2
- 238000004040 coloring Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000002708 enhancing effect Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000010521 absorption reaction Methods 0.000 description 1
- 208000003464 asthenopia Diseases 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000010339 dilation Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 229910052742 iron Inorganic materials 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 238000004806 packaging method and process Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 230000001629 suppression Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V5/00—Prospecting or detecting by the use of ionising radiation, e.g. of natural or induced radioactivity
- G01V5/20—Detecting prohibited goods, e.g. weapons, explosives, hazardous substances, contraband or smuggled objects
- G01V5/22—Active interrogation, i.e. by irradiating objects or goods using external radiation sources, e.g. using gamma rays or cosmic rays
-
- 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
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Multimedia (AREA)
- High Energy & Nuclear Physics (AREA)
- General Life Sciences & Earth Sciences (AREA)
- Geophysics (AREA)
- Analysing Materials By The Use Of Radiation (AREA)
Abstract
The invention discloses X-ray security inspection equipment and an automatic image recognizing method thereof, and belongs to the field of radiation imaging. The automatic image recognition device comprises a basic supporting assembly, a data acquisition assembly and an automatic image recognition assembly. The data acquisition assembly comprises a plurality of groups of mutually independent X-ray detection systems and a detector module consisting of high-energy and low-energy detection plates; the automatic image recognizing assembly comprises an image processing module, a data storage module, a display control module and an automatic contraband recognizing module. By arranging a plurality of groups of independent and mutually independent detection systems, the problem of difficulty in reading due to object placing angles and object overlapping can be effectively avoided; the image processing module can calculate the material type of the goods according to the signal values received by the high-low energy detection plate; and detecting the X-ray image of the contraband in the detected object by adopting a target detection technology based on deep learning, and comparing the X-ray image with the learning data of the contraband to identify the contraband.
Description
Technical Field
The invention belongs to the field of radiation imaging, and particularly relates to X-ray security inspection equipment and an automatic image recognizing method thereof.
Background
The X-ray security inspection equipment adopts a line scanning working mode, and a high-sensitivity X-ray array detector in the X-ray security inspection equipment scans the articles layer by layer under the driving of a mechanical scanning device; after the transmission ray signal is detected and processed, the obtained data is subjected to image reconstruction to obtain an image, the internal information of the article can be represented on the image, and the image is displayed on a display screen. And the security check personnel checks the image displayed on the screen for observation and check to determine whether dangerous goods exist.
After the applicant researches the prior art carefully, the prior art is found to have at least the following defects and shortcomings: in the process of scanning an article through a security check device, 1, a security check worker is required to look at a screen all the time to perform manual resolution, and if the identification experience is insufficient, dangerous articles can pass through the security check smoothly, so that error detection is caused; 2. safety personnel are required to stare at the screen for a long time, so that visual fatigue is caused, the inspection of dangerous goods is easily missed, and the inspection is missed; 3. safety inspection personnel are required to inspect one by one, so that the working efficiency is low; 4. for uncertain goods, manual identification is required to be carried out in an image processing (such as enhancement, transformation and color change) mode, and the security check speed is reduced.
Disclosure of Invention
The purpose of the invention is as follows: the X-ray security inspection equipment and the automatic image recognizing method thereof are provided to solve the problems of error detection, missing detection and the like of contraband caused by easy deviation of manual observation of the conventional security inspection equipment; meanwhile, the security inspection efficiency is improved from the side surface.
The technical scheme is as follows: an X-ray security device comprising: the automatic image recognition device comprises a basic supporting assembly, a data acquisition assembly and an automatic image recognition assembly.
The basic supporting assembly comprises a frame with the cross section in a shape like a Chinese character 'kou', an outer cover plate fixedly arranged outside the frame, a tape machine inserted and installed inside the frame, a plurality of lifting supports and idler wheels fixedly arranged at the bottom of the frame, and lead curtain doors arranged at two ends of the frame;
the data acquisition assembly comprises a photoelectric sensor, an X-ray detection system and an industrial personal computer, wherein the photoelectric sensor is arranged on the adhesive tape machine and positioned on the inner side of the lead curtain door;
the automatic image recognizing assembly comprises an X-ray image receiving module electrically connected with an industrial personal computer, an image processing module connected with the X-ray image receiving module, an automatic contraband recognizing module connected with the image processing module, a data storage module connected with the automatic contraband recognizing module, a display control module connected with the automatic contraband recognizing module, and an alarm connected with the automatic contraband recognizing module.
In a further embodiment, the detection system comprises: the X-ray detector comprises a first X-ray machine mounting frame and a second X-ray machine mounting frame which are respectively arranged on one side and the top of the frame, a first X-ray generator which is fixedly arranged on the first X-ray machine mounting frame and keeps an included angle of 15-20 degrees with the horizontal direction, and a first detector module which is arranged on the other side and the top of the frame and has a cross section shape of '7' which is coplanar with a light source of the first X-ray generator; the second X-ray generator is fixedly installed on the second X-ray machine installation frame and keeps an included angle of 15-20 degrees with the vertical direction, the second detector modules are installed on one side and the bottom of the frame and are in an L shape with the cross section of the same plane as a light source of the second X-ray generator, the collimators are arranged at the light source of each group of X-ray generators, and the shielding boxes are connected to each group of X-ray generators and the detector modules to form a closed ring; the first X-ray machine mounting frame, the first X-ray generator, the first detector module, the second X-ray machine mounting frame, the second X-ray generator and the second detector module are arranged at a preset distance in the conveying direction of the adhesive tape machine to form at least two groups of mutually independent detection systems.
In a further embodiment, the detection plate is a high-energy and low-energy detection plate, the high-energy and low-energy X-ray detector units are arranged in an up-and-down overlapping mode, and a layer of filtering copper sheets is arranged between the high-energy and low-energy X-ray detector units.
In another aspect, an automatic image recognizing method for an X-ray security inspection apparatus includes:
s1, after the detected article enters the channel, blocking the photoelectric sensor, and sending a detection signal of the photoelectric sensor to the control unit to start X-ray emission;
s2, scanning the cargo layer with the thickness of about 1 mm by the X-ray fan-shaped beam passing through the collimator at the frequency of 3-10 ms each time, and penetrating through the module moving at a constant speed along with the conveyor belt;
s3, the detector module receives the X-ray signal penetrating through the scanning area and carries out photoelectric conversion, the data acquisition system receives, amplifies and digitizes the output signal of the detector and transmits the output signal to the industrial personal computer, and finally the output signal is sent to the automatic image recognition assembly;
s4, the image processing module performs background-air correction, gray level fusion, material classification, geometric correction and the like on the digital signal of each pixel, and then performs basic image processing functions of coloring, enhancing and the like according to the corrected 16-bit high-energy and low-energy signals;
s5, the display control module displays the image data on a color display to present a high-quality X-ray scanning image;
s6, simultaneously, the automatic contraband identification module identifies the contraband, performs sound-light alarm through the alarm, and frames the position information and the type information of the contraband on the display control module;
in a further embodiment, the automatic contraband identification module performs deep neural network training on the selected standard sample of the contraband by using a deep learning method in advance, and stores learning data in the data storage module.
In a further embodiment, the deep learning adds a clean parcel picture without contraband as a negative example for training; and the number of the training pictures is expanded by methods of image scaling, image rotation, image turning change and the like.
In a further embodiment, the image processing module automatically identifies forbidden articles and adopts a target detection algorithm based on deep learning to detect X-ray images of the forbidden articles in the detected articles; the method for automatically identifying the contraband by the automatic contraband identification module comprises the following steps:
s601, firstly, predicting the position of an extraction frame of an object in an X-ray image through an RPN network, and then mapping the extraction frame to an image feature table to intercept the feature corresponding to the detected object;
s602, performing further convolution operation on the features, comparing the features with the features of the learning data, predicting the type of the detected object, and further correcting the position of the extraction frame;
s603, after the position of the extraction frame is corrected, repeating the steps in the S6, comparing the feature similarity, judging whether the feature similarity is larger than a first threshold value, and if the feature similarity is larger than the first threshold value, determining that the object is forbidden;
s604, if the detected object is a prohibited object, the data storage module stores the image, the position information and the type information.
In a further embodiment, in the target detection algorithm, the position of the extraction frame of the object is predicted by setting the size and the aspect ratio of the candidate extraction frame and performing logistic regression based on the candidate extraction frame.
In a further embodiment, when the target detection algorithm detects multi-scale contraband of the X-ray image, a research method of a multi-receptive-field branch model is adopted; the detection method of the multi-receptive-field branch model comprises the following steps: taking an analysis residual error network as a frame, forming 3 branches sharing parameters of the network for the convolution layer of the network, namely a first sparse convolution branch, a second sparse convolution branch and a third sparse convolution branch; each branch adopts RPN and R-CNN network, and RPN and R-CNN network parameters are shared, then the detection results of the three branches are fused through NMS, and the final detection result of the model is obtained.
In a further embodiment, the deep convolutional neural network adopted by the deep learning method needs to compress and optimize the model, and the pruning compression method is adopted to reduce the number of filters in the convolutional network.
In a further embodiment, the method for the automatic contraband identification module to automatically identify the X-ray image of the suspicious region of the contraband is designed as follows:
a1, acquiring suspicious region image data, namely intercepting an X-ray image of which the identification degree is smaller than a first threshold value but larger than a second threshold value in an automatic image identification method of an image contraband automatic identification module;
a2, partitioning the suspicious region image by adopting a partitioning algorithm to obtain a plurality of block images, and numbering each block image;
a3, selecting partial block images for processing, acquiring characteristic parameters of the partial block images, and comparing the characteristic parameters in a database;
a4, when the similarity between a block image and the articles in the database is higher than the expected value, selecting the adjacent block image for comparison;
a5, if the image similarity of the adjacent blocks is higher than the expected value, merging the block image and the adjacent images into the same block image; continuing the steps A4-A5 until no tile image with the neighboring side of the finally synthesized regional image data higher than the expected similarity exists;
a6, comparing the image data obtained finally after combination with the articles in the database, and judging whether the articles are forbidden articles; and stopping detection when the similarity between the block images with the database items is lower than an expected value, wherein the similarity exceeds the expected number.
In this regard, the method for the automatic contraband identification module to automatically identify the X-ray image of the contraband on the suspicious region may further be designed as follows:
b1, acquiring suspicious region image data, namely intercepting an X-ray image of which the similarity is smaller than a first threshold value but larger than a second threshold value in the automatic image recognizing method of the image contraband 21;
b2, partitioning the suspicious region image by adopting a splitting algorithm to obtain a plurality of block images, and numbering each block image;
b3, selecting partial block images for processing, uniformly selecting a plurality of pixel points in the block images, acquiring the gray values of the pixel points, comparing the gray values of two adjacent pixel points, judging whether the gray values are larger than an expected value, and identifying and picking up images which are inconsistent with the gray values at two sides;
b4, when some block image appears with the image whose two sides are not consistent with the gray scale, selecting the adjacent block image to repeat B3;
b5, if the adjacent block image has an image with inconsistent gray scales at two sides, combining the block image and the image with inconsistent gray scales at two sides in the adjacent image into a new image, and communicating image areas; continuing to carry out the steps B4-B5 until no image with the gray difference value of two adjacent pixel points larger than the expected value exists on the adjacent side of the finally synthesized regional image data;
b6, comparing the combined finally picked image data with the outline of the article in the database, and judging whether the article is a contraband article; and stopping detection when the similarity between the block images with the database items is lower than an expected value, wherein the similarity exceeds the expected number.
Has the advantages that: the invention relates to an X-ray security inspection device and an automatic image recognizing method thereof.A plurality of groups of independent and mutually independent detection systems are arranged on the aspect of the device, so that the difficulty in image reading caused by object placing angles and object overlapping can be effectively avoided, and the conditions of missing inspection and manual unpacking inspection of forbidden articles are effectively reduced; by adopting the high-low energy detection plate and according to the signal value received by the high-low energy detection plate, the image processing module can calculate the material type of the goods, and the material composition of the goods to be detected can be conveniently identified. In the aspect of the method, an X-ray image of the contraband in the detected object is detected by adopting a target detection technology based on deep learning, and compared with the standard of the contraband by using a convolutional neural network to identify the contraband; and the multi-scale contraband detection identification precision is improved by adopting a multi-receptive-field branch model research method. The problems of error detection, missing detection and the like of contraband due to easy deviation of manual observation of the conventional security inspection equipment are solved; meanwhile, the security inspection efficiency is improved from the side surface.
Drawings
Fig. 1 is a schematic structural view of the present invention.
Fig. 2 is a schematic cross-sectional view of the present invention.
Fig. 3 is a schematic diagram of the connection of the security device of the present invention.
FIG. 4 is a schematic diagram of the multi-sensor branch model of the present invention.
Fig. 5 is an X-ray image of a pressure tank in embodiment 2 of the present invention.
The reference signs are: the device comprises a frame 1, an outer cover plate 2, a belt conveyor 3, a lifting support 4, a roller 5, a lead curtain door 6, a photoelectric sensor 7, a first X-ray machine mounting frame 8, a second X-ray machine mounting frame 9, a first X-ray generator 10, a first detector module 11, a second X-ray generator 12, a second detector module 13, a collimator 14, a shielding box 15, a detection plate 16, an industrial personal computer 17, an image receiving module 18, an image processing module 19, a data storage module 20, an automatic contraband identification module 21, a display control module 22, an alarm 23, a first sparse convolution branch 24, a second sparse convolution branch 25 and a third sparse convolution branch 26.
Detailed Description
In the following description, numerous specific details are set forth in order to provide a more thorough understanding of the present invention. It will be apparent, however, to one skilled in the art, that the present invention may be practiced without one or more of these specific details. In other instances, well-known features have not been described in order to avoid obscuring the invention.
In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings, and are only for convenience of description and simplicity of description, but do not indicate or imply that the device or element being referred to must have a particular orientation, be constructed and operated in a particular orientation, and thus, should not be construed as limiting the present invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.
Example 1
As shown in fig. 1 to 3, an X-ray security inspection apparatus (hereinafter referred to as "the security inspection apparatus") includes: the automatic image recognition device comprises a basic supporting assembly, a data acquisition assembly and an automatic image recognition assembly.
A base support assembly comprising: the device comprises a frame 1, an outer cover plate 2, a tape machine 3, a lifting support 4, a roller 5 and a lead curtain door 6. The section of the frame 1 is in a shape of a Chinese character 'kou', the outer cover plate 2 is fixedly installed outside the frame 1, the adhesive tape machine 3 is installed inside the frame 1 in an inserted mode, the lifting supports 4 and the rollers 5 are fixedly installed at the bottom of the frame 1, and the lead curtain doors 6 are arranged at two ends of the frame 1; the distance between the frame 1 and the supporting plane can be adjusted up and down by adjusting the lifting support 4 through a tool such as a wrench, and the lifting support plays a role in lifting. When the lifting support 4 is lower than the height of the roller 5, the roller 5 can be used for moving the security inspection equipment. Meanwhile, the lead curtain door 6 and the outer shield plate 2 serve to reduce the radiation of X-rays.
A data acquisition assembly comprising: the device comprises a photoelectric sensor 7, an X-ray detection system and an industrial personal computer 17, wherein the photoelectric sensor 7 is arranged on the adhesive tape machine 3 and is positioned on the inner side of the lead curtain door 6, the X-ray detection system is arranged inside the frame 1, the industrial personal computer 17 is arranged on one side of the frame 1, and the industrial personal computer 17 is electrically connected with the X-ray detection system and the photoelectric sensor 7. Wherein the detection system comprises: the X-ray detector comprises a first X-ray machine mounting frame 8, a second X-ray machine mounting frame 9, a first X-ray generator 10, a first detector module 11, a second X-ray generator 12, a second detector module 13, a collimator 14, a shielding box 15, a detection plate 16 and an industrial personal computer 17. A first X-ray machine mounting frame 8 and a second X-ray machine mounting frame 9 are respectively arranged on one side and the top of the frame 1, the first X-ray generator 8 is fixedly mounted on the first X-ray machine mounting frame 8, an included angle between the first X-ray machine mounting frame 8 and the horizontal direction is 15-20 degrees, the first detector module 11 is mounted on the other side and the top of the frame 1, and the cross section of the first detector module and the cross section of the light source of the first X-ray generator 10 on the same plane are in a 7 shape; second X-ray generator 12 fixed mounting is in on the second X-ray machine mounting bracket 9, and keep the contained angle to be 15 ~ 20 with vertical direction, second detector module 13 is installed one side and the bottom of frame 1, and with the coplanar cross sectional shape of the light source of second X-ray generator 12 is "L", and collimator 14 sets up in every group X-ray generator light source department, and shielding box 15 connects at every group X-ray generator and detector module, constitutes closed ring for reduce the radiation of X ray and reveal. The first X-ray machine mounting frame 8, the first X-ray generator 10, the first detector module 11, the second X-ray machine mounting frame 9, the second X-ray generator 12 and the second detector module 13 are separated by a predetermined distance along the transportation direction of the adhesive tape machine 3 to form two groups of mutually independent detection systems. The security inspection equipment adopts a plurality of groups of independent detection systems which are mutually independent, so that the difficulty in reading due to the arrangement angle of objects and the overlapping of the objects can be effectively avoided, and the conditions of missing inspection of forbidden articles and manual unpacking inspection can be effectively reduced; and each group of detector modules are arranged in an L shape and are arranged at the positions opposite to the X-ray generator, and the X-ray emitted by the X-ray generator can scan the cross section of the whole inspection channel, so that the probability of missed inspection is further reduced.
Preferably, the detection board 16 is a high-energy and low-energy detection board, and is formed by stacking detector units for high-energy and low-energy X-rays up and down, and a layer of filter copper sheet is arranged between the detector units for high-energy and low-energy X-rays. Keeping the power of the X-ray generator at a constant value, and sensing X-rays and receiving low-energy components in the medium X-rays because the low-energy X-ray detector unit is arranged at one end close to a light source away from the X-ray generator; the low-energy components in the X-rays are filtered by the filter copper sheet, the high-energy X-ray detector unit receives the high-energy components in the X-rays, the image processing module 19 can calculate the material type of the goods according to the signal values received by the high-energy and low-energy detection board 16, different materials are respectively marked with different colors on the display screen, and the inspectors can conveniently identify the material composition of the goods to be inspected. Compared with the traditional dual-energy X-ray generator, the space layout of the whole security inspection equipment is simpler, and the economic cost is more saved.
Automatic picture subassembly of knowing includes: an image receiving module 18, an image processing module 19, a data storage module 20, a display control module 22 and an alarm 23. The image receiving module 18 is electrically connected with the industrial personal computer 17 and is used for receiving the X-ray image; the image processing module 19 is connected with the image receiving module 18 and is used for the X-ray image processing module 19; the automatic contraband identification module 21 is connected with the image processing module 19 and is used for identifying and detecting contraband; the data storage module 20 is connected with the contraband automatic identification module 21 and used for storing an original image, a deep learning example and an automatic image identification result, the auxiliary image processing module 19 is used for identifying images, and the display control module 22 is connected with the contraband automatic identification module 21 and used for outputting images; the alarm 23 is connected to the contraband automatic identification module 21, and includes an LED lamp and a buzzer, and when the contraband passes through, the alarm 23 gives out a photoacoustic alarm.
Example 2
On the other hand, further explaining an automatic image recognizing method for the security inspection equipment, specifically:
and S1, after the detected article enters the channel, blocking the photoelectric sensor, and sending a detection signal of the photoelectric sensor to the control unit to start X-ray emission.
S2, scanning the cargo layer with the thickness of about 1 mm by the X-ray fan-shaped beam passing through the collimator 14 at the frequency of 3-10 ms each time, and penetrating the cargo moving with the conveyor belt at a constant speed to reach the detector module.
S3, the detector module receives the X-ray signal penetrating through the scanning area and carries out photoelectric conversion, the data acquisition system receives, amplifies and digitizes the output signal of the detector and transmits the data to the industrial personal computer 17, and finally the output signal is sent to the automatic image recognition assembly;
s4, the image processing module 19 carries out background-air correction, gray level fusion and material classification, geometric correction and the like on the digital signal of each pixel, and then carries out basic image processing functions of coloring, enhancing and the like according to the corrected 16-bit high-energy and low-energy signals.
S5, the display control module 22 displays the X-ray image data on a color display to show a high-quality X-ray scanning image.
And S6, identifying the contraband by the automatic contraband identification module 21, giving an audible and visual alarm by the alarm 23, and framing the position information and the type information of the contraband on the display control module 22.
The collimated X-ray fan beam scans a cargo layer with a thickness of about 1 mm at a frequency of 3-10 ms each time, a signal of each layer is converted into a column of images on the display control module 22, and the gray level value of the images represents the degree of absorption of the X-ray by the object. The goods to be detected are scanned one by one, the X-ray images of the goods obtained after the transmission data of the layers are processed are also displayed on the screen in a row, and after all the layers of the goods to be detected are combined, the complete X-ray images of the goods to be detected are formed.
The automatic contraband identification module 21 needs to perform deep neural network training on the selected standard sample of the contraband in advance by using a deep learning method, and store the data in the data storage module 20.
As a preferred scheme, a clean parcel picture without contraband is added in the deep neural network training as a negative example for training. In a traditional deep neural network training model, only pictures marked with positive examples are input for training. However, in the actual implementation process and the experimental comparison process, the misrecognition rate of the training model only using the normal example picture is too high. Because the model only learns the features of the positive example object during the training process, and other negative example objects may have the same features, the model may misidentify the negative example object as a positive example. As shown in fig. 5, if the blue and circular features of the pressure tank are learned in the training process, since there is no other circular or blue object other than the pressure tank as training, the difference between the features other than blue and circular cannot be learned, and since the circular and blue objects are easily recognized as the pressure tank during detection, the misrecognition rate is high. By introducing the characteristics of the negative typical object, the false recognition rate can be greatly reduced, and the accuracy and the security check efficiency of automatic image recognition are improved. Meanwhile, the number of the training pictures is expanded through methods such as image scaling, image rotation, image turning change and the like; besides increasing the number of images to be trained, the adaptability of the model to objects with different dimensions can be increased.
After the X-ray image is sent to the automatic image recognition component, the automatic contraband recognition module 21 automatically recognizes the contraband and detects the X-ray image of the contraband in the detected object by using a target detection technology based on deep learning. The automatic contraband identification module 21 automatic image identification method specifically comprises the following steps:
s601, firstly, predicting the position of an extraction frame of an object in an X-ray image through an RPN (Region projection Network), and then mapping the extraction frame to an image feature table to intercept the feature corresponding to the detected object.
S602, then, by performing further convolution operation on the features and comparing the features with the features of the learning data, the reference image processing module 19 calculates the material type of the cargo according to the signal values received by the high/low energy detection board 16, so as to predict the type of the detected object, and further correct the position of the extraction frame.
And S603, after the position of the extraction frame is corrected, repeating the steps from S601 to S602, comparing the feature similarity, judging whether the feature similarity is greater than a first threshold value, and if the feature similarity is greater than the first threshold value, determining that the object is forbidden.
S604, if the detected object is a contraband, the data storage module 20 stores the image, the position information and the type information, namely, the found contraband image can be inquired, the original image can be traced, and the standard contrast neural network data can be enriched.
The target detection method is two-stage target detection (two-stage target detection), and the Fast-RCNN algorithm is adopted in the embodiment, but not limited to the Fast-RCNN algorithm, and can also be algorithms such as R-CNN, SPPNet, Fast R-CNN, FPN and the like. the detection problem is divided into two stages by two-stage target detection, wherein a candidate region is firstly generated in the first stage and contains approximate position information of the target, and then the candidate region is classified and position refined in the second stage. Firstly, one-stage target detection is only an RPN network with multiple classifications in terms of network structure, and corresponds to the first stage of the two-stage target detection method. The prediction result is obtained by predicting from the corresponding feature in the image feature table, and the two-stage target detection method further refines the extracted candidate frame according to the result, so that the method is more accurate. For example, a candidate box may cover only 50% of an object but is considered to be a complete positive sample, and thus its prediction must be in error. Secondly, the one-stage algorithm has a poor detection effect on small targets, and if all candidate marquees do not cover the target, the target is missed. If a larger extraction candidate box covers the target, the larger extraction candidate box weakens the real features of the target, and the degree of reality is not high. the candidate box in the two-stage algorithm can amplify the target, the characteristics of the small target are amplified, the characteristic outline is clearer, and therefore the detection is more accurate.
As a preferred scheme, the automatic contraband identification module 21 optimizes a model and improves the detection accuracy according to the distribution of contraband samples in the X-ray image; the specific method comprises the following steps: in the target detection algorithm, the size and the aspect ratio of the extraction candidate frame are set, and logistic regression is performed on the basis of the extraction candidate frame, so that the position of the extraction frame of the object is predicted. The running speed of the automatic image recognizing module is increased, and the security inspection speed is increased.
Because the detection effects of the receptive fields with different sizes on the objects with different scales are different, the detection performance of the objects with different scales is positively correlated with the sizes of the receptive fields. Therefore, in the target detection algorithm of the X-ray image, a research method of a multi-receptive-field branch model is adopted in the detection of the multi-scale contraband. As shown in fig. 4, namely: the analysis residual network is used as a frame 1, three branches are formed on convolution layers of the network, all the three branches share parameters of the network, and only ordinary convolution in each branch is replaced by expansion convolution respectively, namely a first sparse convolution branch 24, a second sparse convolution branch 25 and a third sparse convolution branch 26. Thus the receptive field will be different for each branch, but the network structure and parameters will be identical. And an RPN and an R-CNN network (Regions with CNN features, a target detection technology based on algorithms such as a convolutional neural network, a linear regression and a support vector machine) are adopted after each branch, RPN and R-CNN network parameters are shared, and then the detection results of the three branches are fused through NMS (Non-Maximum Suppression) to obtain the final detection result of the model.
Branches with different reception fields are constructed through different sparse convolutions of the dilation convolution, the size of each branch reception field is different and used for solving the problem of multi-scale object detection, and the network model parameters of each branch are consistent. The shared network parameters of different branches have the following advantages that firstly, compared with the original model, the parameters are not increased, and overfitting caused by excessive parameters is prevented. And secondly, the same feature expression capability is realized for objects with different dimensions, because the parameters of different branches are consistent. And thirdly, because the network parameters of the three branches are shared, objects with various scales are utilized in the network training process, and the shared network parameters are updated through back propagation. In the FPN training process, an image feature table of a certain layer is only trained by using an object with a certain size in a certain interval, and there are few object samples in the certain size interval, which may result in insufficient training of the certain layer. This problem is avoided by sharing network parameters among different branches.
In order to avoid the condition that the receptive field and the object dimension are not matched, each branch only trains samples with the dimension within a certain range, and the influence of the object with the extreme dimension on the test performance is avoided. The intervals of the areas of the training samples of the three sparse convolution branches are [0, 90], [90, 180], [180, ∞ ] respectively, and better area interval division can be determined through comparative experiments. Although each branch is trained by using samples with different scales, because the parameters of the branches are shared, the model fully utilizes the samples with various sizes, so that the target detection problem of a multi-scale object can be better solved.
As a preferable scheme, in order to solve the problem that a part of people disguise contraband articles to evade the detection tool and illegally bring the articles into a safe area, the following preferable scheme is provided. In the related inspection, some people can carry out detachable split design on forbidden articles or can locally disguise the forbidden articles. For example, the shank of the tool is removed, the shape of the tool is designed, and iron sheets and cylindrical objects are displayed on a security inspection machine, so that detection can be avoided, and great inconvenience is brought to work. Still other personnel disguise the contraband, for example through a packing unit with the same shape, the same material with the contraband, with the gomphosis of contraband in packing unit, the gap between contraband and the packing unit is less, is difficult to distinguish it. For this purpose, the image processing method is designed as follows:
a1, acquiring image data of a suspicious region, namely intercepting an X-ray image of which the identification degree is smaller than a first threshold value but larger than a second threshold value in the automatic image identification method of the image contraband 21;
a2, partitioning the suspicious region image by adopting a partitioning algorithm to obtain a plurality of block images, and numbering each block image;
a3, selecting partial block images for processing, acquiring characteristic parameters of the partial block images, and comparing the characteristic parameters in a database;
a4, when the similarity between a block image and the articles in the database is higher than the expected value, selecting the adjacent block image for comparison;
a5, if the image similarity of the adjacent blocks is higher than the expected value, merging the block image and the adjacent images into the same block image; continuing the steps A4-A5 until no tile image with the neighboring side of the finally synthesized regional image data higher than the expected similarity exists;
a6, comparing the image data obtained finally after combination with the articles in the database, and judging whether the articles are forbidden articles; and stopping detection when the similarity between the block images with the database items is lower than an expected value, wherein the similarity exceeds the expected number.
By the method, the articles with camouflage in partial areas can be detected, in this case, the area images, namely the whole data are compared and analyzed, the similarity of the area images and the whole data is possibly lower than an expected value, and the condition of missing detection occurs. Through local communication, a block image is formed, and partial areas are detected, so that suspicious parts can be detected.
Some people disguise the forbidden articles in a jogged mode, so that the shapes of the forbidden articles are different from those of the forbidden articles in a normal state, and detection is missed. To solve this problem. By means of material detection or image outer edge contour detection, when an X-ray penetrates through a chimeric part of a forbidden article and a chimeric packaging unit, a certain gap is left between the chimeric parts, so that a gap inconsistent with the gray scales of two sides can appear in the X-ray image, and therefore, the image processing method can be further designed as follows:
b1, acquiring suspicious region image data, namely intercepting an X-ray image of which the similarity is smaller than a first threshold value but larger than a second threshold value in the automatic image recognizing method of the image contraband 21;
b2, partitioning the suspicious region image by adopting a splitting algorithm to obtain a plurality of block images, and numbering each block image;
b3, selecting partial block images for processing, uniformly selecting a plurality of pixel points in the block images, acquiring the gray values of the pixel points, comparing the gray values of two adjacent pixel points, judging whether the gray values are larger than an expected value, and identifying and picking up images which are inconsistent with the gray values at two sides;
b4, when some block image appears with the image whose two sides are not consistent with the gray scale, selecting the adjacent block image to repeat B3;
b5, if the adjacent block image has an image with inconsistent gray scales at two sides, combining the block image and the image with inconsistent gray scales at two sides in the adjacent image into a new image, and communicating image areas; continuing to carry out the steps B4-B5 until no image with the gray difference value of two adjacent pixel points larger than the expected value exists on the adjacent side of the finally synthesized regional image data;
b6, comparing the combined finally picked image data with the outline of the article in the database, and judging whether the article is a contraband article; and stopping detection when the similarity between the block images with the database items is lower than an expected value, wherein the similarity exceeds the expected number.
As a preferred scheme, the deep convolutional neural network adopted by the deep learning method has intrinsic computational complexity, and in a specific implementation process, a model needs to be compressed and optimized, so that the frame rate is improved without sacrificing too much accuracy. The method of pruning compression is used for reducing the number of filters in a convolution network and reducing the amount of calculation and parameters so as to improve the running speed of an automatic image recognition module and improve the security inspection speed.
It should be noted that the various features described in the above embodiments may be combined in any suitable manner without departing from the scope of the invention. The invention is not described in detail in order to avoid unnecessary repetition.
Claims (1)
1. An automatic image recognizing method of an X-ray security check device is characterized in that the security check device comprises:
the basic supporting assembly comprises a frame with the cross section in a shape like a Chinese character 'kou', an outer cover plate fixedly arranged outside the frame, a tape machine inserted and installed inside the frame, a plurality of lifting supports and idler wheels fixedly arranged at the bottom of the frame, and lead curtain doors arranged at two ends of the frame;
the data acquisition assembly comprises a photoelectric sensor, an X-ray detection system and an industrial personal computer, wherein the photoelectric sensor is arranged on the adhesive tape machine and positioned on the inner side of the lead curtain door;
the automatic image recognizing assembly comprises an X-ray image receiving module electrically connected with an industrial personal computer, an image processing module connected with the X-ray image receiving module, an automatic contraband recognizing module connected with the image processing module, a data storage module connected with the automatic contraband recognizing module, a display control module connected with the automatic contraband recognizing module and an alarm connected with the automatic contraband recognizing module;
the detection system comprises: the X-ray detector comprises a first X-ray machine mounting frame and a second X-ray machine mounting frame which are respectively arranged on one side and the top of the frame, a first X-ray generator which is fixedly arranged on the first X-ray machine mounting frame and keeps an included angle of 15-20 degrees with the horizontal direction, and a first detector module which is arranged on the other side and the top of the frame and has a cross section shape of '7' which is coplanar with a light source of the first X-ray generator; the second X-ray generator is fixedly installed on the second X-ray machine installation frame and keeps an included angle of 15-20 degrees with the vertical direction, the second detector modules are installed on one side and the bottom of the frame and are in an L shape with the cross section of the same plane as a light source of the second X-ray generator, the collimators are arranged at the light source of each group of X-ray generators, and the shielding boxes are connected to each group of X-ray generators and the detector modules to form a closed ring; the first X-ray machine mounting frame, the first X-ray generator, the first detector module, the second X-ray machine mounting frame, the second X-ray generator and the second detector module are separated by a preset distance along the transportation direction of the adhesive tape machine to form at least two groups of mutually independent detection systems;
the detection plate is a high-low energy detection plate, detector units of high-energy X-rays and low-energy X-rays are arranged in an up-down overlapping mode, a layer of filtering copper sheet is arranged between the high-energy X-ray detector units and the low-energy X-ray detector units, and the power of the X-ray generator is a fixed value;
the automatic graph recognizing method comprises the following steps:
s1, after the detected article enters the channel, blocking the photoelectric sensor, and sending a detection signal of the photoelectric sensor to the control unit to start X-ray emission;
s2, scanning the cargo layer with the thickness of about 1 mm by the X-ray fan-shaped beam passing through the collimator at the frequency of 3-10 ms each time, and penetrating through the module moving at a constant speed along with the conveyor belt;
s3, the detector module receives the X-ray signal penetrating through the scanning area and carries out photoelectric conversion, the data acquisition system receives, amplifies and digitizes the output signal of the detector and transmits the output signal to the industrial personal computer, and finally the output signal is sent to the automatic image recognition assembly;
s4, the image processing module performs background-air correction, gray level fusion, material classification and geometric correction on the digital signal of each pixel, and then performs color assignment and basic image processing enhancement according to the corrected 16-bit high-energy and low-energy signals;
s5, the display control module displays the image data on a color display to present a high-quality X-ray scanning image;
s6, at the same time, predicting the position of an extraction frame of the object in the X-ray image through an RPN network, and mapping the extraction frame to an image feature table to intercept the feature corresponding to the detected object; then, performing further convolution operation on the features, comparing the features with the features of the learning data, predicting the type of the detected object, and further correcting the position of the extraction frame; after the position of the extraction frame is corrected, repeating the steps in the step S6, comparing the feature similarity, judging whether the feature similarity is larger than a first threshold value, and if the feature similarity is larger than the first threshold value, determining that the object is forbidden; if the detected object is a contraband object, storing the image, the position information and the type information in the data storage module, carrying out sound-light alarm through an alarm, and framing the position information and the type information of the contraband object on the display control module;
in addition, for the highly suspected contraband, the following image processing method is further carried out:
a1, acquiring image data of a suspicious region, namely intercepting an X-ray image of which the similarity is smaller than a first threshold value but larger than a second threshold value in an automatic image recognizing method of an image contraband automatic recognition module;
a2, partitioning the suspicious region image by adopting a partitioning algorithm to obtain a plurality of block images, and numbering each block image;
a3, selecting partial block images for processing, acquiring characteristic parameters of the partial block images, and comparing the characteristic parameters in a database;
a4, when the similarity between a block image and the articles in the database is higher than the expected value, selecting the adjacent block image for comparison;
a5, if the image similarity of the adjacent blocks is higher than the expected value, merging the block image and the adjacent images into the same block image; continuing the steps A4-A5 until no tile image with the neighboring side of the finally synthesized regional image data higher than the expected similarity exists;
a6, comparing the image data obtained finally after combination with the articles in the database, and judging whether the articles are forbidden articles; when the similarity between the block images with the number exceeding the expected number and the database articles is lower than an expected value, stopping detection;
the automatic contraband identification module performs deep neural network training on the selected standard sample of the contraband in advance by using a deep learning method, and stores learning data in the data storage module;
adding a clean parcel picture without contraband as a negative example for the deep learning, and training; the number of the training pictures is expanded through an image scaling method, an image rotation method and an image turning change method;
in the target detection algorithm of S6, the position of the extraction frame of the object is predicted by setting the size and aspect ratio of the candidate extraction frame and performing logistic regression based on the candidate extraction frame;
when the target detection algorithm detects multi-scale contraband of the X-ray image, a research method of a multi-receptive-field branch model is adopted; the detection method of the multi-receptive-field branch model comprises the following steps: taking an analysis residual error network as a frame, forming 3 branches sharing parameters of the network for the convolution layer of the network, namely a first sparse convolution branch, a second sparse convolution branch and a third sparse convolution branch; each branch adopts an RPN and an R-CNN network, and the RPN and R-CNN network parameters are shared, and then the detection results of the three branches are fused through NMS to obtain the final detection result of the model;
the deep convolutional neural network adopted by the deep learning method needs to compress and optimize the model, and the pruning compression method is adopted to reduce the number of filters in the convolutional network.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910269269.2A CN110133739B (en) | 2019-04-04 | 2019-04-04 | X-ray security check equipment and automatic image recognizing method thereof |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910269269.2A CN110133739B (en) | 2019-04-04 | 2019-04-04 | X-ray security check equipment and automatic image recognizing method thereof |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110133739A CN110133739A (en) | 2019-08-16 |
| CN110133739B true CN110133739B (en) | 2021-07-27 |
Family
ID=67569256
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910269269.2A Active CN110133739B (en) | 2019-04-04 | 2019-04-04 | X-ray security check equipment and automatic image recognizing method thereof |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110133739B (en) |
Families Citing this family (23)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110458166A (en) * | 2019-08-19 | 2019-11-15 | 广东工业大学 | A kind of hazardous material detection method, device and equipment based on deformable convolution |
| CN110531435B (en) * | 2019-08-29 | 2024-05-31 | 公安部第一研究所 | Test box and test method for testing contraband detection algorithm |
| CN110909604B (en) * | 2019-10-23 | 2024-04-19 | 深圳市重投华讯太赫兹科技有限公司 | Security check image detection method, terminal equipment and computer storage medium |
| WO2021087962A1 (en) * | 2019-11-08 | 2021-05-14 | 中国科学院深圳先进技术研究院 | Automatic identification device and method for restricted articles |
| CN112861888A (en) * | 2019-11-08 | 2021-05-28 | 中国科学院深圳先进技术研究院 | Automatic identification device and method for limiter |
| CN111123375A (en) * | 2019-12-25 | 2020-05-08 | 重庆特斯联智慧科技股份有限公司 | Security inspection platform and method |
| CN110956225B (en) * | 2020-02-25 | 2020-05-29 | 浙江啄云智能科技有限公司 | Contraband detection method and system, computing device and storage medium |
| CN111429410B (en) * | 2020-03-13 | 2023-09-01 | 杭州电子科技大学 | Object X-ray image material discrimination system and method based on deep learning |
| CN111337989A (en) * | 2020-03-20 | 2020-06-26 | 安徽启新明智科技有限公司 | Active double-visual-angle screening method and device for multistage security inspection instrument |
| JP7422023B2 (en) * | 2020-07-03 | 2024-01-25 | 株式会社日立ソリューションズ | X-ray image processing device and X-ray image processing method |
| CN112073009B (en) * | 2020-08-06 | 2024-09-20 | 重庆易斯特半导体研究院有限公司 | Analog signal acquisition processing system and security check machine signal processing system |
| CN112034523B (en) * | 2020-08-17 | 2022-07-19 | 长沙理工大学 | Hazardous chemical substance detection method and detection system |
| CN112162324B (en) * | 2020-09-02 | 2024-08-02 | 海深智能科技(上海)有限公司 | Intelligent security inspection method for effectively improving contraband identification rate |
| CN112415619B (en) * | 2020-09-14 | 2024-03-19 | 上海奕瑞光电子科技股份有限公司 | High-speed big data transmission system, method, terminal and medium based on linear array detector |
| CN112884755B (en) * | 2021-03-11 | 2023-07-04 | 北京理工大学 | Method and device for detecting contraband |
| CN113220927B (en) * | 2021-05-08 | 2024-08-06 | 百度在线网络技术（北京）有限公司 | Image detection method, device, equipment and storage medium |
| CN113281814A (en) * | 2021-06-23 | 2021-08-20 | 重庆世纪科怡科技股份有限公司 | Security inspection device based on image recognition and security inspection method thereof |
| CN113552118A (en) * | 2021-07-15 | 2021-10-26 | 海口市骨科与糖尿病医院(上海市第六人民医院海口骨科与糖尿病医院) | Detection method for indicating card in surgical instrument package |
| CN113759433B (en) * | 2021-08-12 | 2024-02-27 | 浙江啄云智能科技有限公司 | Prohibited article screening method and device and security inspection equipment |
| CN116092096A (en) * | 2021-11-05 | 2023-05-09 | 同方威视技术股份有限公司 | Method, system, device and medium for verifying the authenticity of a declared message |
| CN114399482A (en) * | 2022-01-04 | 2022-04-26 | 北京理工大学 | Contraband detection method based on weighted bidirectional fusion feature pyramid network |
| CN117518282B (en) * | 2023-11-09 | 2024-06-07 | 上海物影科技有限公司 | Miniaturized security inspection CT system |
| CN118294476A (en) * | 2024-04-15 | 2024-07-05 | 深圳市法拉姆实业有限公司 | Electronic product detection method and device and computer readable storage medium |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101644686B (en) * | 2008-08-08 | 2013-07-24 | 李海洋 | United on-line detector for explosive and poison and application thereof |
| CN101592579B (en) * | 2009-07-03 | 2011-07-06 | 公安部第一研究所 | Method and device for automatically detecting explosive substances in luggage by using multi-view X rays |
| CN104065889B (en) * | 2014-06-25 | 2017-06-27 | 山东大学 | A kind of pseudo- dual-energy x-ray linear array imaging system |
| CN105784733A (en) * | 2015-01-26 | 2016-07-20 | 公安部第研究所 | Channel type double-view-angle X-ray safety inspection device |
| CN105478373B (en) * | 2016-01-28 | 2018-02-27 | 北矿机电科技有限责任公司 | A kind of ore intelligence Grading System based on transmission of radiation identification |
-
2019
- 2019-04-04 CN CN201910269269.2A patent/CN110133739B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN110133739A (en) | 2019-08-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110133739B (en) | X-ray security check equipment and automatic image recognizing method thereof | |
| CN101799431B (en) | Machine visual on-line detection method and equipment for tank cover quality in high-speed production line | |
| EP1980839B1 (en) | Inspection device for food and its use | |
| AU2019201619B2 (en) | Extensible Millimeter Wave Security Inspection System, Scanning Unit and Security Inspection Method for Human Body | |
| CN1940555A (en) | Internal and external quality inspecting method and device for agricultural products | |
| CN109977877A (en) | A kind of safety check is intelligent to be assisted sentencing drawing method, system and system control method | |
| EP1766379A2 (en) | Non-destructive inspection of material in container | |
| CN109709108B (en) | Bottled transparent liquid impurity detection equipment and method | |
| WO2021087962A1 (en) | Automatic identification device and method for restricted articles | |
| CN113435543B (en) | Visible light and X-ray image matching method and device based on conveyor belt identification | |
| CN110850493A (en) | Linear array high-speed security inspection machine for visual image judgment of scanning imaging machine | |
| CN102445455B (en) | Machine-vision online detection device of cover quality on high-speed production line | |
| GB2613476A (en) | Device and method for security check | |
| CN111323845A (en) | Millimeter wave imaging security inspection equipment | |
| CN110766642B (en) | Artifact removing method | |
| JP2003279503A (en) | X-ray inspection apparatus | |
| CN113177925A (en) | Method for nondestructive detection of fruit surface defects | |
| CN210690842U (en) | Double-view-angle intelligent X-ray security inspection equipment | |
| CN112162324A (en) | Intelligent security inspection method for effectively improving contraband identification rate | |
| JP2022013217A (en) | X-ray image processing system and x-ray image processing method | |
| CN111458344A (en) | Mask defect visual detection method, equipment and storage medium | |
| CN108535287B (en) | Foreign matter detection system, method and device based on X-Ray imaging technology | |
| CN113790685B (en) | Automatic size detection method of to-be-detected article for double-light-source X-ray security inspection machine | |
| CN212180649U (en) | Paste appearance defect detection equipment | |
| CN106291738B (en) | A kind of safe examination system on bus |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |