CN110133739A - A kind of X-ray rays safety detection apparatus and its drawing method is known automatically - Google Patents
A kind of X-ray rays safety detection apparatus and its drawing method is known automatically Download PDFInfo
- Publication number
- CN110133739A CN110133739A CN201910269269.2A CN201910269269A CN110133739A CN 110133739 A CN110133739 A CN 110133739A CN 201910269269 A CN201910269269 A CN 201910269269A CN 110133739 A CN110133739 A CN 110133739A
- Authority
- CN
- China
- Prior art keywords
- ray
- module
- image
- prohibited items
- 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.)
- Granted
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 89
- 238000000034 method Methods 0.000 title claims abstract description 54
- 238000012545 processing Methods 0.000 claims abstract description 19
- 238000013135 deep learning Methods 0.000 claims abstract description 13
- 239000000463 material Substances 0.000 claims abstract description 13
- 238000013527 convolutional neural network Methods 0.000 claims description 14
- 238000000605 extraction Methods 0.000 claims description 11
- 238000012937 correction Methods 0.000 claims description 9
- 238000012360 testing method Methods 0.000 claims description 8
- 238000013528 artificial neural network Methods 0.000 claims description 6
- 238000001914 filtration Methods 0.000 claims description 5
- RYGMFSIKBFXOCR-UHFFFAOYSA-N Copper Chemical compound [Cu] RYGMFSIKBFXOCR-UHFFFAOYSA-N 0.000 claims description 4
- 229910052802 copper Inorganic materials 0.000 claims description 4
- 239000010949 copper Substances 0.000 claims description 4
- 238000013461 design Methods 0.000 claims description 4
- 230000004927 fusion Effects 0.000 claims description 4
- 238000011160 research Methods 0.000 claims description 4
- 241001269238 Data Species 0.000 claims description 3
- 230000003321 amplification Effects 0.000 claims description 3
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 238000012505 colouration Methods 0.000 claims description 3
- 238000007906 compression Methods 0.000 claims description 3
- 230000006835 compression Effects 0.000 claims description 3
- 238000013480 data collection Methods 0.000 claims description 3
- 230000002708 enhancing effect Effects 0.000 claims description 3
- 230000006870 function Effects 0.000 claims description 3
- 238000007477 logistic regression Methods 0.000 claims description 3
- 238000003199 nucleic acid amplification method Methods 0.000 claims description 3
- 238000013138 pruning Methods 0.000 claims description 3
- 238000013100 final test Methods 0.000 claims 1
- 239000000203 mixture Substances 0.000 abstract description 8
- 238000003384 imaging method Methods 0.000 abstract description 3
- 238000012549 training Methods 0.000 description 16
- 238000007689 inspection Methods 0.000 description 13
- 230000003466 anti-cipated effect Effects 0.000 description 4
- 238000005520 cutting process Methods 0.000 description 4
- 238000004806 packaging method and process Methods 0.000 description 4
- 230000000875 corresponding effect Effects 0.000 description 3
- 241001270131 Agaricus moelleri Species 0.000 description 2
- XEEYBQQBJWHFJM-UHFFFAOYSA-N Iron Chemical group [Fe] XEEYBQQBJWHFJM-UHFFFAOYSA-N 0.000 description 2
- 230000007812 deficiency Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000001965 increasing effect Effects 0.000 description 2
- 238000003672 processing method Methods 0.000 description 2
- 230000005855 radiation Effects 0.000 description 2
- 230000001629 suppression Effects 0.000 description 2
- 241001465754 Metazoa Species 0.000 description 1
- 230000005540 biological transmission Effects 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
- 238000009826 distribution Methods 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000006698 induction Effects 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 229910052742 iron Inorganic materials 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 238000000465 moulding Methods 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000005693 optoelectronics Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
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 a kind of X-ray rays safety detection apparatus and its automatically know drawing method, belong to radiographic imaging arts.Including base support component, data acquisition components and automatic knowledge figure component three parts.Wherein, data acquisition components include the mutually independent x-ray detection system of multiple groups, and using the detector module of high and low energy detecting plate composition；Automatic knowledge figure component includes image processing module, data memory module, display control module and prohibited items automatic identification module.The present invention can be avoided effectively by the setting independent mutually independent detection system of multiple groups since object placement angle and the interpreting blueprints of overlapped object bring are difficult；By the way that according to high and low energy detecting plate received signal value, the material classification of cargo can be calculated in image processing module；The prohibited items radioscopic image in inspected object is detected using the target detection technique based on deep learning, is compared with the learning data of prohibited items, identifies prohibited items.
Description
Technical field
The invention belongs to radiographic imaging arts, especially a kind of X-ray rays safety detection apparatus and its drawing method is known automatically.
Background technique
X-ray rays safety detection apparatus uses line scanning work mode, and internal highly sensitive X-ray detector array is swept in machinery
Article is successively scanned under the driving of imaging apparatus；Transmitted ray signal is detected, after processing, carries out image weight to the data of acquisition
It builds and has just obtained image, the internal information of article can be showed on image, image is shown on a display screen.Security staff checks screen
The image of curtain display is observed inspection, it is determined whether dangerous product.
Applicant is after carefully studying the prior art, and the discovery prior art at least has defect and deficiency below: in object
Product are by rays safety detection apparatus scanning process, and 1, need security staff to stare at screen always to see, manually differentiated, if identification passes through
Deficiency is tested, dangerous material can be allowed to pass through safety check, cause false retrieval；2, it needs security staff to stare at screen for a long time, causes vision
Fatigue is easy to miss the inspection of dangerous material, causes missing inspection；3, security staff is needed piecemeal to check, working efficiency is low；4,
For uncertain article, needs to carry out manual identified by way of image procossing (such as enhance, convert, change colour), reduce peace
Examine speed.
Summary of the invention
Goal of the invention: providing a kind of X-ray rays safety detection apparatus and its know drawing method automatically, with solve existing rays safety detection apparatus by
It is easy in the deviation of artificial observation, the problems such as causing false retrieval, the missing inspection of contraband；Safety check efficiency is improved from side simultaneously.
Technical solution: a kind of X-ray rays safety detection apparatus, comprising: base support component, data acquisition components and automatic knowledge figure group
Part three parts.
Base support component, including cross sectional shape are " mouth " glyph framework, the outer cover being fixedly mounted on outside the frame
Plate, grafting are mounted on the sealing-tape machine of the lower portion, are fixedly mounted on the multiple lifting supports and idler wheel of the base of frame,
And the lead curtain door at the frame both ends is set.
Data acquisition components, including the photoelectric sensor on the sealing-tape machine and being located on the inside of the lead curtain door is arranged in,
The x-ray detection system of the lower portion is set, and the industrial personal computer of frame side is set, wherein the industrial personal computer with
Electrically connect between the x-ray detection system, photoelectric sensor.
It is automatic to know figure component, including the radioscopic image receiving module being electrically connected with industrial personal computer, with the radioscopic image
The image processing module of receiving module connection, the prohibited items automatic identification module being connect with described image processing module, with institute
The data memory module for stating the connection of prohibited items automatic identification module, the display being connect with the prohibited items automatic identification module
Control module, and the alarm device being connect with the prohibited items automatic identification module.
In a further embodiment, the detection system includes: the first X that the frame side and top is respectively set
Ray machine mounting rack and the second X-ray machine mounting rack are fixedly mounted on the first X-ray machine mounting rack and keep with horizontal direction
The first X-ray emitter that angle is 15~20 °, be mounted on the frame the other side and top and with first X-ray
The coplanar cross sectional shape of the light source of generator is in first detector module of " 7 "；It is fixedly mounted on the second X-ray machine installation
The second X-ray emitter that angle is 15~20 ° is kept on frame and with vertical direction, is mounted on the side and bottom of the frame
Portion and with the coplanar cross sectional shape of the light source of second X-ray emitter be in " L " the second detector module, setting exist
Collimator at every group of X-ray emitter light source, and it is connected to every group of X-ray emitter and detector module, composition closure
The shielding box of annular；Wherein, the first X-ray machine mounting rack, the first X-ray emitter, the first detector module and the second X-ray
The second X-ray emitter of machine mounting rack, the second detector module are separated by pre- spacing on the transporting direction along the sealing-tape machine
From the composition mutually independent detection system of at least two groups.
In a further embodiment, the detecting plate is high and low energy detecting plate, by the detector of high energy and low energy X ray
Unit is superimposed placement up and down, and one layer of filtering copper sheet is provided between high energy and low energy X ray detector cells.
On the other hand, the automatic knowledge drawing method of a kind of X-ray rays safety detection apparatus, comprising:
After S1, inspected object enter channel, photoelectric sensor is blocked, sensor detection signal is sent to control unit starting X and penetrates
Line transmitting；
S2, the X-ray fladellum by collimator are penetrated with the cargo level of 3~10ms each about 1 millimeters thick of frequency scanning
The module at the uniform velocity moved with conveyer belt；
S3, detector module receive the X-ray signal through scanning area and carry out photoelectric conversion, complete by data collection system
At the reception of detector output signal, amplification, digitlization and to the data of industrial personal computer transmit, be ultimately delivered to automatic knowledge figure component
In；
S4, image processing module to the digital signal of each pixel carry out background-Air correction, gray scale fusion and material classification,
Geometric correction etc. carries out the primary images such as colouration, enhancing and handles function then according to 16 high energy and low energy signal after correction
Energy；
S5, display control module show these image datas on color monitor that the X-ray for showing high quality is swept
Trace designs picture；
S6, simultaneously, prohibited items automatic identification module identify prohibited items, and carry out sound-light alarm by alarm device, and
Contraband location information and type information are outlined on display control module.
In a further embodiment, the prohibited items automatic identification module advances with deep learning method to selection
The master samples of prohibited items carry out deep neural network training, and learning data is stored in data memory module.
In a further embodiment, the deep learning increases the clean package picture without prohibited items as negative example,
It is trained；And passing through image scaling, image rotation, the methods of Image Reversal variation goes to expand the quantity of trained picture.
In a further embodiment, described image processing module automatic identification prohibited items are used based on deep learning
Algorithm of target detection detects the prohibited items radioscopic image in inspected object；Wherein, prohibited items automatic identification mould
The method of block automatic identification prohibited items are as follows:
S601, pass through RPN network first, predict the position of the extraction frame of object in radioscopic image, then mapped by extracting frame
The corresponding feature of the object detected is intercepted into characteristics of image table；
S602, then again by doing further convolution operation to feature, the Characteristic Contrast with learning data, so that predicted detection arrives
The classification of object, and further the position for extracting frame is modified；
After S603, amendment extract the position of frame, above-mentioned S601~S602 step is repeated, contrast characteristic's similarity judges whether big
It is then prohibited items if more than first threshold in first threshold；
If S604, being detected article as prohibited items, image and location information, type information are saved inside data memory module.
In a further embodiment, in the algorithm of target detection, by size and length and width that candidate extraction frame is arranged
Than doing logistic regression based on candidate extraction frame to predict the position for extracting frame of object.
In a further embodiment, when the multiple dimensioned prohibited items of the algorithm of target detection detection radioscopic image,
Using the research method of more receptive field branch models；The detection method of more receptive field branch models are as follows: to analyze residual error net
Network forms the branch of the parameter of 3 shared networks, i.e. respectively the first sparse convolution point to the convolutional layer of network as frame
Branch, the second sparse convolution branch, third sparse convolution branch；Each branch is all made of RPN and R-CNN network, and RPN and R-
CNN network parameter be also it is shared, then by NMS merge three branches testing result, obtain the final detection knot of model
Fruit.
In a further embodiment, the depth convolutional neural networks that the deep learning method uses need to model into
Row compression and optimization reduce the filter quantity in convolutional network using beta pruning compression method.
In a further embodiment, the prohibited items automatic identification module automatic identification prohibited items are to suspicious region
The method of radioscopic image designed as follows:
A1, suspicious region image data is obtained, i.e. interception image prohibited items automatic identification module 21 knows phase in drawing method automatically
Knowledge and magnanimity are less than first threshold, but are greater than second threshold；
A2, suspicious region image is carried out by piecemeal using cutting algorithm, obtains several block images, each block image is carried out
Number；
A3, selected part block image are handled, and obtain its characteristic parameter, and it is compared in the database；
A4, when the similarity of article in a certain block image and database is higher than desired value, choose its adjacent block image
It compares；
If the image similarity of A5, adjacent block is higher than desired value, the block image and adjacent image are merged into same area
Block image；Continue step A4~A5, until the adjacent side of the region image data finally synthesized is not above expected similarity
Block image；
A6, the article in image data finally obtained after merging and database is compared, judges whether it is prohibited items；When super
When the similarity of the block image and database article of crossing anticipated number is lower than desired value, stop detection.
In this regard, the prohibited items automatic identification module automatic identification prohibited items are to the radioscopic image of suspicious region
Method can also be designed as follows:
B1, suspicious region image data is obtained, i.e. interception image prohibited items automatic identification module 21 knows phase in drawing method automatically
It is less than first threshold like degree, but is greater than the radioscopic image of second threshold；
B2, suspicious region image is carried out by piecemeal using cutting algorithm, obtains several block images, each block image is carried out
Number；
B3, selected part block image are handled, multiple pixels in uniform selected block image, and obtain its gray scale
Value, compares the gray value of two neighboring pixel, judges whether to be greater than desired value, identification, pickup are inconsistent with the gray scale of two sides
Image；
B4, when there is the image inconsistent with the gray scales of two sides in a certain block image, choose its adjacent block image and repeat
B3；
If the image inconsistent with the gray scale of two sides occur in B5, adjacent block image, by the block image and adjacent image
The inconsistent image of the gray scale of middle two sides is merged into a new image, Contiguous graphics region；Continue step B4~B5, directly
There is no the gray scale difference value of two neighboring pixel greater than the image of desired value to the adjacent side of the region image data finally synthesized；
B6, by the silhouette contrast of the article in the image data and database finally picked up after merging, judge whether it is violated object
Product；When the similarity of the block image and database article that are more than anticipated number is lower than desired value, stop detection.
The utility model has the advantages that the present invention relates to a kind of X-ray rays safety detection apparatus and its knowing drawing method automatically and passing through in equipment aspect
The independent mutually independent detection system of multiple groups is set, can effectively avoid bringing due to object placement angle and overlapped object
Interpreting blueprints it is difficult, thus the case where effectively lowering the missing inspection and artificial open box to check of prohibited items；By using high and low energy
For detecting plate according to high and low energy detecting plate received signal value, the material classification of cargo can be calculated in image processing module, side
Just the material composition of tested cargo is identified.In terms of method, using the target detection technique based on deep learning to inspected object
In prohibited items radioscopic image detected, compared, identified separated with convolutional neural networks with the Comparison of standards of prohibited items
Prohibited cargo product；The accuracy of identification of multiple dimensioned prohibited items detection is improved using the research method of more receptive field branch models.It solves
The problems such as existing rays safety detection apparatus is easy due to the deviation of artificial observation, is caused false retrieval, the missing inspection of contraband；Simultaneously from side
Improve safety check efficiency.
Detailed description of the invention
Fig. 1 is structural schematic diagram of the invention.
Fig. 2 is schematic cross-section of the invention.
Fig. 3 is the connection schematic diagram of rays safety detection apparatus of the invention.
Fig. 4 is more receptive field branch model schematic illustrations of the invention.
Fig. 5 is pressurized tank radioscopic image in the embodiment of the present invention 2.
Appended drawing reference are as follows: frame 1, outer cover plate 2, sealing-tape machine 3, lifting support 4, idler wheel 5, lead curtain door 6, photoelectric sensor 7,
First X-ray machine mounting rack 8, the second X-ray machine mounting rack 9, the first X-ray emitter 10, the first detector module 11, the 2nd X are penetrated
Line generator 12, the second detector module 13, collimator 14, shielding box 15, detecting plate 16, industrial personal computer 17, image receiver module
18, image processing module 19, data memory module 20, prohibited items automatic identification module 21, display control module 22, alarm device
23, the first sparse convolution branch 24, the second sparse convolution branch 25, third sparse convolution branch 26.
Specific embodiment
In the following description, a large amount of concrete details are given so as to provide a more thorough understanding of the present invention.So
And it is obvious to the skilled person that the present invention may not need one or more of these details and be able to
Implement.In other examples, in order to avoid confusion with the present invention, for some technical characteristics well known in the art not into
Row description.
In the description of the present invention, it should be noted that term " center ", "upper", "lower", "left", "right", "vertical",
The orientation or positional relationship of the instructions such as "horizontal", "inner", "outside" be based on the orientation or positional relationship shown in the drawings, merely to
Convenient for description the present invention and simplify description, rather than the device or element of indication or suggestion meaning must have a particular orientation,
It is constructed and operated in a specific orientation, therefore is not considered as limiting the invention.In addition, term " first ", " second ",
" third " is used for descriptive purposes only and cannot be understood as indicating or suggesting relative importance.
Embodiment 1
As shown in attached drawing 1~3, a kind of X-ray rays safety detection apparatus, (hereinafter referred to as " rays safety detection apparatus ") include: base support component,
Data acquisition components and automatic knowledge figure component three parts.
Base support component, comprising: frame 1, outer cover plate 2, sealing-tape machine 3, lifting support 4, idler wheel 5, lead curtain door 6.Wherein,
The cross sectional shape of frame 1 is " mouth " font, and outer cover plate 2 is fixedly mounted on outside the frame 1, and 3 grafting of sealing-tape machine is mounted on institute
It states inside frame 1, is fixedly mounted on the multiple lifting supports 4 and idler wheel 5 of 1 bottom of frame, lead curtain door 6 is arranged in the frame
1 both ends of frame；The lifting support 4 is adjusted by tools such as spanners, can be adjusted up and down between the frame 1 and supporting plane
Distance plays the role of lifting.When lifting support 4 is less than 5 height of idler wheel, set using the mobile safety check of idler wheel 5
It is standby.Meanwhile lead curtain door 6 and outer cover plate 2 are used to reduce the radiation of X-ray.
Data acquisition components, comprising: be arranged on the sealing-tape machine 3 and be located at the optoelectronic induction of 6 inside of lead curtain door
The x-ray detection system inside the frame 1 is arranged in device 7, and the industrial personal computer 17 of 1 side of frame is arranged in, wherein described
It is electrically connected between industrial personal computer 17 and the x-ray detection system, photoelectric sensor 7.Wherein, the detection system includes:
One X-ray machine mounting rack 8, the second X-ray machine mounting rack 9, the first X-ray emitter 10, the first detector module 11, the second X-ray
Generator 12, the second detector module 13, collimator 14, shielding box 15, detecting plate 16, industrial personal computer 17.First X-ray machine mounting rack
8 and second X-ray machine mounting rack 9 be respectively set 1 side of frame and top, the first X-ray emitter 8 is fixedly mounted on described
On first X-ray machine mounting rack 8, and keeping angle with horizontal direction is 15~20 °, and the first detector module 11 is mounted on the frame
The other side and top of frame 1, and be in " 7 " word with the coplanar cross sectional shape of the light source of first X-ray emitter 10；Second
X-ray emitter 12 is fixedly mounted on the second X-ray machine mounting rack 9, and keeping angle with vertical direction is 15~20 °,
Second detector module 13 is mounted on the side and bottom and same with the light source of second X-ray emitter 12 of the frame 1
The cross sectional shape of plane is in " L ", and collimator 14 is arranged at every group of X-ray emitter light source, and shielding box 15 is connected to every group of X and penetrates
Line generator and detector module, composition closed annular, for reducing the radiation leaks of X-ray.Wherein, first X-ray machine
Mounting rack 8,9 second X-ray emitter of the first X-ray emitter 10, the first detector module 11 and the second X-ray machine mounting rack
12, the second detector module 13 predetermined distance on the transporting direction along the sealing-tape machine 3, two groups of composition are mutually indepedent
Detection system.The rays safety detection apparatus uses the independent mutually independent detection system of multiple groups, can effectively avoid putting due to object
It puts angle and the interpreting blueprints of overlapped object bring is difficult, to effectively lower the missing inspection and artificial open box to check of prohibited items
The case where；And every group of L-shaped arrangement of detector module, it is mounted on the position diagonal with X-ray emitter, X-ray emitter
The X-ray of sending can scan entire inspection channel cross section, further decrease the probability of missing inspection.
As a preferred embodiment, the detecting plate 16 is high and low energy detecting plate, by the detector of high energy and low energy X ray
Unit is superimposed placement up and down, and one layer of filtering copper sheet is provided between high energy and low energy X ray detector cells.X is kept to penetrate
The power of line generator is definite value, and the light source one apart from X-ray emitter is positioned close to due to low energy X ray detector cells
End first senses X ray, and the low-energy component in reception in X ray；Through filtering copper sheet filtering X ray in low energy at
Point, sigmatron detector cells receive the high energy composition in X ray, according to 16 received signal value of high and low energy detecting plate,
The material classification of cargo can be calculated in image processing module 19, and different materials are marked by different face respectively on a display screen
Color facilitates inspection personnel to identify the material composition of tested cargo.For more traditional dual-energy x-ray generator, entire safety check is set
Standby space layout is more simple, and economic cost is more saved.
It is automatic to know figure component, comprising: image receiver module 18, image processing module 19, data memory module 20, display control
Molding block 22, alarm device 23.Image receiver module 18 and industrial personal computer 17 are electrically connected, for receiving radioscopic image；Image procossing
Module 19 is connect with described image receiving module 18, is used for radioscopic image processing module 19；Prohibited items automatic identification module 21
Be connected with described image image processing module 19, for identification with detection prohibited items；Data memory module 20 and violated object
Product automatic identification module 21 connects, for storing original image, deep learning example and automatic knowledge figure as a result, assistant images are handled
Module 19 knows figure, and display control module 22 is connect with prohibited items automatic identification module 21, exports for image；Alarm device 23 with
Prohibited items automatic identification module 21 connects, including a LED light and buzzer, when there is prohibited items to pass through, alarm device 23
Issue optoacoustic alarm.
Embodiment 2
On the other hand, the automatic knowledge drawing method about the rays safety detection apparatus is further illustrated, specifically:
After S1, inspected object enter channel, photoelectric sensor is blocked, sensor detection signal is sent to control unit starting X and penetrates
Line transmitting；
S2, the X-ray fladellum by collimator 14 are worn with the cargo level of 3~10ms each about 1 millimeters thick of frequency scanning
Detector module is reached with the cargo that conveyer belt at the uniform velocity moves thoroughly；
S3, detector module receive the X-ray signal through scanning area and carry out photoelectric conversion, complete by data collection system
At the reception of detector output signal, amplification, digitlization and to the data of industrial personal computer 17 transmit, be ultimately delivered to automatic knowledge figure component
In；
S4, image processing module 19 carry out background-Air correction, gray scale fusion and substance point to the digital signal of each pixel
Class, geometric correction etc. carry out the processing of the primary images such as colouration, enhancing then according to 16 high energy and low energy signal after correction
Function；
S5, display control module 22 show these X-ray image datas on color monitor, show high quality
X-ray scanning image；
S6, simultaneously, prohibited items automatic identification module 21 identifies prohibited items, and carries out sound-light alarm by alarm device 23,
And contraband location information and type information are outlined on display control module 22.
Wherein, the collimated X-ray fladellum cargo level that frequency scans about 1 millimeters thick each with 3~10ms,
The signal of each level is converted into 22 previous column image of display control module, and gray value of image represents X-ray and absorbed by object
Degree.Tested cargo is scanned to level one by one, the cargo radioscopic image that the transmission data of level obtains after treatment
It shows on the screen line by line, after the structure at all levels image of tested cargo combines, forms its complete X-ray
Image.
Wherein, the prohibited items automatic identification module 21 needs to advance with deep learning method to the violated object of selection
The master sample of product carries out deep neural network training, and stores data in data memory module 20.
As a preferred embodiment, increases the clean package picture without prohibited items in the deep neural network training and make
Be negative example, is trained.In the model of conventional depth neural metwork training, merely enters and indicate the picture of positive example and be trained.But
It is to be found during actual implementation process and Experimental comparison, the false recognition rate for only using positive example picture training pattern is too high.Cause
For in training process, model only learns the feature to positive example object, and other negative example objects may also have identical feature, so
This may be born example object and be mistakenly identified as positive example by model.As shown in Fig. 5, blue of the study to pressurized tank in training process
And circular feature, then due to there is no other round in training process and blue but not be pressurized tank object as training, that
It is just arrived without calligraphy learning, the difference of other features other than blue and these round features, because handle can be easy when detection
Round and blue object identification is at pressurized tank, so false recognition rate is relatively high.By introducing the feature of negative typical subject, energy
It is enough greatly to reduce misclassification rate, improve the automatic accuracy rate and safety check efficiency for knowing figure.Meanwhile by image scaling, image rotation,
The methods of Image Reversal variation goes to expand the quantity of trained picture；Picture number in addition to increasing training, can also increase in this way
Adaptability of the model to the object of different scale.
After radioscopic image is sent into automatic knowledge figure component, the 21 violated object of automatic identification of prohibited items automatic identification module
Product detect the prohibited items radioscopic image in inspected object using the target detection technique based on deep learning.Wherein,
Prohibited items automatic identification module 21 knows drawing method automatically specifically:
S601, pass through RPN network (Region Proposal Network, zone scheme network), prediction radioscopic image first
The position of the extraction frame of middle object, is then mapped to the corresponding spy of object for intercepting and detecting in characteristics of image table by extracting frame
Sign；
S602, then again by doing further convolution operation to feature, the Characteristic Contrast with learning data, reference picture handle mould
The material classification of cargo is calculated according to 16 received signal value of high and low energy detecting plate for block 19, so that predicted detection is to object
Classification, and further the position for extracting frame is modified；
After S603, amendment extract the position of frame, above-mentioned S601~S602 step is repeated, contrast characteristic's similarity judges whether big
It is then prohibited items if more than first threshold in first threshold；
If S604, being detected article as prohibited items, image and location information, type letter are saved inside data memory module 20
Breath, it can inquire the contraband image of discovery, trace original image, and Comparison of standards Neural Network Data can be enriched.
Wherein, the object detection method is a kind of two-stage target detection (second order target detection), in the present embodiment
Using Faster-RCNN algorithm, but it is not limited to Faster-RCNN algorithm, can also be R-CNN, SPPNet, Fast R-
CNN, FPN scheduling algorithm.Two-stage target detection will test problem and be divided into two stages, and first stage generates time first
Favored area, comprising the general location information of target, then second stage carries out classification and position refine to candidate region, compares
For traditional one-stage target detection (single order target detection), one-stage target detection is based on homing method mesh
Mark test problems, into being handled, so not needing to generate the candidate region stage, but use network direct as regression problem
Input picture is handled, output detects classification, position and the confidence level of target.Firstly, since one-stage target is examined
An only polytypic RPN network from network structure is surveyed, the first rank of two-stage object detection method is equivalent to
Section.Its prediction result is to predict to obtain from feature corresponding in characteristics of image table, and two-stage object detection method meeting
Further fining after being done again to extraction candidate frame to the above results, thus it is more accurate.For example, a candidates selection frame has
The 50% of a target may be covered only, but is used as complete positive sample, therefore its prediction has error certainly.Secondly,
One-stage algorithm is poor to small target deteection effect, if all candidates selection frames are all not covered with this target, that
This target will missing inspection.If a bigger extraction candidate frame covers this target, biggish extraction is waited
Select frame that can weaken the real features of target, validity will not be high.Candidate frame in two-stage algorithm can put target
Greatly, the feature of Small object is amplified, and feature contour is also relatively sharp, therefore is detected also more accurate.
As a preferred embodiment, the prohibited items automatic identification module 21 is according to the prohibited items sample of radioscopic image
This distribution situation, Optimized model improve Detection accuracy；Method particularly includes: it is extracted and is waited by setting in algorithm of target detection
The size and length-width ratio for selecting frame do logistic regression based on candidate frame is extracted to predict the position for extracting frame of object.To improve
The automatic speed of service for knowing module, improves safety check rate.
Since different size of receptive field is different to different scale object detection effect, the detection of the object of different scale
The being positively correlated property of size of performance and receptive field.Therefore in the algorithm of target detection of radioscopic image, detect multiple dimensioned violated
In Articles detecting, using the research method of more receptive field branch models.As shown in Fig. 4, it may be assumed that analyze residual error network as frame
Frame 1 forms three branches, the parameter of all shared network of these three branches, only in each branch to the convolutional layer of network
Common convolution is substituted for expansion convolution respectively, i.e. respectively the first sparse convolution branch 24, the second sparse convolution branch 25, third
Sparse convolution branch 26.The receptive field of branch each in this way can be different, but network structure is all consistent with parameter.Each point
Using RPN and R-CNN network, (Regions with CNN features is based on convolutional neural networks, linear regression after branch
With a kind of target detection technique of support vector machines scheduling algorithm), and RPN and R-CNN network parameter be also it is shared, then pass through
NMS(Non-Maximum Suppression, non-maxima suppression) fusion three branches testing result, it is final to obtain model
Testing result.
The branch of different size receptive field is constructed by the sparse convolution of different expansion convolution, each branch's receptive field is big
It is small different for solving the problems, such as multiple dimensioned object detection, but the network model parameter of each branch is consistent.Different
The shared network parameter of branch has the following advantages that, first is that comparing original model is increased without parameter, prevents parameter from excessively making
At over-fitting.Second is that having same feature representation ability to the object of different scale, because the parameter of different branches is consistent.
Third is that because three branches network parameter be it is shared, the object of various scales all utilizes in network training process
, shared network parameter can be all updated by backpropagation.And traditional training process generally uses FPN(feature
Pyramid networks, feature pyramid), in FPN training process, the characteristics of image table of a certain layer only can be with size at certain
One section object is trained, may object sample in this size section it is fewer, the training that will lead to this layer is insufficient.
And the mode that network parameter is shared by different branches then can be to avoid this problem.
In order to avoid receptive field and the unmatched situation of object scale, each branch only trains size sample in a certain range
This, is avoided the object of extreme scale for the influence of test performance.The area of the sample of the training of three sparse convolution branches
Section is respectively [0,90], [90,180], [180, ∞], and better area interval division can be by comparative experiments come really
It is fixed.Although each branch use different scale sample training because branch parameter be it is shared, model all fills
Divide the sample using each size, so as to preferably solve the target detection problems of multiple dimensioned object.
As a preferred embodiment, pretend to solve some of the staff to contraband, to escape detection instrument, disobeys
Rule bring article into safety zone, provide following preferred embodiment.In coherence check, have some of the staff by by prohibited items into
The detachable separately-loaded design of row, or part is pretended.Such as the shank of cutter is removed, and the shape of cutter is carried out
Design shows it is iron plate and cylindrical object on screening machine, may escape detection, bring big inconvenience to work.
There are also personnel to pretend prohibited items, such as, material identical packaging unit consistent with prohibited items shape by one,
Prohibited items are entrenched in packaging unit, the gap between prohibited items and packaging unit is smaller, is difficult to be distinguished.
For this purpose, being designed as follows image processing method:
A1, suspicious region image data is obtained, i.e. interception image prohibited items automatic identification module 21 knows phase in drawing method automatically
Knowledge and magnanimity are less than first threshold, but are greater than second threshold；
A2, suspicious region image is carried out by piecemeal using cutting algorithm, obtains several block images, each block image is carried out
Number；
A3, selected part block image are handled, and obtain its characteristic parameter, and it is compared in the database；
A4, when the similarity of article in a certain block image and database is higher than desired value, choose its adjacent block image
It compares；
If the image similarity of A5, adjacent block is higher than desired value, the block image and adjacent image are merged into same area
Block image；Continue step A4~A5, until the adjacent side of the region image data finally synthesized is not above expected similarity
Block image；
A6, the article in image data finally obtained after merging and database is compared, judges whether it is prohibited items；When super
When the similarity of the block image and database article of crossing anticipated number is lower than desired value, stop detection.
By the above method, the Articles detecting that partial region has camouflage can be come out, in this case, administrative division map
Picture, i.e., whole data are compared and are analyzed, and similarity is likely lower than desired value, to the case where missing inspection occur.Pass through
Locally connected forms block image, detects to partial region, so as to detected suspicious portion.
There are some of the staff all to pretend prohibited items by way of chimeric, to make its shape and normal condition not
Together, lead to missing inspection.In order to solve this problem.By way of material tests or the detection of image external periphery outline, when X-ray is worn
When crossing the chimeric place of prohibited items and chimeric packaging unit, due between chimeric there are certain gap, meeting in radioscopic image
There is a gap inconsistent with the gray scale of two sides, for this purpose, can also be designed as follows image processing method:
B1, suspicious region image data is obtained, i.e. interception image prohibited items automatic identification module 21 knows phase in drawing method automatically
It is less than first threshold like degree, but is greater than the radioscopic image of second threshold；
B2, suspicious region image is carried out by piecemeal using cutting algorithm, obtains several block images, each block image is carried out
Number；
B3, selected part block image are handled, multiple pixels in uniform selected block image, and obtain its gray scale
Value, compares the gray value of two neighboring pixel, judges whether to be greater than desired value, identification, pickup are inconsistent with the gray scale of two sides
Image；
B4, when there is the image inconsistent with the gray scales of two sides in a certain block image, choose its adjacent block image and repeat
B3；
If the image inconsistent with the gray scale of two sides occur in B5, adjacent block image, by the block image and adjacent image
The inconsistent image of the gray scale of middle two sides is merged into a new image, Contiguous graphics region；Continue step B4~B5, directly
There is no the gray scale difference value of two neighboring pixel greater than the image of desired value to the adjacent side of the region image data finally synthesized；
B6, by the silhouette contrast of the article in the image data and database finally picked up after merging, judge whether it is violated object
Product；When the similarity of the block image and database article that are more than anticipated number is lower than desired value, stop detection.
As a preferred embodiment, the depth convolutional neural networks that the deep learning method uses, due to depth convolution
There is neural network the computational complexity of essence to need to compress model and optimized, not sacrificial in the specific implementation process
Frame per second is improved under the premise of the too many accuracy of domestic animal.In the method that beta pruning is compressed, the filter quantity in convolutional network is reduced, is reduced
Operand and parameter amount improve safety check rate to improve the automatic speed of service for knowing module.
It is further to note that specific technical features described in the above specific embodiments, in not lance
In the case where shield, it can be combined in any appropriate way.In order to avoid unnecessary repetition, the present invention to it is various can
No further explanation will be given for the combination of energy.
Claims (10)
1. a kind of X-ray rays safety detection apparatus characterized by comprising
Base support component, including cross sectional shape are " mouth " glyph framework, and the outer cover plate being fixedly mounted on outside the frame is inserted
The sealing-tape machine for being mounted on the lower portion is connect, the multiple lifting supports and idler wheel of the base of frame are fixedly mounted on, and
The lead curtain door at the frame both ends is set；
Data acquisition components, including the photoelectric sensor on the sealing-tape machine and being located on the inside of the lead curtain door, setting is arranged in
The x-ray detection system of the lower portion, and the industrial personal computer of frame side is set, wherein the industrial personal computer and the X
It is electrically connected between ray detection system, photoelectric sensor；
It is automatic to know figure component, including the radioscopic image receiving module being electrically connected with industrial personal computer, received with the radioscopic image
The image processing module of module connection, the prohibited items automatic identification module connecting with described image processing module are disobeyed with described
The data memory module for prohibiting the connection of Automatic identification module, the display control being connect with the prohibited items automatic identification module
Module, and the alarm device being connect with the prohibited items automatic identification module.
2. X-ray rays safety detection apparatus according to claim 1, which is characterized in that the detection system includes: that institute is respectively set
The the first X-ray machine mounting rack and the second X-ray machine mounting rack for stating frame side and top are fixedly mounted on the first X-ray machine peace
It shelves and keeps the first X-ray emitter that angle is 15~20 ° with horizontal direction, be mounted on the other side of the frame
It is in first detector module of " 7 " with top and with the coplanar cross sectional shape of the light source of first X-ray emitter；Gu
Dingan County keeps the second X-ray emitter that angle is 15~20 ° on the second X-ray machine mounting rack and with vertical direction,
Be mounted on the frame side and bottom and with the coplanar cross sectional shape of the light source of second X-ray emitter in " L "
The second detector module, the collimator at every group of X-ray emitter light source is set, and is connected to every group of X-ray and occurs
Device and detector module, the shielding box for forming closed annular；Wherein, the first X-ray machine mounting rack, the first X-ray emitter,
First detector module and second the second X-ray emitter of X-ray machine mounting rack, the second detector module are along the sealing-tape machine
Transporting direction on predetermined distance, form the mutually independent detection system of at least two groups.
3. X-ray rays safety detection apparatus according to claim 2, which is characterized in that the detecting plate is high and low energy detecting plate, by
The detector cells of high energy and low energy X ray are superimposed placement up and down, and are arranged between high energy and low energy X ray detector cells
There is one layer of filtering copper sheet.
4. a kind of automatic knowledge drawing method of X-ray rays safety detection apparatus characterized by comprising
After S1, inspected object enter channel, photoelectric sensor is blocked, sensor detection signal is sent to control unit starting X and penetrates
Line transmitting；
S2, the X-ray fladellum by collimator are penetrated with the cargo level of 3~10ms each about 1 millimeters thick of frequency scanning
The module at the uniform velocity moved with conveyer belt；
S3, detector module receive the X-ray signal through scanning area and carry out photoelectric conversion, complete by data collection system
At the reception of detector output signal, amplification, digitlization and to the data of industrial personal computer transmit, be ultimately delivered to automatic knowledge figure component
In；
S4, image processing module to the digital signal of each pixel carry out background-Air correction, gray scale fusion and material classification,
Geometric correction etc. carries out the primary images such as colouration, enhancing and handles function then according to 16 high energy and low energy signal after correction
Energy；
S5, display control module show these image datas on color monitor that the X-ray for showing high quality is swept
Trace designs picture；
S6, simultaneously, prohibited items automatic identification module identify prohibited items, and carry out sound-light alarm by alarm device, and
Contraband location information and type information are outlined on display control module.
5. the automatic knowledge drawing method of X-ray rays safety detection apparatus according to claim 4, which is characterized in that the prohibited items
Automatic identification module advances with deep learning method and carries out deep neural network instruction to the master sample of the prohibited items of selection
Practice, and learning data is stored in data memory module.
6. the automatic knowledge drawing method of X-ray rays safety detection apparatus according to claim 5, which is characterized in that the deep learning
Increase the clean package picture without prohibited items as negative example, is trained；And passing through image scaling, image rotation, image turns over
The methods of transformationization goes to expand the quantity of trained picture.
7. the automatic knowledge drawing method of X-ray rays safety detection apparatus according to claim 4, which is characterized in that described image processing
Module automatic identification prohibited items use the algorithm of target detection based on deep learning, penetrate to the prohibited items X in inspected object
Line image is detected；Wherein, the method for prohibited items automatic identification module automatic identification prohibited items are as follows:
S601, pass through RPN network first, predict the position of the extraction frame of object in radioscopic image, then mapped by extracting frame
The corresponding feature of the object detected is intercepted into characteristics of image table；
S602, then again by doing further convolution operation to feature, the Characteristic Contrast with learning data, so that predicted detection arrives
The classification of object, and further the position for extracting frame is modified；
After S603, amendment extract the position of frame, above-mentioned S601~S602 step is repeated, contrast characteristic's similarity judges whether big
It is then prohibited items if more than first threshold in first threshold；
If S604, being detected article as prohibited items, image and location information, type information are saved inside data memory module.
8. the automatic knowledge drawing method of X-ray rays safety detection apparatus according to claim 7, which is characterized in that the target detection
In algorithm, by the way that the size and length-width ratio of candidate extraction frame is arranged, logistic regression is done to predict object based on candidate extraction frame
Extraction frame position.
9. the automatic knowledge drawing method of X-ray rays safety detection apparatus according to claim 7, which is characterized in that the target detection
When algorithm detects the multiple dimensioned prohibited items of radioscopic image, using the research method of more receptive field branch models；More senses
By the detection method of wild branch model are as follows: to analyze residual error network as frame, form 3 shared networks to the convolutional layer of network
Parameter branch, i.e. respectively the first sparse convolution branch, the second sparse convolution branch, third sparse convolution branch；Each
Branch is all made of RPN and R-CNN network, and RPN and R-CNN network parameter be also it is shared, then pass through NMS and merge three points
The testing result of branch, obtains the final testing result of model.
10. the automatic knowledge drawing method of the X-ray rays safety detection apparatus according to claim 5 or 7, which is characterized in that the depth
The depth convolutional neural networks that learning method uses need to compress model and optimized, and using beta pruning compression method, reduce
Filter quantity in 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 true CN110133739A (en) | 2019-08-16 |
| CN110133739B 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) |
Cited By (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 |
| CN110531435A (en) * | 2019-08-29 | 2019-12-03 | 公安部第一研究所 | It is a kind of for testing the test box and test method of contraband probe algorithm |
| CN110909604A (en) * | 2019-10-23 | 2020-03-24 | 深圳市华讯方舟太赫兹科技有限公司 | Security image detection method, terminal device and computer storage medium |
| CN110956225A (en) * | 2020-02-25 | 2020-04-03 | 浙江啄云智能科技有限公司 | Contraband detection method and system, computing device and storage medium |
| 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 |
| 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 |
| CN111429410A (en) * | 2020-03-13 | 2020-07-17 | 杭州电子科技大学 | Object X-ray image material distinguishing system and method based on deep learning |
| CN112034523A (en) * | 2020-08-17 | 2020-12-04 | 长沙理工大学 | Hazardous chemical substance detection method and detection system |
| CN112073009A (en) * | 2020-08-06 | 2020-12-11 | 重庆东微电子股份有限公司 | Analog signal acquisition and processing system and security check machine signal processing system |
| CN112162324A (en) * | 2020-09-02 | 2021-01-01 | 海深智能科技(上海)有限公司 | Intelligent security inspection method for effectively improving contraband identification rate |
| CN112415619A (en) * | 2020-09-14 | 2021-02-26 | 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 |
| 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 |
| CN112884755A (en) * | 2021-03-11 | 2021-06-01 | 北京理工大学 | Method and device for detecting contraband |
| CN113220927A (en) * | 2021-05-08 | 2021-08-06 | 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 |
| CN113759433A (en) * | 2021-08-12 | 2021-12-07 | 浙江啄云智能科技有限公司 | Method and device for screening contraband and security inspection equipment |
| CN113960687A (en) * | 2020-07-03 | 2022-01-21 | 株式会社日立解决方案 | X-ray image processing device and X-ray image processing method |
| 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 |
| WO2023078044A1 (en) * | 2021-11-05 | 2023-05-11 | 同方威视技术股份有限公司 | Method, system and device for checking authenticity of declaration information, and medium |
| CN117518282A (en) * | 2023-11-09 | 2024-02-06 | 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 |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101592579A (en) * | 2009-07-03 | 2009-12-02 | 公安部第一研究所 | The method and the device that utilize multi-view X ray that explosive substances in luggage is surveyed automatically |
| CN101644686A (en) * | 2008-08-08 | 2010-02-10 | 李海洋 | United on-line detector for explosive and poison and application thereof |
| CN104065889A (en) * | 2014-06-25 | 2014-09-24 | 山东大学 | Pseudo dual energy X-ray line array imaging system |
| CN105478373A (en) * | 2016-01-28 | 2016-04-13 | 北矿机电科技有限责任公司 | Ore intelligent sorting control system based on ray transmission recognition |
| 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 |
-
2019
- 2019-04-04 CN CN201910269269.2A patent/CN110133739B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101644686A (en) * | 2008-08-08 | 2010-02-10 | 李海洋 | United on-line detector for explosive and poison and application thereof |
| CN101592579A (en) * | 2009-07-03 | 2009-12-02 | 公安部第一研究所 | The method and the device that utilize multi-view X ray that explosive substances in luggage is surveyed automatically |
| CN104065889A (en) * | 2014-06-25 | 2014-09-24 | 山东大学 | Pseudo dual energy X-ray line 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 |
| CN105478373A (en) * | 2016-01-28 | 2016-04-13 | 北矿机电科技有限责任公司 | Ore intelligent sorting control system based on ray transmission recognition |
Non-Patent Citations (3)
| Title |
|---|
| 姚群力: "深度卷积神经网络在目标检测中的研究进展", 《计算机工程与应用》 * |
| 王慧玲: "基于深度卷积神经网络的目标检测技术的研究进展", 《计算机科学》 * |
| 黄俊洁: "基于全局和局部卷积特征融合的车辆目标检测", 《西南科技大学学报》 * |
Cited By (32)
| 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 |
| CN110531435A (en) * | 2019-08-29 | 2019-12-03 | 公安部第一研究所 | It is a kind of for testing the test box and test method of contraband probe algorithm |
| 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 |
| CN110909604A (en) * | 2019-10-23 | 2020-03-24 | 深圳市华讯方舟太赫兹科技有限公司 | Security image detection method, terminal device and computer storage medium |
| 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 |
| CN110956225A (en) * | 2020-02-25 | 2020-04-03 | 浙江啄云智能科技有限公司 | Contraband detection method and system, computing device and storage medium |
| CN111429410A (en) * | 2020-03-13 | 2020-07-17 | 杭州电子科技大学 | Object X-ray image material distinguishing system and method based on deep learning |
| 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 |
| CN113960687B (en) * | 2020-07-03 | 2024-08-27 | 株式会社日立解决方案 | X-ray image processing device and X-ray image processing method |
| CN113960687A (en) * | 2020-07-03 | 2022-01-21 | 株式会社日立解决方案 | X-ray image processing device and X-ray image processing method |
| CN112073009A (en) * | 2020-08-06 | 2020-12-11 | 重庆东微电子股份有限公司 | Analog signal acquisition and processing system and security check machine signal processing system |
| CN112034523A (en) * | 2020-08-17 | 2020-12-04 | 长沙理工大学 | Hazardous chemical substance detection method and detection system |
| CN112162324A (en) * | 2020-09-02 | 2021-01-01 | 海深智能科技(上海)有限公司 | Intelligent security inspection method for effectively improving contraband identification rate |
| 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 |
| CN112415619A (en) * | 2020-09-14 | 2021-02-26 | 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 |
| CN112884755A (en) * | 2021-03-11 | 2021-06-01 | 北京理工大学 | Method and device for detecting contraband |
| CN113220927A (en) * | 2021-05-08 | 2021-08-06 | 百度在线网络技术（北京）有限公司 | Image detection method, device, equipment and storage medium |
| 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 |
| CN113759433A (en) * | 2021-08-12 | 2021-12-07 | 浙江啄云智能科技有限公司 | Method and device for screening contraband and security inspection equipment |
| 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 |
| WO2023078044A1 (en) * | 2021-11-05 | 2023-05-11 | 同方威视技术股份有限公司 | Method, system and device for checking authenticity of declaration information, and medium |
| 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 |
| CN117518282A (en) * | 2023-11-09 | 2024-02-06 | 上海物影科技有限公司 | Miniaturized security inspection CT system |
| 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 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110133739B (en) | 2021-07-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110133739A (en) | A kind of X-ray rays safety detection apparatus and its drawing method is known automatically | |
| US6218943B1 (en) | Contraband detection and article reclaim system | |
| JP7250331B2 (en) | Image generation device, inspection device and learning device | |
| EP2189785B1 (en) | Method and device for detecting a liquid article | |
| CN110186940B (en) | Security check identification method and device, computer equipment and storage medium | |
| CN101614683B (en) | Real-time calibration equipment and method for substance identification system | |
| CN108872085B (en) | Navel orange heart rot nondestructive testing method based on blue wave band transmission image detection | |
| US7218105B2 (en) | Nuclear quadrupole resonance measurements | |
| KR20180115214A (en) | Detection and quarantine inspection system and method | |
| CN101858734A (en) | Method and device for detecting PET bottleneck quality | |
| CN110163191A (en) | A kind of dangerous material intelligent identification Method, system and dangerous material safe examination system | |
| CN114419311B (en) | Multi-source information-based passion fruit maturity nondestructive testing method and device | |
| CN106442525A (en) | Online detecting method for wizened defects of insides of walnuts | |
| CN104181178A (en) | Channel type two-viewing-angle X-ray liquid substance safety check system | |
| WO2021087962A1 (en) | Automatic identification device and method for restricted articles | |
| CN104662412A (en) | Method of operating a radiographic inspection system with a modular conveyor chain | |
| CN111337989A (en) | Active double-visual-angle screening method and device for multistage security inspection instrument | |
| CN218743310U (en) | Multi-vision integrated intelligent sorting equipment | |
| CN111175840A (en) | Hazardous article security inspection system based on terahertz imaging technology and security inspection method thereof | |
| CN107976408A (en) | A kind of historical relic non-destructive testing device | |
| CN208795671U (en) | The structure division of image capturing system for the rotten heart non-destructive testing of navel orange | |
| JP2003279503A (en) | X-ray inspection apparatus | |
| CN210690842U (en) | Double-view-angle intelligent X-ray security inspection equipment | |
| CN117274623A (en) | Identification method and system for intelligent retrieval and identification of forbidden articles | |
| CN111458344A (en) | Mask defect visual detection method, equipment and storage medium |
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 |