CN109978827A - Violated object recognition methods, device, equipment and storage medium based on artificial intelligence - Google Patents
Violated object recognition methods, device, equipment and storage medium based on artificial intelligence Download PDFInfo
- Publication number
- CN109978827A CN109978827A CN201910136323.6A CN201910136323A CN109978827A CN 109978827 A CN109978827 A CN 109978827A CN 201910136323 A CN201910136323 A CN 201910136323A CN 109978827 A CN109978827 A CN 109978827A
- Authority
- CN
- China
- Prior art keywords
- violated
- ray
- packing material
- pixel
- violated object
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 37
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 23
- 238000012856 packing Methods 0.000 claims abstract description 163
- 239000000463 material Substances 0.000 claims abstract description 145
- 238000012216 screening Methods 0.000 claims description 44
- 238000005070 sampling Methods 0.000 claims description 39
- 210000002569 neuron Anatomy 0.000 claims description 29
- 238000012545 processing Methods 0.000 claims description 23
- 238000013527 convolutional neural network Methods 0.000 claims description 20
- 230000005540 biological transmission Effects 0.000 claims description 11
- 239000000284 extract Substances 0.000 claims description 9
- 238000000605 extraction Methods 0.000 claims description 7
- 238000001514 detection method Methods 0.000 abstract description 7
- 238000007689 inspection Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 4
- 238000012549 training Methods 0.000 description 4
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 230000000306 recurrent effect Effects 0.000 description 3
- 230000009977 dual effect Effects 0.000 description 2
- 239000002360 explosive Substances 0.000 description 2
- 239000007788 liquid Substances 0.000 description 2
- 239000004065 semiconductor Substances 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 238000010521 absorption reaction Methods 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000012790 confirmation Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 239000010985 leather Substances 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 238000004806 packaging method and process Methods 0.000 description 1
- 238000005295 random walk Methods 0.000 description 1
- 230000003252 repetitive effect Effects 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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/12—Edge-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/194—Segmentation; Edge detection involving foreground-background segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
- G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10016—Video; Image sequence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10116—X-ray image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Theoretical Computer Science (AREA)
- Quality & Reliability (AREA)
- High Energy & Nuclear Physics (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Life Sciences & Earth Sciences (AREA)
- Geophysics (AREA)
- Analysing Materials By The Use Of Radiation (AREA)
Abstract
This application involves violated object recognition methods, device, equipment and storage medium based on artificial intelligence, method includes: acquisition x-ray image;It is marked off from x-ray image and the one-to-one X-ray subgraph of each packing material;If identifying, there are suspicious violated objects in the packing material in X-ray subgraph, match corresponding preset violated probability value for each suspicious violated object;The corresponding violated probability value of each suspicious violated object is overlapped, probability value summation is obtained;If probability value summation is greater than violated object threshold value, determine that there are violated objects in the packing material in X-ray subgraph.By being identified to each in packing material in x-ray image by safety check article, obtain the violated probability value of each suspected contraband, and it sums to each violated probability value, whether judged in packing material according to probability value summation against prohibited cargo, realize the automatic detection of safety check, safety check efficiency is improved, human cost is reduced.
Description
Technical field
This application involves field of image recognition, more particularly to the violated object recognition methods based on artificial intelligence, device, equipment
And storage medium.
Background technique
Screening machine (or safety inspection instrument) is a kind of examined luggage to be sent into X-ray examination channel by means of conveyer belt
And complete the electronic equipment checked.After luggage enters X-ray examination channel, it will stop to wrap up detection sensor;Package detection passes
Sensor issues detection signal, and detection signal is sent to systems control division point;Systems control division point generates X-ray trigger signal, touching
Send out x-ray source emitting x-ray.A branch of fan-shaped x-ray beam by collimator pass through conveyer belt on by safety check article, X is penetrated
Line is absorbed by safety check article, finally bombards the dual energy semiconductor detector being mounted in channel.Dual energy semiconductor detector
X-ray is changed into signal, these very weak signals are amplified, and are sent to signal processing cabinet and are further processed, and X-ray is formed
Video.X-ray is a kind of electromagnetic wave that can penetrate the opaque articles such as timber, cardboard, leather.Safety check function is according to object pair
The image of different colours is presented in the degree of absorption of X-ray on fluorescent screen.At this moment, safety inspector quickly checks that X-ray is swept by fluorescent screen
The X-ray video retouched, by virtue of experience judges whether there is violated object.
But by the method for violated object in manual identified X-ray video and unreliable, because the places such as subway, airport are every
Its passing personnel's amount is very big, and security staff is difficult observation x-ray image of concentrating one's energy constantly, may miss dress unless you give your whole attention to it
It is possible that the luggage of violated object.Therefore, existing violated object identification method is there are violated object recognition efficiency is low, and labor intensive at
This problem of.
Summary of the invention
The main purpose of the application is to solve existing violated object identification method low efficiency and the high technology of human cost
Problem obtains the violated general of each suspected contraband by being identified to each in packing material in x-ray image by safety check article
Rate value, and sum to each violated probability value, judged to realize peace whether against prohibited cargo in packing material according to probability value summation
The automatic detection of inspection, improves safety check efficiency, reduces human cost.
A kind of violated object recognition methods based on artificial intelligence, comprising: clapped with preset sampling time interval from screening machine
X-ray image is acquired in the X-ray video taken the photograph;Each packing material region in the x-ray image is identified, according to where each packing material
Region marks off and the one-to-one X-ray subgraph of each packing material from the x-ray image;Pass through convolutional neural networks model
Convolutional layer extracts target data from the X-ray subgraph, by the pond layer of the convolutional neural networks model to the mesh
It marks data and carries out de-redundancy processing, obtain article characteristics information;By the article characteristics information input to violated object identification model;
Each in the packing material in the X-ray subgraph is identified by safety check article by the violated object identification model, if knowing
Not Chu there are suspicious violated objects in the packing material in the X-ray subgraph, then it is corresponding pre- for each suspicious violated object matching
If violated probability value;The corresponding violated probability value of each suspicious violated object is overlapped by accumulator, obtains probability value
Summation;If the probability value summation is greater than the violated object threshold value, determines to exist in the packing material in the X-ray subgraph and disobey
Prohibited cargo.
Optionally, X-ray figure is acquired in the X-ray video shot with preset sampling time interval from the screening machine
Before picture, the method also includes:
According between setting sampling time time required for the shooting area for passing through the screening machine by safety check article
Every.
Optionally, described in the setting of the time according to required for the shooting area for passing through the screening machine by safety check article
Sampling time interval, comprising:
Obtain the length of the shooting area of the screening machine;Obtain the transmission speed of the conveyer belt of the screening machine;By institute
It states length to be divided by with the transmission speed, obtains sampling time reference value;By the sampling time reference value and preset constant a
It is multiplied, obtains the sampling time interval;The constant a is greater than 0, and is less than or equal to 1.
Optionally, each packing material region in the identification x-ray image, according to each packing material region from
It is marked off in the x-ray image and the one-to-one X-ray subgraph of each packing material, comprising:
It is reference with the gray value of background area in the x-ray image, identifies the picture of the packing material in the x-ray image
Vegetarian refreshments;The all pixels point on each packing contour in the x-ray image is identified respectively, according on each packing contour
All pixels point marks off and each packing material X-ray subgraph correspondingly from the x-ray image.
Optionally, all pixels point on each packing contour identified in the x-ray image respectively, according to each
All pixels point on packing contour marks off and each packing material X-ray subgraph correspondingly from the x-ray image
Picture, comprising:
The first pixel of any packing material not being traversed in the x-ray image is extracted at random;First pixel
For any pixel point of packing material;Traverse each pixel around first pixel;If traverse the second pixel it is upper,
Under, there are the pixels of the background area in left and right four adjacent pixels, it is determined that second pixel be packaging
Pixel on object profile;Second pixel is any pixel point around first pixel;Traverse out described
All pixels point on the profile of packing material belonging to one pixel;Extract the profile of packing material belonging to first pixel
On all pixels point area defined image, as corresponding X-ray of packing material belonging to first pixel
Image.
Optionally, if being greater than the violated object threshold value in the probability value summation, determine in the X-ray subgraph
Packing material in there are after violated object, the method also includes:
Issue violated object prompt.
It is described to issue violated object prompt including at least one of following implementations: to issue voice prompt;Open warning light;?
Violated object prompting frame is popped up on the fluorescent screen of the screening machine;Violated object is identified in the x-ray image.
Optionally, the expression formula of the violated object identification model are as follows:
Wherein, I is the dimension of input vector, and V is the dimension of the article in the X-ray subgraph Jing Guo vectorization, and H is hidden
The neuron number of layer, K are the neuron number of output layer, and x is the object that the convolutional neural networks model extraction comes out
Product characteristic information, v are the vector data that the violated object identification model is converted to the article characteristics information recognition result,
For the input at hidden layer neuron current time in the violated object identification model,To be implied in the violated object identification model
The output at layer neuron current time;For the input at output layer neuron current time in the violated object identification model;
For the output at output layer neuron current time in the violated object identification model,For violated probability value.
Optionally, the article characteristics information includes contoured article characteristic information and item color characteristic information.
Based on the same technical idea, the present invention also provides a kind of violated object identification device based on artificial intelligence, packet
Include transceiver module and processing module.The processing module is used to control the transmitting-receiving operation of the transceiver module.
The transceiver module, for acquiring X-ray figure from the X-ray video that screening machine is shot with preset sampling time interval
Picture.
The processing module, each packing material region in the x-ray image for identification, according to where each packing material
Region marks off and the one-to-one X-ray subgraph of each packing material from the x-ray image;Pass through convolutional neural networks model
Convolutional layer extracts target data from the X-ray subgraph, by the pond layer of the convolutional neural networks model to the mesh
It marks data and carries out de-redundancy processing, obtain article characteristics information;By the article characteristics information input to violated object identification model;
Each in the packing material in the X-ray subgraph is identified by safety check article by the violated object identification model, if knowing
Not Chu there are suspicious violated objects in the packing material in the X-ray subgraph, then it is corresponding pre- for each suspicious violated object matching
If violated probability value;The corresponding violated probability value of each suspicious violated object is overlapped by accumulator, obtains probability value
Summation;If the probability value summation is greater than the violated object threshold value, determines to exist in the packing material in the X-ray subgraph and disobey
Prohibited cargo.
Optionally, the processing module is specifically used for taking the gray value of background area in the x-ray image as reference, identification
The pixel of packing material in the x-ray image out;It identifies respectively all on each packing contour in the x-ray image
Pixel is marked off from the x-ray image according to all pixels point on each packing contour and is corresponded with each packing material
The X-ray subgraph.
Optionally, the processing module is specifically used for extracting any packing material not being traversed in the x-ray image at random
The first pixel;First pixel is any pixel point of packing material;Traverse each picture around first pixel
Vegetarian refreshments;If traversing in the adjacent pixel in four, upper and lower, left and right of the second pixel, there are the pixels of the background area
Point, it is determined that second pixel is the pixel on packing contour;Second pixel is first pixel
Any pixel point of surrounding;Traverse out all pixels point on the profile of packing material belonging to first pixel;Extract institute
The all pixels point area defined image on the profile of packing material belonging to the first pixel is stated, as first pixel
The corresponding X-ray subgraph of packing material belonging to point.
Optionally, the processing module is also used to if it is determined that there are violated objects in packing material in the X-ray subgraph, then
Violated object prompt is issued by the transceiver module.
It is described to issue violated object prompt including at least one of following implementations: to issue voice prompt;Open warning light;?
Violated object prompting frame is popped up on the fluorescent screen of the screening machine;Violated object is identified in the x-ray image.
Optionally, the expression formula of the violated object identification model are as follows:
Wherein, I is the dimension of input vector, and V is the dimension of the article in the X-ray subgraph Jing Guo vectorization, and H is hidden
The neuron number of layer, K are the neuron number of output layer, and x is the object that the convolutional neural networks model extraction comes out
Product characteristic information, v are the vector data that the violated object identification model is converted to the article characteristics information recognition result,
For the input at hidden layer neuron current time in the violated object identification model,To be implied in the violated object identification model
The output at layer neuron current time;For the input at output layer neuron current time in the violated object identification model;
For the output at output layer neuron current time in the violated object identification model,For violated probability value.
Optionally, the article characteristics information includes contoured article characteristic information and item color characteristic information.
Based on the same technical idea, the present invention also provides a kind of computer equipments, including transceiver, memory and place
Device is managed, be stored with computer-readable instruction in the memory makes when the computer-readable instruction is executed by the processor
The processor is obtained to execute such as the step of the above-mentioned violated object recognition methods based on artificial intelligence.
Based on the same technical idea, the present invention also provides a kind of storage medium for being stored with computer-readable instruction,
When the computer-readable instruction is executed by one or more processors, it is based on so that one or more processors are executed as above-mentioned
The step of violated object recognition methods of artificial intelligence.
The application's by being identified to each in packing material in x-ray image by safety check article the utility model has the advantages that obtained each
The violated probability value of suspected contraband, and sum to each violated probability value, judged in packing material according to probability value summation
Whether against prohibited cargo, the automatic detection of safety check is realized, safety check efficiency is improved, reduces human cost.
Detailed description of the invention
Fig. 1 is the flow diagram of the violated object recognition methods based on artificial intelligence in the embodiment of the present application.
Fig. 2 is the flow diagram of step S2 in Fig. 1.
Fig. 3 is the structural schematic diagram of the violated object identification device based on artificial intelligence in the embodiment of the present application.
Fig. 4 is the structural schematic diagram of computer equipment in the embodiment of the present application.
Specific embodiment
It should be appreciated that specific embodiment described herein is not used to limit the application only to explain the application.
Those skilled in the art of the present technique are appreciated that unless expressly stated, singular " one " used herein, " one
It is a ", " described " and "the" also may include plural form.It is to be further understood that used in the description of the present application
Wording " comprising " refers to that there are the feature, program, step, operation, element and/or component, but it is not excluded that in the presence of or add
Add other one or more features, program, step, operation, element, component and/or their group.
Fig. 1 is a kind of flow chart of the violated object recognition methods based on artificial intelligence in some embodiments of the application, should
Violated object recognition methods is executed by violated object identification device, and violated object identification device can integrate in screening machine, can also be with
Screening machine electrical connection, as shown in Figure 1, may include steps of S1-S6:
S1, x-ray image is acquired from the X-ray video that screening machine is shot with preset sampling time interval.
The X-ray video that screening machine is shot is sampled with preset sampling time interval, obtains x-ray image.X-ray image
Carry article characteristics information;It is identified, identifies by safety check by safety check article in x-ray image according to article characteristics information
Whether article is suspicious violated object.
In some embodiments, before step S1, the method also includes step S11: passing through according to by safety check article
Time required for the shooting area of the screening machine sets the sampling time interval.
Due to it is each by safety check article be in the case where the conveyer belt of screening machine drives at the uniform velocity by screening machine shooting area, because
This, the time required for shooting area of the safety check article by screening machine is constant, that is to say, that respectively by safety check article in X
Presentation time in light video is constant and identical.In order to collect each X-ray figure by safety check article in X-ray video
Picture needs reasonable set sampling time interval.Assuming that being 5 seconds by the time of the shooting area of screening machine;If between the sampling time
It every being greater than 5 seconds, such as is set as 7 seconds or 8 seconds, in this way, being regarded by safety check article in X-ray since the time of acquisition x-ray image is greater than
Presentation time in frequency then leads to not collect certain x-ray images by safety check article from X-ray video, and missing inspection feelings occur
Condition;If sampling time interval is too small, such as is set as 1 second, in this way, since the time of acquisition x-ray image is less than by safety check article
Presentation time in X-ray video then causes containing same by safety check article in 5 seconds presentation times of X-ray video, X
Light image has been collected 5 times, also mean that it is same by safety check article done 5 safety checks identification, undoubtedly increase safety check
Data processing operation amount (including multiple unnecessary repetitive operation) in the process, leads to not timely feedback by safety check object
The x-ray image of product, or even occur having been passed through conveyer belt and taken away by user by safety check article, and future is remembered and it is suspicious to stop
Article., in this way can be in 5 seconds presentation times of X-ray video if sampling time interval is set as 4 seconds, it substantially can be with
Guarantee it is each in X-ray video an x-ray image has only been collected by safety check article, and will not there is a situation where missing inspections.Therefore,
Sampling time interval cannot be greater than the time shown in X-ray video by safety check article, and sampling time interval should not be too short.
Certainly, in practical application, the sampling time interval of x-ray image can carry out debugging setting according to actual needs.
Due to the position of the x-ray inspection equipment in screening machine be it is fixed, so the region that can be shot of X-ray video
Range is also fixed.It can be by screening machine by the time required for regional scope captured by X-ray video by safety check article
Conveyer belt transmission speed and the length computation at position that can be displayed in x-ray image of conveyer belt acquire.
In some embodiments, step S11 includes the following steps S111-S114:
S111, obtain the screening machine shooting area length.
The regional scope that the shooting area i.e. X-ray video of screening machine are shown.The shooting area of screening machine has been determined
Length, also determined that the actual distance that the shooting area of screening machine is passed through by safety check article.
S112, obtain the screening machine conveyer belt transmission speed.
The transmission speed of conveyer belt is can be pre-set, obtains the transmission speed of conveyer belt, has also just obtained quilt
The speed that safety check article passes through the shooting area of screening machine.
S113, the length and the transmission speed are divided by, obtain sampling time reference value.
By the length divided by the transmission speed, the sampling time reference value can be obtained.The sampling time ginseng
It examines value and illustrates the time required for the shooting area for passing through screening machine by safety check article.
S114, the sampling time reference value is multiplied with preset constant a, obtains the sampling time interval.
The constant a is greater than 0, and is less than or equal to 1.The sampling time interval is joined less than or equal to the sampling time
Examine value.
Reasonable sampling time interval is set according to sampling time reference value, so that same crossing safety check by safety check article
Only it is collected an x-ray image during machine, and will not there is a situation where missing inspections.
It, can be automatic after the length of the shooting area for the transmission speed and screening machine for obtaining conveyer belt in the present embodiment
Calculate and be arranged the sampling time interval of x-ray image so that the sampling time interval of the x-ray image of this method with by safety check article
Movement speed be adapted, save the trouble of manual setting sampling time interval.
Each packing material region in S2, the identification x-ray image, according to each packing material region from the X-ray
It is marked off in image and the one-to-one X-ray subgraph of each packing material.
Packing material is for accommodating by safety check article.
Might have in one frame x-ray image multiple passengers by safety check article, each passenger's is usually by safety check article
It is placed in respective packing material, screening machine is then to carry out safety check to the article in packing material one by one.When safety check, typically
Multiple packing materials cross screening machine successively, therefore may have multiple packing materials in a frame x-ray image.In traditional direct surveillance X
In the mode of light video, security staff, which picks out in the packing material of X-ray video, suspicious violated object, then to containing suspicious
Violated object packing material carries out out packet confirmation and checks.Security staff is actually to distinguish one by one to the packing material in X-ray video
Know.Similarly, the application, which will determine, contains suspicious violated object in the packing material of which passenger, it is necessary to first by each packing material area
It separates, whether then identifies respectively in each packing material containing suspicious violated object, when finding that it is suspicious that some packing material is kept
When violated object, need to only be further examined to the packing material for keeping suspicious violated object can.
Packing material includes the device that such as tank, basket, bottle, altar, tank, bag are used to accommodate article.It is to be appreciated that certain
A little independent articles for crossing screening machines, such as instrument and equipment may accommodate although itself is not put in packing material, inside it
Against prohibited cargo, therefore the independent article for crossing screening machine is also considered as packing material.
In some embodiments, as shown in Fig. 2, step S2 includes the following steps S21-S22:
S21, with the gray value of background area in the x-ray image be reference, identify the packing material in the x-ray image
Pixel.
Background frame shown by X-ray video be it is fixed, therefore, each pixel of the background area in x-ray image
Gray value is also fixed.Background area in x-ray image is different with packing material gray value, therefore the back in x-ray image
Scene area and packing material are easy to distinguish.Each pixel of background area in x-ray image is labeled as logical zero, by X-ray figure
Belong to each pixel of packing material as in and be labeled as logic 1, to distinguish each pixel of background area and belong to each picture of packing material
Vegetarian refreshments.
S22, all pixels point on each packing contour in the x-ray image is identified respectively, according to each packing material
All pixels point on profile marks off and each packing material X-ray subgraph correspondingly from the x-ray image.
During crossing screening machine, between the packing material of front and back generally can there are certain intervals, therefore, x-ray image
In background area each packing material is separated.The edge placement for finding each packing material and background area, has also just marked off X
The region of each packing material in light image to get to the one-to-one X-ray subgraph of each packing material, then to X-ray subgraph
In packing material carry out violated object identification.
In some embodiments, step S22 includes the following steps S221-S223:
S221, the first pixel for extracting any packing material not being traversed in the x-ray image at random.
First pixel is any pixel point of packing material.
The method that random walk is used in the application, traverses x-ray image.First in the matrix of x-ray image with
Machine obtains the first pixel of any packing material not being traversed, as packing material starting pixels point in traversal x-ray image.From
Beginning pixel position starts to traverse each pixel in the packing material being traversed where it.
Each pixel around S222, the traversal starting pixels point;If traversing the upper and lower, left and right of the second pixel
There are the pixels of the background area in four adjacent pixels, it is determined that second pixel is on packing contour
Pixel;Second pixel is any pixel point around first pixel;Traverse out first pixel
All pixels point on the profile of affiliated packing material.
Identify that the pixel in packing material beside a certain pixel whether there is the pixel of background area, and if it exists, then
Illustrate the pixel in packing material edge.The pixel in packing material is traversed using which, until traversing packet
Filling all directions of object is all background area.
All pixels point area defined figure on the profile of packing material belonging to S223, extraction first pixel
Picture, as the corresponding X-ray subgraph of packing material belonging to first pixel.
When traversing all directions of packing material is all background area, also determines that and belonged to the packing material all pixels point
Region, also determined that the packing material X-ray subgraph.After a packing material is traversed in x-ray image, return step S221,
Other packing materials not being traversed are found, continue to extract the packing material that this is not traversed, until will be each in x-ray image
Packing material X-ray subgraph marks off one by one to be come.
S3, target data is extracted from the X-ray subgraph by the convolutional layer of convolutional neural networks model, by described
The pond layer of convolutional neural networks model carries out de-redundancy processing to the target data, obtains article characteristics information;It will be described
Article characteristics information input is to violated object identification model;By the violated object identification model to the packet in the X-ray subgraph
Each in dress object is identified that there are suspicious violated in the packing material in the X-ray subgraph if identifying by safety check article
Object then matches corresponding preset violated probability value for each suspicious violated object.
The article characteristics information includes contoured article characteristic information and item color characteristic information etc..
Preset the convolutional neural networks model of 3*3 width, starting picture of the convolutional neural networks model from X-ray subgraph
The position of vegetarian refreshments starts, and using 1 pixel as stride, gradually traverses to the data of X-ray subgraph, runs convolution algorithm,
Extract the article characteristics information in X-ray subgraph.Article characteristics information is spliced into continuous data by convolutional neural networks model,
And by spliced article characteristics information input to violated object identification model.
In some embodiments, the expression formula of violated object identification model are as follows:
Wherein, I is the dimension of input vector, and V is the dimension of the article in the X-ray subgraph Jing Guo vectorization, and H is hidden
The neuron number of layer, K are the neuron number of output layer, and x is the object that the convolutional neural networks model extraction comes out
Product characteristic information, v are the vector data that the violated object identification model is converted to the article characteristics information recognition result,
For the input at hidden layer neuron current time in the violated object identification model,To be implied in the violated object identification model
The output at layer neuron current time;For the input at output layer neuron current time in the violated object identification model;
For the output at output layer neuron current time in the violated object identification model,For violated probability value.
Violated object identification model is trained in advance.Specifically, a certain amount of training X-ray is acquired from X-ray video
Image includes the characteristic information of violated object in trained x-ray image.The violated object in each trained x-ray image is manually marked out,
And corresponding violated probability value is distributed for each violated object of mark.The data for carrying out the training x-ray image manually marked are defeated
Enter to violated object identification model.Violated object identification model passes through the characteristic information of the violated object marked to trained x-ray image
Remembered, remember the weighted value for identifying each characteristic information of each violated object, each violated object is accurately identified to realize, and
Export that each violated object is corresponding to preset violated probability value.The model of article has much for identification, can use convolutional Neural net
Network and Recognition with Recurrent Neural Network realize that convolutional neural networks extract the article characteristics information in x-ray image, by the object of extraction
Product characteristic information is sent to Recognition with Recurrent Neural Network;Recognition with Recurrent Neural Network identification is according to article characteristics information to disobeying in x-ray image
Prohibited cargo is identified.The training general principles of identification model are consistent, are not repeated herein.
A possibility that different articles is violated object is different, for example, the water bottle in packing material, only with having water in packing material
Bottle is not enough to determine the liquid for having inflammable, explosive in water bottle, but also has a possibility that equipped with inflammable, explosive liquid.It therefore, is doubtful
Like the article of water bottle, one corresponding violated probability value is set.It is violated to more accurately judge to whether there is in packing material
Object establishes the identification probability list of a variety of violated objects, carries out violated object by the way of to each suspicious violated object allocation probability value
Identification.
Violated object is not necessarily certain complete article, is also possible to some position of some complete article.In order to more
Accurately judge whether article is violated object, in the violated object identification model of training, a complete article can be divided into
Multiple concrete positions, and a corresponding violated probability value is distributed for each position.For example, the bar in packing material is likely to be separated
The cutter of taboo, it is also possible to other normal goods, to distribute a corresponding violated probability similar to the bar of violated cutter
Value;Violated club can all have the handle object suitable for holding, and be corresponding violated similar to the distribution one of the handle object of violated cutter
Probability value;It will be similar to that the corresponding violated probability value of the bar of violated cutter is corresponding with the handle object of violated cutter is similar to
Violated probability value be overlapped, obtained probability value can accurately more reflect whether identified integral item is violated
Object.To the more specific of the position of article identification, the accuracy rate of identification is higher.
S4, the corresponding violated probability value of each suspicious violated object is overlapped by accumulator, obtains probability value summation.
Accumulator is preset, the corresponding violated probability value of possible violated object each in packing material is carried out by accumulator
Summation, obtained probability value summation is higher, illustrates that a possibility that packing material is included against prohibited cargo is bigger.
If S5, the probability value summation are greater than the violated object threshold value, determine in the packing material in the X-ray subgraph
There are violated objects.
Very low violated probability value is not enough in accurate judgement packing material that there are violated objects, it is likely to violated object phase
As non-violated object carry out identifying obtained violated probability value.Set a violated object threshold value, the probability value summation
Just determine to contain in packing material when the sum of each violated probability value is more than preset violated object threshold value with preset violated object threshold value
Against prohibited cargo, it can more accurately identify whether packing material contains violated object in this way.
S6, if it is determined that there are violated objects in packing material in the X-ray subgraph, then issue violated object prompt.
It is described to issue violated object prompt including at least one of following implementations: to issue prompting sound in some embodiments
Sound;Open warning light;Violated object prompting frame is popped up on the fluorescent screen of the screening machine;It is identified in the x-ray image violated
Object.
In above-described embodiment, background area and packing material in x-ray image are identified by gray value, and from x-ray image
In mark off the corresponding X-ray subgraph of each packing material, by violated object identification model to the quilt in each packing material in X-ray subgraph
Safety check article is identified, identifies violated object that may be present in each packing material, and distribute and correspond to for each suspicious violated object
Violated probability value, and sum to each violated probability value, obtain probability value summation, by probability value summation with it is preset violated
Object threshold value is compared, and whether there is violated object in the violated object of accurate judgement.The human cost during safety check is reduced, is improved
The efficiency of safety check.
Based on the same technical idea, the present invention also provides a kind of violated object identification device based on artificial intelligence, such as
Shown in Fig. 3, which includes transceiver module 1 and processing module 2.The processing module 2 is used to control the receipts of the transceiver module 1
Hair operation.
The transceiver module 1, for acquiring X-ray from the X-ray video that screening machine is shot with preset sampling time interval
Image.
The processing module 2, each packing material region in the x-ray image for identification, according to each packing material institute
It is marked off from the x-ray image and the one-to-one X-ray subgraph of each packing material in region;Pass through convolutional neural networks model
Convolutional layer extract target data from the X-ray subgraph, by the pond layer of the convolutional neural networks model to described
Target data carries out de-redundancy processing, obtains article characteristics information;The article characteristics information input to violated object is identified into mould
Type;Each in the packing material in the X-ray subgraph is identified by safety check article by the violated object identification model, if
Identify that there are suspicious violated objects in the packing material in the X-ray subgraph, then it is corresponding for each suspicious violated object matching
Preset violated probability value;The corresponding violated probability value of each suspicious violated object is overlapped by accumulator, obtains probability
It is worth summation;If the probability value summation is greater than the violated object threshold value, determine exist in the packing material in the X-ray subgraph
Violated object.
In some embodiments, the processing module 2 is specifically used for identifying the X-ray figure according to the gray value of pixel
As in background area each pixel and belong to each pixel of packing material;Each packing material in the x-ray image is identified respectively
All pixels point on profile marks off and each packet according to all pixels point on each packing contour from the x-ray image
Fill the object X-ray subgraph correspondingly.
In some embodiments, the processing module 2 is specifically used for extracting at random any in the x-ray image not being traversed
One pixel of packing material, using the pixel extracted at random as starting pixels point;It traverses around the starting pixels point
Each pixel;Judge to whether there is the background in the adjacent pixel in four, upper and lower, left and right of traversed pixel
The pixel in region;If in the adjacent pixel in four, the upper and lower, left and right of the pixel traversed, there are the background areas
Pixel, then determine the pixel traversed for the pixel on packing contour;Traverse out starting pixels point institute
All pixels point on the packing contour of category;Extract all pixels point on packing contour belonging to the starting pixels point
Area defined image, as X-ray subgraph described in packing material belonging to the starting pixels point.
In some embodiments, the processing module 2 is also used to the interior presence of packing material if it is determined that in the X-ray subgraph
Violated object then issues violated object prompt by the transceiver module 1.
In some embodiments, the expression formula of violated object identification model are as follows:
Wherein, I is the dimension of input vector, and V is the dimension of the article in the X-ray subgraph Jing Guo vectorization, and H is hidden
The neuron number of layer, K are the neuron number of output layer, and x is the object that the convolutional neural networks model extraction comes out
Product characteristic information, v are the vector data that the violated object identification model is converted to the article characteristics information recognition result,
For the input at hidden layer neuron current time in the violated object identification model,To be implied in the violated object identification model
The output at layer neuron current time;For the input at output layer neuron current time in the violated object identification model;
For the output at output layer neuron current time in the violated object identification model,For violated probability value.
In some embodiments, the article characteristics information includes contoured article characteristic information and item color feature letter
Breath.
In above-described embodiment, background area and packing material in x-ray image are identified by gray value, and from x-ray image
In mark off the corresponding X-ray subgraph of each packing material, by violated object identification model to the quilt in each packing material in X-ray subgraph
Safety check article is identified, identifies violated object that may be present in each packing material, and distribute and correspond to for each suspicious violated object
Violated probability value, and sum to each violated probability value, obtain probability value summation, by probability value summation with it is preset violated
Object threshold value is compared, and whether there is violated object in the violated object of accurate judgement.The human cost during safety check is reduced, is improved
The efficiency of safety check.
Based on the same technical idea, the present invention also provides a kind of computer equipments, as shown in figure 4, the computer is set
Standby includes transceiver 901, processor 902 and memory 903, is stored with computer-readable instruction in the memory 903, described
When computer-readable instruction is executed by the processor 902, so that the processor 902 executes the institute in the respective embodiments described above
The step of violated object recognition methods based on artificial intelligence stated.
The corresponding entity device of transceiver module 1 shown in Fig. 3 is transceiver 901 shown in Fig. 4,901 energy of transceiver
Enough realize transceiver module 1 partly or entirely, and the same or similar function.
The corresponding entity device of processing module 2 shown in Fig. 3 is processor 902 shown in Fig. 4,902 energy of processor
Enough realize processing module 2 partly or entirely, and the same or similar function.
Based on the same technical idea, the present invention also provides a kind of storage medium for being stored with computer-readable instruction,
When the computer-readable instruction is executed by one or more processors, so that one or more processors execute above-mentioned each implementation
The step of violated object recognition methods based on artificial intelligence in mode.
Through the above description of the embodiments, those skilled in the art can be understood that above-described embodiment side
Method can be realized by means of software and necessary general hardware platform, naturally it is also possible to by hardware, but in many cases
The former is more preferably embodiment.Based on this understanding, the technical solution of the application substantially in other words does the prior art
The part contributed out can be embodied in the form of software products, which is stored in a storage medium
In (such as ROM/RAM), including some instructions are used so that a terminal (can be mobile phone, computer, server or network are set
It is standby etc.) execute method described in each embodiment of the application.
Embodiments herein is described above in conjunction with attached drawing, but the application be not limited to it is above-mentioned specific
Embodiment, the above mentioned embodiment is only schematical, rather than restrictive, those skilled in the art
Under the enlightenment of the application, when not departing from the application objective and scope of the claimed protection, can also it make very much
Form, it is all using equivalent structure or equivalent flow shift made by present specification and accompanying drawing content, directly or indirectly
Other related technical areas are used in, these are belonged within the protection of the application.
Claims (10)
1. a kind of violated object recognition methods based on artificial intelligence characterized by comprising
X-ray image is acquired from the X-ray video that screening machine is shot with preset sampling time interval;
It identifies each packing material region in the x-ray image, is drawn from the x-ray image according to each packing material region
It separates and the one-to-one X-ray subgraph of each packing material;
Target data is extracted from the X-ray subgraph by the convolutional layer of convolutional neural networks model, passes through the convolution mind
Pond layer through network model carries out de-redundancy processing to the target data, obtains article characteristics information;The article is special
Reference breath is input to violated object identification model;By the violated object identification model in the packing material in the X-ray subgraph
It is each identified by safety check article, if identifying in the packing material in the X-ray subgraph there are suspicious violated object, for
Each suspicious violated object matches corresponding preset violated probability value;
The corresponding violated probability value of each suspicious violated object is overlapped by accumulator, obtains probability value summation;
If the probability value summation is greater than the violated object threshold value, determines to exist in the packing material in the X-ray subgraph and disobey
Prohibited cargo.
2. the violated object recognition methods according to claim 1 based on artificial intelligence, which is characterized in that
Before acquiring x-ray image in the X-ray video shot with preset sampling time interval from screening machine, the method
Further include:
The sampling time interval is set according to the time required for the shooting area for passing through the screening machine by safety check article.
3. the violated object recognition methods according to claim 2 based on artificial intelligence, which is characterized in that
The time according to required for the shooting area for passing through the screening machine by safety check article set between the sampling time
Every, comprising:
Obtain the length of the shooting area of the screening machine;
Obtain the transmission speed of the conveyer belt of the screening machine;
The length and the transmission speed are divided by, sampling time reference value is obtained;
The sampling time reference value is multiplied with preset constant a, obtains the sampling time interval;The constant a is greater than
0, and it is less than or equal to 1.
4. the violated object recognition methods according to claim 1 based on artificial intelligence, which is characterized in that
Each packing material region in the identification x-ray image, according to each packing material region from the x-ray image
In mark off and the one-to-one X-ray subgraph of each packing material, comprising:
It is reference with the gray value of background area in the x-ray image, identifies the pixel of the packing material in the x-ray image
Point;
The all pixels point on each packing contour in the x-ray image is identified respectively, according on each packing contour
All pixels point marks off and each packing material X-ray subgraph correspondingly from the x-ray image.
5. the violated object recognition methods according to claim 4 based on artificial intelligence, which is characterized in that
All pixels point on each packing contour identified in the x-ray image respectively, according to each packing contour
On all pixels point marked off from the x-ray image and each packing material X-ray subgraph correspondingly, comprising:
The first pixel of any packing material not being traversed in the x-ray image is extracted at random;First pixel is packet
Fill any pixel point of object;
Traverse each pixel around first pixel;If four, upper and lower, left and right for traversing the second pixel are adjacent
Pixel in there are the pixels of the background area, it is determined that second pixel be packing contour on pixel
Point;Second pixel is any pixel point around first pixel;It traverses out belonging to first pixel
All pixels point on the profile of packing material;
The all pixels point area defined image on the profile of packing material belonging to first pixel is extracted, as institute
State the corresponding X-ray subgraph of packing material belonging to the first pixel.
6. the violated object recognition methods according to any one of claims 1 to 5 based on artificial intelligence, which is characterized in that
If being greater than the violated object threshold value in the probability value summation, determine in the packing material in the X-ray subgraph
There are after violated object, the method also includes:
Issue violated object prompt;
It is described to issue violated object prompt including at least one of following implementations:
Issue voice prompt;
Open warning light;
Violated object prompting frame is popped up on the fluorescent screen of the screening machine;
Violated object is identified in the x-ray image.
7. the violated object recognition methods according to any one of claims 1 to 5 based on artificial intelligence, which is characterized in that
The expression formula of the violated object identification model are as follows:
Wherein, I is the dimension of input vector, and V is the dimension of the article in the X-ray subgraph Jing Guo vectorization, and H is hidden layer
Neuron number, K are the neuron number of output layer, and x is that the article that the convolutional neural networks model extraction comes out is special
Reference breath, v are the vector data that the violated object identification model is converted to the article characteristics information recognition result,For institute
The input at hidden layer neuron current time in violated object identification model is stated,For hidden layer mind in the violated object identification model
Output through first current time;For the input at output layer neuron current time in the violated object identification model;For institute
The output at output layer neuron current time in violated object identification model is stated,For violated probability value.
8. a kind of violated object identification device based on artificial intelligence characterized by comprising
Transceiver module, for acquiring x-ray image from the X-ray video that screening machine is shot with preset sampling time interval;
Processing module, each packing material region in the x-ray image for identification, according to each packing material region from institute
It states and is marked off in x-ray image and the one-to-one X-ray subgraph of each packing material;By the convolutional layer of convolutional neural networks model from
Extract target data in the X-ray subgraph, by the pond layer of the convolutional neural networks model to the target data into
The processing of row de-redundancy, obtains article characteristics information;By the article characteristics information input to violated object identification model;By described
Violated object identification model is identified each in the packing material in the X-ray subgraph by safety check article, if identifying the X
There are suspicious violated objects in packing material in photon image, then corresponding preset violated general for each suspicious violated object matching
Rate value;The corresponding violated probability value of each suspicious violated object is overlapped by accumulator, obtains probability value summation;If described
Probability value summation is greater than the violated object threshold value, then determines that there are violated objects in the packing material in the X-ray subgraph.
9. a kind of computer equipment, which is characterized in that including transceiver, memory and processor, be stored in the memory
Computer-readable instruction, when the computer-readable instruction is executed by the processor, so that the processor executes such as right
It is required that the step of any described violated object recognition methods based on artificial intelligence in 1 to 7.
10. a kind of storage medium for being stored with computer-readable instruction, which is characterized in that the computer-readable instruction is by one
Or multiple processors are when executing so that one or more processors execute as described in any in claim 1 to 7 based on people
The step of violated object recognition methods of work intelligence.
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910136323.6A CN109978827A (en) | 2019-02-25 | 2019-02-25 | Violated object recognition methods, device, equipment and storage medium based on artificial intelligence |
PCT/CN2019/092678 WO2020173021A1 (en) | 2019-02-25 | 2019-06-25 | Artificial intelligence-based forbidden object identification method, apparatus and device, and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910136323.6A CN109978827A (en) | 2019-02-25 | 2019-02-25 | Violated object recognition methods, device, equipment and storage medium based on artificial intelligence |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109978827A true CN109978827A (en) | 2019-07-05 |
Family
ID=67077268
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910136323.6A Pending CN109978827A (en) | 2019-02-25 | 2019-02-25 | Violated object recognition methods, device, equipment and storage medium based on artificial intelligence |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN109978827A (en) |
WO (1) | WO2020173021A1 (en) |
Cited By (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110543857A (en) * | 2019-09-05 | 2019-12-06 | 安徽启新明智科技有限公司 | Contraband identification method, device and system based on image analysis and storage medium |
CN110751079A (en) * | 2019-10-16 | 2020-02-04 | 北京海益同展信息科技有限公司 | Article detection method, apparatus, system and computer readable storage medium |
CN110852248A (en) * | 2019-11-07 | 2020-02-28 | 江苏弘冉智能科技有限公司 | Flammable and explosive area illegal equipment based on machine vision and action monitoring method |
CN111062252A (en) * | 2019-11-15 | 2020-04-24 | 浙江大华技术股份有限公司 | Real-time dangerous article semantic segmentation method and device and storage device |
CN111290040A (en) * | 2020-03-12 | 2020-06-16 | 安徽启新明智科技有限公司 | Active double-view-angle correlation method based on image recognition |
CN111325114A (en) * | 2020-02-03 | 2020-06-23 | 重庆特斯联智慧科技股份有限公司 | Security image processing method and device for artificial intelligence recognition classification |
CN112016387A (en) * | 2019-07-08 | 2020-12-01 | 杭州芯影科技有限公司 | Contraband identification method and device suitable for millimeter wave security check instrument |
CN112364903A (en) * | 2020-10-30 | 2021-02-12 | 盛视科技股份有限公司 | X-ray machine-based article analysis and multi-dimensional image association method and system |
CN113792665A (en) * | 2021-09-16 | 2021-12-14 | 山东大学 | Method for detecting intrusion of forbidden region aiming at different role authorities |
CN114758259A (en) * | 2022-06-15 | 2022-07-15 | 科大天工智能装备技术(天津)有限公司 | Package detection method and system based on X-ray object image recognition |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114255436B (en) * | 2020-09-11 | 2024-03-19 | 同方威视技术股份有限公司 | Security check image recognition system and method based on artificial intelligence |
CN114758239B (en) * | 2022-04-22 | 2024-06-04 | 安徽工业大学科技园有限公司 | Method and system for monitoring flying of article away from preset travelling line based on machine vision |
CN114994100B (en) * | 2022-06-06 | 2024-06-18 | 赵震远 | Self-service railway passenger security inspection system and detection method thereof |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105654130A (en) * | 2015-12-30 | 2016-06-08 | 成都数联铭品科技有限公司 | Recurrent neural network-based complex image character sequence recognition system |
CN107607562A (en) * | 2017-09-11 | 2018-01-19 | 北京匠数科技有限公司 | A kind of prohibited items identification equipment and method, X-ray luggage security check system |
CN107871122A (en) * | 2017-11-14 | 2018-04-03 | 深圳码隆科技有限公司 | Safety check detection method, device, system and electronic equipment |
CN108198227A (en) * | 2018-03-16 | 2018-06-22 | 济南飞象信息科技有限公司 | Contraband intelligent identification Method based on X-ray screening machine image |
CN108665509A (en) * | 2018-05-10 | 2018-10-16 | 广东工业大学 | A kind of ultra-resolution ratio reconstructing method, device, equipment and readable storage medium storing program for executing |
CN109074889A (en) * | 2016-02-22 | 2018-12-21 | 拉皮斯坎***股份有限公司 | System and method for detecting dangerous material and contraband in cargo |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9547821B1 (en) * | 2016-02-04 | 2017-01-17 | International Business Machines Corporation | Deep learning for algorithm portfolios |
CN106485268B (en) * | 2016-09-27 | 2020-01-21 | 东软集团股份有限公司 | Image identification method and device |
-
2019
- 2019-02-25 CN CN201910136323.6A patent/CN109978827A/en active Pending
- 2019-06-25 WO PCT/CN2019/092678 patent/WO2020173021A1/en active Application Filing
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105654130A (en) * | 2015-12-30 | 2016-06-08 | 成都数联铭品科技有限公司 | Recurrent neural network-based complex image character sequence recognition system |
CN109074889A (en) * | 2016-02-22 | 2018-12-21 | 拉皮斯坎***股份有限公司 | System and method for detecting dangerous material and contraband in cargo |
CN107607562A (en) * | 2017-09-11 | 2018-01-19 | 北京匠数科技有限公司 | A kind of prohibited items identification equipment and method, X-ray luggage security check system |
CN107871122A (en) * | 2017-11-14 | 2018-04-03 | 深圳码隆科技有限公司 | Safety check detection method, device, system and electronic equipment |
CN108198227A (en) * | 2018-03-16 | 2018-06-22 | 济南飞象信息科技有限公司 | Contraband intelligent identification Method based on X-ray screening machine image |
CN108665509A (en) * | 2018-05-10 | 2018-10-16 | 广东工业大学 | A kind of ultra-resolution ratio reconstructing method, device, equipment and readable storage medium storing program for executing |
Cited By (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112016387A (en) * | 2019-07-08 | 2020-12-01 | 杭州芯影科技有限公司 | Contraband identification method and device suitable for millimeter wave security check instrument |
CN110543857A (en) * | 2019-09-05 | 2019-12-06 | 安徽启新明智科技有限公司 | Contraband identification method, device and system based on image analysis and storage medium |
CN110751079A (en) * | 2019-10-16 | 2020-02-04 | 北京海益同展信息科技有限公司 | Article detection method, apparatus, system and computer readable storage medium |
CN110852248A (en) * | 2019-11-07 | 2020-02-28 | 江苏弘冉智能科技有限公司 | Flammable and explosive area illegal equipment based on machine vision and action monitoring method |
CN111062252A (en) * | 2019-11-15 | 2020-04-24 | 浙江大华技术股份有限公司 | Real-time dangerous article semantic segmentation method and device and storage device |
CN111062252B (en) * | 2019-11-15 | 2023-11-10 | 浙江大华技术股份有限公司 | Real-time dangerous goods semantic segmentation method, device and storage device |
CN111325114A (en) * | 2020-02-03 | 2020-06-23 | 重庆特斯联智慧科技股份有限公司 | Security image processing method and device for artificial intelligence recognition classification |
CN111325114B (en) * | 2020-02-03 | 2022-07-19 | 重庆特斯联智慧科技股份有限公司 | Security image processing method and device for artificial intelligence recognition classification |
CN111290040A (en) * | 2020-03-12 | 2020-06-16 | 安徽启新明智科技有限公司 | Active double-view-angle correlation method based on image recognition |
CN112364903A (en) * | 2020-10-30 | 2021-02-12 | 盛视科技股份有限公司 | X-ray machine-based article analysis and multi-dimensional image association method and system |
CN113792665A (en) * | 2021-09-16 | 2021-12-14 | 山东大学 | Method for detecting intrusion of forbidden region aiming at different role authorities |
CN113792665B (en) * | 2021-09-16 | 2023-08-08 | 山东大学 | Forbidden area intrusion detection method aiming at different role authorities |
CN114758259A (en) * | 2022-06-15 | 2022-07-15 | 科大天工智能装备技术(天津)有限公司 | Package detection method and system based on X-ray object image recognition |
CN114758259B (en) * | 2022-06-15 | 2022-09-06 | 科大天工智能装备技术(天津)有限公司 | Package detection method and system based on X-ray object image recognition |
Also Published As
Publication number | Publication date |
---|---|
WO2020173021A1 (en) | 2020-09-03 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109978827A (en) | Violated object recognition methods, device, equipment and storage medium based on artificial intelligence | |
CA2676903C (en) | Method and apparatus for determining a threat status for a liquid product | |
CN104636707B (en) | The method of automatic detection cigarette | |
CN106529602B (en) | A kind of millimeter-wave image automatic target recognition method and device | |
CN110532852A (en) | Subway station pedestrian's accident detection method based on deep learning | |
CN109001833A (en) | A kind of Terahertz hazardous material detection method based on deep learning | |
CN111145177A (en) | Image sample generation method, specific scene target detection method and system thereof | |
US20200097758A1 (en) | Method and system for object detection and classification | |
CN106548263A (en) | A kind of intelligent and safe inspection system and method | |
CN107766829A (en) | A kind of method and apparatus of Articles detecting | |
CN112612066A (en) | Personnel security check method and personnel security check system | |
CN109211951A (en) | A kind of safe examination system and safety inspection method based on image segmentation | |
CN104181178A (en) | Channel type two-viewing-angle X-ray liquid substance safety check system | |
CN104865281B (en) | Human body back scattering inspection method and system | |
CN111257957A (en) | Identification tracking system and method based on passive terahertz imaging | |
CN108957578A (en) | Rotary Terahertz safety check apparatus and safety inspection method | |
CN110119734A (en) | Cutter detecting method and device | |
KR20180082956A (en) | Security Check System and Method | |
CN109187606A (en) | A kind of combined type contraband Intelligent precise testing equipment and detection method | |
WO2023124474A1 (en) | Security check information binding method and apparatus, electronic device, and readable storage medium | |
CN112967289A (en) | Security check package matching method, system, equipment and storage medium | |
CN113706497B (en) | Intelligent contraband identification device and system | |
CN110309823B (en) | Safety inspection method and device | |
CN116563628A (en) | Security check judgment chart identification method, device, equipment and storage medium | |
CN110045349A (en) | Identification method, device, 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 |