WO2022194732A1 - Verfahren zum erkennen von alarmobjekten in gepäckstücken bei einer erkennungsvorrichtung - Google Patents
Verfahren zum erkennen von alarmobjekten in gepäckstücken bei einer erkennungsvorrichtung Download PDFInfo
- Publication number
- WO2022194732A1 WO2022194732A1 PCT/EP2022/056444 EP2022056444W WO2022194732A1 WO 2022194732 A1 WO2022194732 A1 WO 2022194732A1 EP 2022056444 W EP2022056444 W EP 2022056444W WO 2022194732 A1 WO2022194732 A1 WO 2022194732A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- strip
- evaluation
- luggage
- strips
- verification
- Prior art date
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 50
- 238000000034 method Methods 0.000 title claims abstract description 49
- 238000007689 inspection Methods 0.000 claims abstract description 57
- 238000011156 evaluation Methods 0.000 claims abstract description 52
- 238000012795 verification Methods 0.000 claims description 36
- 238000013528 artificial neural network Methods 0.000 claims description 16
- 230000003287 optical effect Effects 0.000 claims description 12
- 238000004590 computer program Methods 0.000 claims description 6
- 230000001537 neural effect Effects 0.000 description 4
- 238000013473 artificial intelligence Methods 0.000 description 2
- 230000005670 electromagnetic radiation Effects 0.000 description 2
- 238000012544 monitoring process Methods 0.000 description 2
- 239000013589 supplement Substances 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 230000007423 decrease Effects 0.000 description 1
- 230000001419 dependent effect Effects 0.000 description 1
- 238000012854 evaluation process Methods 0.000 description 1
- 239000002360 explosive Substances 0.000 description 1
- 238000002594 fluoroscopy Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/10—Image acquisition
- G06V10/16—Image acquisition using multiple overlapping images; Image stitching
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/70—Determining position or orientation of objects or cameras
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/22—Image preprocessing by selection of a specific region containing or referencing a pattern; Locating or processing of specific regions to guide the detection or recognition
- G06V10/225—Image preprocessing by selection of a specific region containing or referencing a pattern; Locating or processing of specific regions to guide the detection or recognition based on a marking or identifier characterising the area
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
- G06V20/64—Three-dimensional objects
- G06V20/647—Three-dimensional objects by matching two-dimensional images to three-dimensional objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/05—Recognition of patterns representing particular kinds of hidden objects, e.g. weapons, explosives, drugs
Definitions
- the present invention relates to a method for detecting alarm objects in pieces of luggage using a detection device, an evaluation device for carrying out such a method, and a corresponding computer program product.
- neural networks are often used as so-called artificial intelligence for the evaluation.
- detection devices are usually equipped with a detection unit which enables the items of luggage to be x-rayed by means of electromagnetic radiation. As a result, inspection images are created in this way, which can then be used as a basis for the evaluation.
- a disadvantage of the known solutions is that you have to wait until the entire piece of luggage has been detected by the detection device.
- a piece of luggage is completely x-rayed and in this way a verification image specific to this piece of luggage is generated.
- This verification image is then checked using a neural network, so that an evaluation with regard to the presence of alarm objects within this piece of luggage can be carried out in an automated manner.
- the disadvantage of this solution is that before starting the evaluation, you always have to wait until the complete inspection image is available. In a case in which several pieces of luggage are directly adjacent to one another or even lie one above the other within the recognition device, it takes a very long time before such a complete verification image is available.
- the object of the present invention is to at least partially eliminate the disadvantages described above.
- the object of the present invention is to improve and/or accelerate the detection of alarm objects in a piece of luggage in a cost-effective and simple manner.
- a method for detecting alarm objects in pieces of luggage is equipped with a detection device. To do this, such a procedure has the following steps:
- a method according to the invention is therefore based on the basic recognition functionality of known recognition devices. If it is used, for example, for a detection device with an electromagnetic detection unit as a detection module, then these electromagnetic detection units are usually designed in such a way that they provide the x-raying of the item of luggage in strips. This can be a permanently installed electromagnetic radiation unit and an associated detection unit. Of course, movable, in particular rotatable, detection modules can also be used for a method according to the invention.
- this two-dimensional checking strip is evaluated directly after a first two-dimensional checking strip has been recorded.
- the evaluation can be carried out manually, algorithmically and/or by using artificial intelligence with the aid of a neural network.
- This evaluation of the first two-dimensional checking strip is also directly followed by the output of an associated checking result.
- the strip-by-strip detection of the two-dimensional checking strips can progress continuously in the known manner. The evaluation of the two-dimensional inspection strips and the output of the inspection result preferably take place subsequently and also sequentially, parallel to the detection.
- the piece of luggage is x-rayed strip by strip, with the first verification strip being automatically evaluated as soon as it has been detected, and the result of this first evaluation process then being output.
- the operating personnel of such a security gate not only build up the recorded inspection strips on an associated inspection monitor, but also the inspection result, which is also output in strips, is already integrated therein. If, for example, a marking or other optical representation of alarm objects is provided as the result of the check, this can also already be done in strips.
- a method according to the invention can identify an alarm object much more quickly than is the case with the known solutions.
- sections of alarm objects are already recognized as such or recognized as belonging to such alarm objects. This makes it possible, even in the case of long pieces of luggage whose inspection images would not fit completely on an associated control monitor, to already display such an alarm object if it has been recognized by the evaluation in sections in an inspection strip.
- test results from individual test strips can be combined with one another directly after the evaluation.
- an alarm object that extends over three inspection strips can be marked as such after the end of the evaluation of this third inspection strip and an alarm is issued, even if the associated piece of luggage is, for example, 50 or more inspection strips long.
- a step of combining individual check results into a common alarm is therefore possible while the further check is still ongoing, as will be explained in more detail with reference to the subclaims.
- alarm objects can be of different sizes, so that with very narrow inspection widths of the inspection strips, the alarm objects are only partially contained in the respective inspection strips.
- An evaluation of the relative position of this detected subsection to the inspection strip makes it clear whether the alarm object is located completely within this detected inspection strip or extends into an adjacent and/or adjoining inspection strip. A distinction can thus be made as to whether an alarm object has already been recognized completely within a check zone or whether it extends beyond this boundary into a neighboring check zone. In particular, this overlapping can be used to generate and/or supplement the marking accordingly during the individual evaluation steps.
- the part section of the alarm object that is present and/or the alarm object that is present is marked optically.
- limitations on an optical image of the verification result are also possible in order to visually highlight a recognized alarm object or the recognized section of the alarm object in terms of color or in some other way.
- this is a geometric shape, for example a rectangular frame, on the output verification result.
- the optical marking is generated as the smallest possible boundary, in particular with a predetermined geometric shape, for an alarm object that extends over at least two checking sections. Since it is fundamentally possible to color the detected alarm object separately, for example, it is also conceivable and above all associated with less computational effort to place an optical marking as a frame around the detected alarm object. For example, it is conceivable that a predetermined geometric shape in the form of a rectangle, a square, an ellipse, a circle or the like is used in order to optically emphasize the alarm object as the smallest possible boundary. In such a case, the result of the check is again provided as a visual output, for example on the monitor of an operator.
- the checking strips are connected to one another, in particular along a detection direction. overlap.
- double recognition steps are carried out in this way, with the overlapping areas being taken into account when the check result is output.
- this takes place when a continuous or essentially continuous detection of the individual check strips is carried out, for example by continuously moving the item of luggage along the detection direction.
- a movable detection device for example a rotating detection module.
- the inspection strips recorded have identical or essentially identical inspection widths.
- the length of the checking strips across the checking width is identical or essentially identical.
- the identical or essentially identical geometric configuration of the checking strips simplifies subsequent processing in the evaluation and also in the output of the checking result, since essentially the same basic geometry can always be taken into account or processed.
- a checking width can be varied for at least one checking strip to be detected.
- Such a possibility of variation can be provided in a manual manner, but also in an automated manner.
- a checking width can be adapted to a respective situation. For example, it is conceivable, when recognizing a section of an alarm object that extends beyond the lateral boundary of the inspection strip, to increase the inspection width of this inspection strip to such an extent that the alarm object is completely or essentially completely within this inspection strip in the next run. If a particularly complex recognition task is present, it can also make sense to significantly reduce the width of the check, so that the computational effort in the evaluation of each check strip decreases as a result of the reduction in the width of the check.
- the check result is output continuously after each evaluation of an individual check strip.
- the output is also continuous in this way, so that not only is the inspection image built up in strips on the monitor at a security checkpoint, but the evaluation is also already integrated in this striped structure of the inspection image.
- the result of the check can be output together with the recorded check strip, but also subsequently. There is thus, so to speak, an optical passage that corresponds or substantially corresponds to the real passage of the piece of luggage along the detection direction.
- the check strips are evaluated by means of a neural network, in particular for all check strips with an identical neural network.
- This is preferably the same network, so that a neural chip with such a neural network can sequentially evaluate each check strip one after the other.
- a neural chip is to be understood as a computing unit on which the neural network is stored and/or executed.
- a set of training data constructed in strips is preferably also used for training such a neural network. It is of course also possible for such a neural network to be designed to be self-learning, ie when alarm objects are detected and verified as such, this information is returned to the neural network for self-learning functionality.
- the checking strips are directly or essentially directly adjacent to one another. This means that there is neither an overlap nor a spacing between inspection strips. In this way, uninterrupted monitoring can be ensured in particular when baggage passes through the detection device along the detection direction. As soon as there is no longer any piece of luggage, it can make sense for the inspection strips to be checked at a distance are generated in relation to one another in order to reduce the effort involved in monitoring without luggage.
- an evaluation device for detecting alarm objects in pieces of luggage with a detection device.
- Such an evaluation device has a detection module for detecting a first two-dimensional verification strip from a first part of the luggage item and for detecting at least one second two-dimensional verification strip from a second part of the luggage item.
- the evaluation device is also equipped with an evaluation module for evaluating the first two-dimensional checking strip for the presence of sections of alarm objects and for evaluating the second two-dimensional checking strip for the presence of sections of alarm objects.
- an output module is provided for outputting a check result based on the check strips.
- the evaluation module, the detection module and/or the output module is/are preferably designed to carry out a method according to the invention.
- the present invention also relates to a computer program product comprising instructions which, when the program is executed by a computer, cause the latter to carry out the steps of a method according to the invention.
- a computer program product according to the invention thus brings with it the same advantages as have been explained in detail with reference to a method according to the invention.
- FIG. 1 A security lock which is equipped with a detection device 100 is shown schematically in FIG. Baggage items G run on a baggage carousel via a conveyor belt through a detection chamber in which a detection module 110 emits electromagnetic rays. With a corresponding recognition section of the recognition module 110, a fluoroscopy result can be generated strip by strip.
- FIG. 1 it can be seen clearly in FIG. 1 that a large number of two-dimensional checking strips US1, US2 and US3 are now generated with the aid of the recognition module 110. In this embodiment, each of these checking strips US1, US2, US3 is directly connected to the previous checking strips US1, US2 and US3.
- a section TA of an alarm object AO, which is arranged in the item of luggage G, can be seen in each of the checking strips US1, US2 and US3.
- a check result UE is generated from these detected check strips US1, US2, US3, in which the alarm object AO as such in the piece of luggage G is optically represented, marked and/or displayed.
- FIG. 2 shows a possibility similar to FIG. 1, but here it is checked in which relative position the partial section TA of the alarm object AO is located in the respective checking strips US1, US2 and US3. It is thus recognized here in FIG. 2 on the left that the alarm object AO extends beyond the first checking strip US1 to the left beyond its lateral boundary BG. Once this is recognized, a connection can be made of this first Check strip US1 done with the second check strip US2. It is now recognized that the alarm object AO is even larger and also extends beyond its lateral boundary BG. Thus, the checking strips US1, US2 and US3 are combined with each other to form a common checking strip according to FIG. 2 and accordingly also output as a common checking result UE.
- FIG. 3 A variant is shown in FIG. 3, in which the checking strips US1 and US2 overlap one another along the detection direction ER.
- a freely formed alarm object AO can be seen schematically here, which is now partially located in an overlapping section.
- the two checking strips US1 and US2 for further evaluation.
- these overlapping verification strips US1 and US2 will be combined for the output of the verification result UE.
- the alarm object AO it is thus also possible for the alarm object AO to be highlighted optically and visually on a monitor at a security checkpoint with the aid of an optical marking OM, in this case a predetermined rectangle.
- the optical marking OM is a rectangle which has a minimum extent in order to delimit the alarm object AO.
- FIGS. 4 and 5 reference is made to the geometric shape of the checking strips US1 and US2.
- all inspection strips US1 and US2 are equipped with an identical inspection width UB.
- the second checking strip US2 is equipped with an increased checking width UB, in particular in order to integrate the complete alarm object AO into these two checking strips US1 and US2.
- the variation of the verification width UB can either be automated, predetermined or varied manually.
- FIGS. 6, 7 and 8 show schematically how a check result UE can build up continuously.
- a first checking strip US1 is detected and a corresponding checking result UE according to FIG. 6 is output for each strip.
- a further checking strip US2 is then detected with the aid of the recognition module 110, which then continuously expands the checking result UE in FIG. 7 from FIG or supplemented.
- a third check strip US3 is recorded in the same way and also supplements the check result UE, which thus increases and builds up further and is then shown at the bottom of FIG.
- FIG. 9 schematically shows the structure of an evaluation device 10 according to the invention. This is equipped with a detection module 20, an evaluation module 30 and an output module 40 for carrying out the method steps described there.
- a neural network NN in particular in the form of a neural chip, is provided in the evaluation module 30 in order to provide the desired evaluation and the preparation for the output of the check result.
- evaluation device 20 detection module 30 evaluation module 40 output module
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Multimedia (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Geophysics And Detection Of Objects (AREA)
Abstract
Description
Claims
Priority Applications (3)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
EP22714419.3A EP4309145A1 (de) | 2021-03-15 | 2022-03-14 | Verfahren zum erkennen von alarmobjekten in gepäckstücken bei einer erkennungsvorrichtung |
CN202280021247.4A CN117501320A (zh) | 2021-03-15 | 2022-03-14 | 用于在识别装置中识别行李件中的警报物体的方法 |
US18/282,107 US20240242459A1 (en) | 2021-03-15 | 2022-03-14 | Method for detecting alarm objects in items of luggage in a detection apparatus |
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
DE102021202510.2A DE102021202510A1 (de) | 2021-03-15 | 2021-03-15 | Verfahren zum Erkennen von Alarmobjekten in Gepäckstücken bei einer Erkennungsvorrichtung |
DE102021202510.2 | 2021-03-15 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2022194732A1 true WO2022194732A1 (de) | 2022-09-22 |
Family
ID=81327054
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/EP2022/056444 WO2022194732A1 (de) | 2021-03-15 | 2022-03-14 | Verfahren zum erkennen von alarmobjekten in gepäckstücken bei einer erkennungsvorrichtung |
Country Status (5)
Country | Link |
---|---|
US (1) | US20240242459A1 (de) |
EP (1) | EP4309145A1 (de) |
CN (1) | CN117501320A (de) |
DE (1) | DE102021202510A1 (de) |
WO (1) | WO2022194732A1 (de) |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20050198226A1 (en) * | 2003-11-19 | 2005-09-08 | Delia Paul | Security system with distributed computing |
WO2014003782A1 (en) * | 2012-06-29 | 2014-01-03 | Analogic Corporation | Automatic occlusion region identification using radiation imaging modality |
-
2021
- 2021-03-15 DE DE102021202510.2A patent/DE102021202510A1/de active Pending
-
2022
- 2022-03-14 CN CN202280021247.4A patent/CN117501320A/zh active Pending
- 2022-03-14 EP EP22714419.3A patent/EP4309145A1/de active Pending
- 2022-03-14 WO PCT/EP2022/056444 patent/WO2022194732A1/de active Application Filing
- 2022-03-14 US US18/282,107 patent/US20240242459A1/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20050198226A1 (en) * | 2003-11-19 | 2005-09-08 | Delia Paul | Security system with distributed computing |
WO2014003782A1 (en) * | 2012-06-29 | 2014-01-03 | Analogic Corporation | Automatic occlusion region identification using radiation imaging modality |
Non-Patent Citations (1)
Title |
---|
GU BANGZHONG ET AL: "Automatic and Robust Object Detection in X-Ray Baggage Inspection Using Deep Convolutional Neural Networks", IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, IEEE SERVICE CENTER, PISCATAWAY, NJ, USA, vol. 68, no. 10, 29 September 2020 (2020-09-29), pages 10248 - 10257, XP011864370, ISSN: 0278-0046, [retrieved on 20210630], DOI: 10.1109/TIE.2020.3026285 * |
Also Published As
Publication number | Publication date |
---|---|
DE102021202510A1 (de) | 2022-09-15 |
CN117501320A (zh) | 2024-02-02 |
EP4309145A1 (de) | 2024-01-24 |
US20240242459A1 (en) | 2024-07-18 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
DE69208411T2 (de) | Verfahren und Vorrichtung zum Bearbeiten von Geflügel | |
EP2114630B2 (de) | Sensorvorrichtung sowie anlage mit einem förderer und einer sensorvorrichtung | |
DE19643997A1 (de) | Verfahren zur Überwachung eines Fördergurtes und System zur Durchführung des Verfahrens | |
EP3123208A1 (de) | Detektion von gegenständen in einem objekt | |
EP1984795A1 (de) | Verfahren und vorrichtung zur überwachung einer produktionslinie | |
DE102007049702A1 (de) | Pickerstrasse | |
WO2014023626A2 (de) | Verfahren und vorrichtung zur überwachung einer fleischverarbeitungsmaschine | |
DE102017110861A1 (de) | Vorrichtung und Verfahren zum Steuern eines Materialflusses an einem Materialfluss-Knotenpunkt | |
EP2107336A1 (de) | Verfahren und Vorrichtung zur Bilderkennung von bewegten Objekten, wie etwa in Förderanlagen | |
DE102017120730A1 (de) | Vorrichtung und Verfahren zur Ausrichtung von Gebinden | |
EP2108879A1 (de) | Verfahren und Vorrichtung zur Überwachung eines Durchgangs | |
WO2022194732A1 (de) | Verfahren zum erkennen von alarmobjekten in gepäckstücken bei einer erkennungsvorrichtung | |
DE102019135620B4 (de) | Optisches Anlernen von kameraerfassten Wegstrecken | |
DE2647061C3 (de) | Personenschleuse zur Kontrolle des Zugangs von Anlagen mit verschärften Sicherheitsbestimmungen | |
EP3298439B1 (de) | Korrektur von röntgenbildern | |
DE4332645C2 (de) | Getränkefüllmaschine | |
DE102012205000B3 (de) | Verfahren und Vorrichtung zur Zählung von Objekten | |
WO2017016689A1 (de) | Förderanlage | |
DE10128722C1 (de) | Vorrichtung zur Kontrolle von Objekten | |
DE202020005501U1 (de) | Reifenprüfanlage | |
EP4309146A1 (de) | Verfahren für eine überprüfung von gepäckstücken zur erkennung von objekten | |
DE202013002483U1 (de) | Vorrichtung zur Vermessung einer Schlachttierkörperhälfte | |
DE102019106218B3 (de) | Verfahren und Vorrichtung zum Klassifizieren von Holzprodukten | |
DE102022111734A1 (de) | Vorrichtung und Verfahren zum Inspizieren von Behältnissen mit Positionserfassung | |
DE102021202511A1 (de) | Verfahren zum Erzeugen von dreidimensionalen Trainingsdaten für eine Erkennungsvorrichtung zum Erkennen von Alarmobjekten in Gepäckstücken |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 22714419 Country of ref document: EP Kind code of ref document: A1 |
|
WWE | Wipo information: entry into national phase |
Ref document number: 202280021247.4 Country of ref document: CN |
|
WWE | Wipo information: entry into national phase |
Ref document number: 2022714419 Country of ref document: EP |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
ENP | Entry into the national phase |
Ref document number: 2022714419 Country of ref document: EP Effective date: 20231016 |