CN115620097A

CN115620097A - Intelligent security inspection testing method based on digital twinning

Info

Publication number: CN115620097A
Application number: CN202211315452.XA
Authority: CN
Inventors: 谢群; 张雷雷; 张恩伟; 尹宇鹤; 姬光; 蒙移发
Original assignee: BEIJING TELESOUND ELECTRONICS CO LTD
Current assignee: BEIJING TELESOUND ELECTRONICS CO LTD
Priority date: 2022-10-26
Filing date: 2022-10-26
Publication date: 2023-01-17

Abstract

The embodiment of the application provides an intelligent security inspection testing method based on digital twins, which comprises the following steps: determining a real-time pushing speed in a testing process according to image acquisition information of at least one target security inspection machine, and acquiring a testing image from an alternative image set based on the real-time pushing speed; the alternative image set comprises images collected by a plurality of preset security inspection machines; pushing the test image to an artificial image judging end and an intelligent image judging end respectively based on the real-time pushing speed; and determining whether the intelligent image judging end passes the test or not based on first image judging information of the manual image judging end on the test image and second image judging information of the intelligent image judging end on the test image in the test process. The real-time simulation security inspection scene highly similar to the real security inspection process is constructed based on the digital twin, and whether the intelligent judgment graph end passes the test or not is judged reasonably, accurately and efficiently.

Description

Intelligent security inspection testing method based on digital twinning

Technical Field

The application relates to the technical field of security inspection, in particular to an intelligent security inspection testing method based on digital twins.

Background

With the development of science and technology, transportation becomes an important part in production and life. In order to guarantee the safety of transportation, the package needs to be subjected to security inspection. In order to improve the security check speed, intelligent image judging equipment is generated, and the intelligent image judging equipment is applied to an intelligent recognition algorithm, so that the intelligent security check can be realized without manpower. The advantages and disadvantages of the intelligent identification algorithm relate to whether effective security inspection can be carried out or not, and whether manual detection can be replaced or not.

In the related technology, a passenger is simulated in real time to place a package into a security check machine, and the security check machine sends an acquired image to an intelligent image judging end; and evaluating the intelligent image judging end based on the recognition result of the intelligent image judging end.

However, articles in a package are complex during real security inspection, and the difference of acquisition parameters of a related security inspection machine is large, so that the difference between the test process and the real security inspection is large, and the intelligent image judging end is difficult to be accurately tested.

Disclosure of Invention

The embodiment of the application provides an intelligent security inspection testing method based on digital twins, which is used for accurately testing an intelligent image judging end.

In a first aspect, an embodiment of the present application further provides an intelligent security inspection testing method based on digital twins, including:

determining a real-time pushing speed in the testing process according to image acquisition information of at least one target security inspection machine, and acquiring a testing image from an alternative image set based on the real-time pushing speed; the alternative image set comprises images collected by a plurality of preset security inspection machines;

respectively pushing the test image to a manual image judging end and an intelligent image judging end based on the real-time pushing speed;

and determining whether the intelligent image judging end passes the test or not based on first image judging information of the manual image judging end on the test image and second image judging information of the intelligent image judging end on the test image in the test process.

According to the scheme, the real-time pushing speed in the testing process is simulated according to the image acquisition information of at least one target security check machine, the testing image is acquired from the alternative image set based on the real-time pushing speed and is pushed to the manual image judging end and the intelligent image judging end, and a real-time simulation security check scene which is highly similar to the real security check process is constructed based on the digital twin; the safety inspection is carried out on the basis of the test image through the manual image judging end and the intelligent image judging end, and the safety inspection effect of the intelligent image judging end compared with that of the manual image judging end can be determined on the basis of the first image judging information of the manual image judging end and the second image judging information of the intelligent image judging end, so that whether the intelligent image judging end passes the test or not can be judged reasonably, accurately and efficiently.

In some optional embodiments, determining whether the intelligent judging graph end passes the test based on first judging graph information of the manual judging graph end on the test image and second judging graph information of the intelligent judging graph end on the test image in the test process includes:

selecting a target test image from all test images in the test process based on the image judging parameters of the test images; the image judging parameters of the test images comprise the total number of contraband articles contained in all the test images, the types of the contraband articles contained in all the test images, pixel parameters of all the test images, image judging modes corresponding to all the test images and part or all of information representing artificial experience corresponding to an artificial image judging end;

aiming at any target test image, determining an image security check result of the target test image; the image security inspection result is obtained by comparing the image judging information of the target test image with the actual forbidden information of the target test image; the judging image information comprises first judging image information and second judging image information of the target test image; the image security inspection result comprises part or all of positive inspection, negative inspection, false inspection and missing inspection;

and determining whether the intelligent image judging end passes the test or not based on the image security inspection results of all the target test images.

In some optional embodiments, selecting a target test image from all test images of the test process based on the mapping parameter of the test image includes:

dividing all test images into a plurality of test image sets through a decision tree model; the judging parameters of all the test images in any test image set are the same;

aiming at any test image set, determining identification difficulty information of the test image set based on a judgment parameter of a test image in the test image set;

and determining whether the test image in the test image set is a target test image or not based on the identification difficulty information of the test image set.

In some optional embodiments, the identifying difficulty information includes an identifying difficulty level, and determining whether the test image in the test image set is the target test image based on the identifying difficulty information of the test image set includes:

if the identification difficulty level is less than or equal to a preset level, determining all the test images in the test image set as target test images; the identification difficulty level of the test image set is smaller when the identification difficulty of the image judging parameter representation of the test image in the test image set is larger.

In some optional embodiments, determining whether the intelligent image judging end passes the test based on the image security inspection results of all target test images includes:

determining the test parameters of the manual image judging end in each test item based on the image security inspection results of all target test images corresponding to the manual image judging end; determining the test parameters of the intelligent image judging end in each test item based on the image security inspection results of all target test images corresponding to the intelligent image judging end; the test items comprise part or all of accuracy, false alarm rate and false alarm rate;

aiming at any test item, comparing the test parameters of the test item at the manual graph judging end with the test parameters of the test item at the intelligent graph judging end, and determining a score corresponding to the test parameter comparison result of the test item;

and if the total score of all the test items is greater than the preset score, determining that the intelligent graph judging end passes the test.

In some optional embodiments, before pushing the test image to the manual graph judging end and the intelligent graph judging end respectively based on the real-time pushing speed, the method further includes:

and adjusting the pixels of the test images based on the acquisition pixels of the target security inspection machine, and filtering the test images after the pixels are adjusted.

In some optional embodiments, if there are a plurality of target security inspection machines, adjusting pixels of each test image based on the collected pixels of the target security inspection machine includes:

aiming at any test image, adjusting an original pixel of the test image to be an acquisition pixel of any target security inspection machine; or alternatively

And aiming at any test image, selecting a target acquisition pixel closest to an original pixel of the test image from acquisition pixels of all target security inspection machines, and adjusting the original pixel of the test image to the target acquisition pixel.

In some optional embodiments, the filtering process is performed on the test image after the pixel adjustment, and includes:

and filtering the test image after the pixels are adjusted by a Gaussian filtering method.

In some alternative embodiments, each test image includes a front view and a side view; based on the real-time pushing speed, the test image is respectively pushed to an artificial image judging end and an intelligent image judging end, and the method comprises the following steps:

if the real-time pushing speed is higher than a preset speed, respectively pushing a main view in the test image to the manual image judging end and the intelligent image judging end;

otherwise, respectively pushing the front view and the side view in the test image to the manual image judging end and the intelligent image judging end.

In some optional embodiments, determining a real-time pushing speed in a test process according to image acquisition information of at least one target security inspection machine includes:

aiming at any test time in the test process, determining the corresponding time of the test time in a preset time period;

and determining the real-time pushing speed of the test moment based on the image acquisition speed of the at least one target security inspection machine at the corresponding moment in the preset time period.

In a second aspect, an embodiment of the present application provides an intelligent security inspection testing apparatus based on a digital twin, including:

the speed determining module is used for determining the real-time pushing speed in the testing process according to the image acquisition information of at least one target security inspection machine and acquiring a testing image from the alternative image set based on the real-time pushing speed; the alternative image set comprises images collected by a plurality of preset security inspection machines;

the pushing module is used for pushing the test image to a manual image judging end and an intelligent image judging end respectively based on the real-time pushing speed;

and the test module is used for determining whether the intelligent image judging end passes the test or not based on first image judging information of the manual image judging end on the test image and second image judging information of the intelligent image judging end on the test image in the test process.

In a third aspect, an embodiment of the present application provides an electronic device, which includes at least one processor and at least one memory, where the memory stores a computer program, and when the program is executed by the processor, the processor is caused to execute the intelligent security inspection testing method based on digital twin according to any one of the above first aspects.

In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, which stores a computer program executable by an electronic device, and when the program runs on the electronic device, the program causes the electronic device to execute the intelligent security inspection testing method based on digital twin according to any one of the above first aspects.

Drawings

In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings without creative efforts.

Fig. 1 is a schematic flowchart of a first digital twin-based intelligent security inspection testing method according to an embodiment of the present disclosure;

fig. 2 is a schematic flowchart of a second intelligent security inspection testing method based on digital twins according to an embodiment of the present disclosure;

fig. 3 is a schematic flowchart of a target test image determination method according to an embodiment of the present application;

fig. 4 is a schematic flowchart of a third intelligent security inspection testing method based on digital twins according to an embodiment of the present application;

fig. 5 is a schematic flowchart of a fourth intelligent security inspection testing method based on digital twins according to an embodiment of the present disclosure;

fig. 6 is a schematic flowchart of a fifth intelligent security inspection testing method based on digital twinning according to an embodiment of the present application;

FIG. 7 is a schematic diagram of a real-time pushing speed in a testing process according to an embodiment of the present application;

fig. 8 is a schematic structural diagram of an intelligent security inspection testing device based on digital twins according to an embodiment of the present application;

fig. 9 is a schematic structural diagram of an electronic device according to an embodiment of the present application.

Detailed Description

To make the objects, technical solutions and advantages of the present application clearer, the present application will be described in further detail with reference to the accompanying drawings, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.

The terms "first", "second" and "first" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of the present application, the meaning of "a plurality" is two or more unless otherwise specified.

In transportation, in order to improve the security check speed, intelligent image judging equipment is generated, and the intelligent image judging equipment is applied to an intelligent recognition algorithm, so that the intelligent security check can be realized without manpower. The advantages and disadvantages of the intelligent identification algorithm relate to whether effective security inspection can be carried out or not, and whether manual detection can be replaced or not.

In the related technology, a passenger is simulated in real time to put a package into a security check machine, and the security check machine sends an acquired image to an intelligent image judging end; and evaluating the intelligent image judging end based on the identification result of the intelligent image judging end.

However, the articles in the package are complex during real security inspection, and the difference of the acquisition parameters of the related security inspection machine is large, so that the difference between the test process and the real security inspection is large, and the intelligent image judging end is difficult to be accurately tested.

In view of this, the embodiment of the present application provides an intelligent security inspection testing method based on digital twins, which includes: determining a real-time pushing speed in a testing process according to image acquisition information of at least one target security inspection machine, and acquiring a testing image from an alternative image set based on the real-time pushing speed; the alternative image set comprises images collected by a plurality of preset security inspection machines; pushing the test image to an artificial image judging end and an intelligent image judging end respectively based on the real-time pushing speed; and determining whether the intelligent image judging end passes the test or not based on first image judging information of the manual image judging end on the test image and second image judging information of the intelligent image judging end on the test image in the test process.

According to the scheme, the real-time pushing speed in the testing process is simulated according to the image acquisition information of at least one target security inspection machine, the testing image is obtained from the alternative image set based on the real-time pushing speed and is pushed to the manual image judging end and the intelligent image judging end, namely, a real-time simulation security inspection scene which is highly similar to the real security inspection process is constructed based on the digital twins; the safety inspection is carried out on the basis of the test images by the artificial image judging end and the intelligent image judging end respectively, and the safety inspection effect of the intelligent image judging end compared with that of the artificial image judging end can be determined on the basis of the first image judging information of the artificial image judging end and the second image judging information of the intelligent image judging end, so that whether the intelligent image judging end passes the test or not can be judged reasonably, accurately and efficiently.

The following detailed description will be given with reference to the accompanying drawings and specific embodiments to explain the technical solutions of the present application and how to solve the above technical problems. The following several specific embodiments may be combined with each other, and details of the same or similar concepts or processes may not be repeated in some embodiments.

Fig. 1 is a schematic flowchart of a first intelligent security inspection testing method based on digital twins provided in an embodiment of the present application, and as shown in fig. 1, the method includes the following steps:

step S101: and determining the real-time pushing speed in the testing process according to the image acquisition information of at least one target security inspection machine, and acquiring a testing image from the alternative image set based on the real-time pushing speed.

The alternative image set comprises images acquired through a plurality of preset security inspection machines.

The alternative image set is acquired by simulating the process that the package passes through a preset security inspection machine. In order to carry out intelligent security inspection test more accurately, above-mentioned simulation security inspection process relates to the parcel of different materials, and parcel (first parcel) of a certain amount only contains non-contraband (like books, articles for daily use such as treasured, computer charge), and other parcels (second parcel) contain contraband and are regarded as the non-contraband of interference item. In practical applications, a user usually does not place more contraband in one package, and therefore, the number of the contraband in each second package does not exceed the preset number.

The following is a specific example:

the method comprises the following steps of (1) recording M preset security machines as a preset security machine 1, a preset security machine 2, \8230, a preset security machine 8230, and a preset security machine M;

30000 test packages are preset, wherein the package materials comprise leather packages, cloth packages, plastic packages and the like, and the package types comprise backpacks, handbags, luggage cases and the like; the second parcel accounts for 10% of the parcels, and each second parcel comprises 1-2 contraband articles and some living goods; the quantity ratio of the second package containing 1 contraband to the second package containing 2 contrabands in the embodiment is 7:3;

respectively randomly placing the 30000 packages into a preset security inspection machine for simulation security inspection;

wherein, the preset security inspection machine 1 acquires the images of n1 parcels and records the images as the image 1 ₁ Image 1 ₂ 823060, 8230, image 1 _n1 ；

Presetting an image of n2 parcels collected by a security inspection machine 2, recording the image as an image 2 ₁ Image 2 ₂ 823060, 8230two, image 2 _n2 ；

Presetting an image of n3 parcels collected by a security inspection machine 3, recording the image as an image 3 ₁ Image 3 ₂ 823060, 82303 _n3 ；……；

Presetting an image that a security inspection machine M acquires nM packages and recording the image as an image M ₁ Image M ₂ 823060, 8230min, image M _nM ；

30000 images collected by the M preset security inspection machines form the alternative image set; a test image may be randomly selected from a set of candidate images during the test.

The process of constructing the candidate image set is only an exemplary description, and the number of the preset security inspection machines, the number of the preset packages, the distribution of the articles in the packages, and the like can be set according to the actual application scenario, and will not be described herein again.

In the embodiment, a real-time simulation security check scene which is highly similar to the real security check process needs to be constructed based on digital twins, and based on the real-time simulation security check scene, the image acquisition condition in the real security check process needs to be determined firstly, namely the image acquisition information of the target security check machine is confirmed; and then according to the image acquisition information of at least one target security inspection machine, simulating the real-time pushing speed in the testing process, and acquiring a testing image from the testing image set based on the real-time pushing speed.

The number of the target security inspection machines is not specifically limited in this embodiment. For example, all security machines in an area where the intelligent judgment drawing device needs to be installed can be selected as the target security machine, for example, there are 4 entrances to a subway entrance, and the 4 entrances are respectively provided with one security machine, and the 4 security machines are determined as the target security machines.

Step S102: and respectively pushing the test image to a manual image judging end and an intelligent image judging end based on the real-time pushing speed.

In implementation, in addition to simulating the real-time pushing speed in the test process, the test image needs to be acquired from the test image set based on the real-time pushing speed and pushed to the manual image judging end and the intelligent image judging end to construct a real-time simulation security check scene.

Exemplarily, the manual image judging end is a display device which needs to judge images manually, the manual image judging end is provided with a display screen, a test image is displayed through the display screen, and related personnel manually determine whether the test image has prohibited articles or not and trigger first image judging information;

the intelligent image judging end is an intelligent device provided with an intelligent recognition algorithm, and after receiving the test image, the intelligent image judging end determines whether the test image has prohibited articles or not based on the intelligent recognition algorithm and triggers second image judging information.

Step S103: and determining whether the intelligent image judging end passes the test or not based on first image judging information of the manual image judging end on the test image and second image judging information of the intelligent image judging end on the test image in the test process.

As described above, the manual judging end triggers the first judging graph information (manual judging graph result) through the relevant personnel, and the intelligent judging graph end triggers the second judging graph information (intelligent judging graph result) through the intelligent recognition algorithm; based on the first graph judging information and the second graph judging information of the artificial graph judging end, the security check effect of the intelligent graph judging end compared with the security check effect of the artificial graph judging end can be determined, and therefore whether the intelligent graph judging end passes the test or not is determined.

It can be understood that, in the embodiment, the intelligent graph judging end is tested to actually determine whether an intelligent recognition algorithm (intelligent recognition technology) of the intelligent graph judging end can achieve an expected effect, and then the intelligent graph judging end is put into use.

Fig. 2 is a schematic flowchart of a second digital twin-based intelligent security inspection testing method provided in an embodiment of the present application, and as shown in fig. 2, the method includes the following steps:

step S201: and determining the real-time pushing speed in the testing process according to the image acquisition information of at least one target security inspection machine, and acquiring a testing image from the alternative image set based on the real-time pushing speed.

Step S202: and pushing the test image to a manual image judging end and an intelligent image judging end respectively based on the real-time pushing speed.

The specific implementation manner of steps S201 to S202 may refer to the above embodiments, and will not be described herein.

Step S203: and selecting a target test image from all the test images in the test process based on the judging parameters of the test images.

The image judging parameters of the test images comprise part or all of the total number of the contraband contained in all the test images, the types of the contraband contained in all the test images, the pixel parameters of all the test images, the image judging modes corresponding to all the test images and the information representing the artificial experience corresponding to the artificial image judging end.

In the embodiment, the image judging parameters of the test images represent the identification difficulty of the test images, the identification difficulty of some test images is too simple, the effect on subsequent tests is small, the test images need to be filtered, and only the target test images are reserved, so that the security check effect of the intelligent image judging end compared with the manual image judging end can be more efficiently determined.

For example, since the more the total number of contraband articles, the greater the identification difficulty, the chart judging parameter may include the total number of contraband articles contained in all the test images;

since the more types of contraband, the greater the identification difficulty, the chart judging parameters may include the types of contraband contained in all the test images;

since the lower the pixels of the test image (the smaller the resolution), the greater the recognition difficulty, the image judgment parameters may include pixel parameters of the test image;

in some optional embodiments, each test image includes a main view and a side view, and a judging mode can be determined according to the real-time pushing speed (the single-view judging mode only pushes the main view, and the double-view judging mode pushes the main view and the side view); in the graph judging process, the main view is mainly referred to, but the side view can play a role of assisting in judging the graph, and the single-view graph judging mode is more difficult to identify compared with the double-view graph judging mode, so that the graph judging parameters can comprise the graph judging mode of the test image;

the image judging parameter may include information representing artificial experience corresponding to the manual image judging end, because the longer the experience of the relevant person corresponding to the manual image judging end is, the greater the recognition difficulty is.

Step S204: and determining an image security check result of the target test image aiming at any target test image.

The image security check result is obtained by comparing the judgment image information of the target test image with the actual forbidden information of the target test image; the image judging information comprises first image judging information and second image judging information of the target test image; the image security inspection result comprises part or all of positive inspection, negative inspection, false inspection and missing inspection.

By comparing the judgment information of the target test image with the actual contraband information of the target test image, the following four cases may occur:

1) The actual contraband information of the target test image represents that contraband exists (namely the target test image contains the contraband), the judgment chart information also represents that the contraband exists, and the image security inspection result of the target test image is positive inspection;

2) The actual contraband information of the target test image represents that no contraband exists (namely, no contraband exists in the target test image), the judgment image information also represents that no contraband exists, and the image security inspection result of the target test image is negative inspection;

3) The actual contraband information of the target test image represents that no contraband exists (namely, no contraband exists in the target test image), but the judgment image information represents that the contraband exists, and the image security inspection result of the target test image is false inspection;

4) The actual contraband information of the target test image represents that contraband exists (namely, the target test image contains the contraband), but the judgment map information represents that the contraband does not exist, and the image security inspection result of the target test image is missing inspection.

Step S205: and determining whether the intelligent image judging end passes the test or not based on the image security inspection results of all the target test images.

In this embodiment, the image security check result represents a graph judgment effect, and based on the image security check results of all target test images, whether an intelligent graph judgment end passes the test or not can be accurately determined.

According to the scheme, the image judging parameters of the test images represent the identification difficulty of the test images, the identification difficulty of some test images is too simple, the effect on subsequent tests is small, the test images need to be filtered, and the security check effect of the intelligent image judging end compared with the manual image judging end can be more efficiently determined only by reserving the target test images; and aiming at each target test image, comparing the judgment image information with the actual illegal information to determine the image security check result representing the security check effect, so that whether the intelligent judgment image end passes the test or not can be accurately determined based on the image security check results of all the target test images.

Referring to fig. 3, in some alternative embodiments, the target test image may be determined by, but is not limited to:

step S301: and dividing all the test images into a plurality of test image sets through the decision tree model.

And the judging parameters of all the test images in any test image set are the same.

Exemplarily, collecting and counting specific judging graph parameters of each test image by taking the judging graph parameters as the characteristic attributes of a decision tree; a Gini function is adopted as a splitting attribute of the decision tree, namely, the splitting of child nodes of the decision tree is realized through the Gini function until all the nodes are leaf nodes, and a plurality of test image sets are obtained; the characteristic attributes of the test image sets are the same, that is, the judging parameters of all the test images in any test image set are the same.

The above-mentioned kini function is:

d is an image set of the same type of characteristic attributes in all test images; p _k The ratio of the number of the kth identical characteristic attribute in the same type of characteristic attributes in the test image to the total number of the type of characteristic attributes; y is the number of different classes in the same class of feature attributes in the parameter set. The smaller Gini (D) the higher the purity of the data set representing the same class of feature attributes in the parameter set.

Explaining how to divide all the test images into a plurality of test image sets by a decision tree model by using a specific embodiment, specifically, the judging parameters of the test images comprise the total number of contraband contained in all the test images, the types of the contraband contained in all the test images, the pixel parameters of all the test images, the judging mode corresponding to all the test images and all the information representing the artificial experience corresponding to the artificial judging end, namely 5 types of characteristic attributes; and respectively calculating a Gini value (D) of each characteristic attribute through a Gini function, sorting the characteristic attributes of each type from small to large according to the calculated Gini value (D), wherein the characteristic attribute with the minimum Gini value (D) is used as a root node of a decision tree (a first characteristic attribute of decision tree splitting), and further obtaining a second characteristic attribute, a third characteristic attribute, a fourth characteristic attribute and a fifth characteristic attribute according to the sorting. When all test images are divided into a plurality of test image sets through the decision tree model, firstly, selecting a first characteristic attribute as a first standard of the decision tree model for dividing all test images into the test image sets to perform primary classification, wherein the first characteristic attribute is the types of contraband contained in all test images as shown in table 1; then, selecting a second characteristic attribute as a second standard of the decision tree model for dividing all the test images into the test image set for primary classification, wherein the second characteristic attribute is the total number of contraband contained in all the test images as shown in table 1; further, the third characteristic attribute to the fifth characteristic attribute are respectively pixel parameters of each test image, a judgment mode corresponding to each test image, and information representing artificial experience corresponding to an artificial judgment end.

Step S302: and aiming at any test image set, determining the identification difficulty information of the test image set based on the figure judging parameters of the test images in the test image set.

As described above, the graph judging parameters of the test images represent the identification difficulty of the test images, and the graph judging parameters of all the test images in the same test image set are the same, so that the identification difficulty information of the test image set can be determined based on the graph judging parameters of the test images in the test image set.

Step S303: and determining whether the test image in the test image set is a target test image or not based on the identification difficulty information of the test image set.

According to the scheme, all the test images are divided into a plurality of test image sets through the decision tree model, the image judging parameters of all the test images in the same test image set are the same, and the image judging parameters of the test images represent the identification difficulty of the test images, so that the identification difficulty information representing the identification difficulty of the images in the test image sets can be determined based on the image judging parameters of the test images in the test image sets; and accurately determining whether the test image in each test image set is a target test image or not based on the identification difficulty information.

In some optional embodiments, the identification difficulty information includes an identification difficulty level, and correspondingly, the step S303 may be implemented by, but is not limited to, the following manners:

For example, referring to table 1, for any test image set, various image judgment parameters of the test image set are compared with corresponding preset parameter thresholds to determine difficulty information of the various image judgment parameters:

TABLE 1

Comparison result of figure judgment parameters and corresponding preset parameter threshold values	Difficulty and ease information
		Kinds of contraband are more than or equal to 280	Difficulty in
Total number of contraband articles is greater than or equal to 3000	Difficulty in
		Pixel parameters of the test image are less than or equal to 100 x 100	Difficulty in
Judging mode = single view angle mode corresponding to test image	Difficulty in
		The information representing the artificial experience corresponding to the artificial graph judging end is more than or equal to 3 years	Difficulty in

It can be understood that when various graph judgment parameters do not belong to the conditions of table 1, the difficulty information is easy.

Several specific examples are described below:

1) And if the difficulty information of 4-5 types of judging graph parameters of the test image set is difficult, the difficulty level is 1 level.

For example, the type of contraband is 290, the total number of contraband is 3000, the pixel parameter (resolution) of the test image is 100 × 100, the judgment mode corresponding to the test image is a single-view mode, the information representing the artificial experience corresponding to the manual judgment end is 4 years, the difficulty information of all the judgment parameters (5 types) is difficult, and the difficulty level is 1 level.

2) And if the difficulty information of the 2-3 types of judging parameters of the test image set is difficult, the difficulty level is 2.

For example, the type of contraband is 100, the total number of contraband is 3000, the resolution of the test image is 200 × 100, the graph judging mode corresponding to the test image is a single-view mode, the information representing the artificial experience corresponding to the manual graph judging end is 1 year, the difficulty information with 2 types of graph judging parameters is difficult, and the difficulty level is 2.

3) And if the difficulty information of the 0-1 type figure judging parameters of the test image set is difficult, the difficulty level is 3.

For example, the type of contraband is 100, the total number of contraband is 2700, the resolution of the test image is 200 × 200, the graph judging mode corresponding to the test image is a dual-view mode, the information representing the artificial experience corresponding to the manual graph judging end is 1 year, the difficulty information of all graph judging parameters is easy, and the difficulty level is 3 levels.

The above difficulty levels are merely exemplary and the application is not limited thereto.

In some optional embodiments, the step S205 may be implemented by, but is not limited to, the following manners:

aiming at any test item, comparing the test parameters of the artificial image judging end on the test item with the test parameters of the intelligent image judging end on the test item, and determining the score corresponding to the test parameter comparison result of the test item;

Exemplarily, the image security inspection result of each target test image corresponding to the manual image judgment end is any one of positive inspection, negative inspection, false inspection and missing inspection, the number (positive inspection number) of the target test images with the positive inspection result of the image security inspection result at the manual image judgment end is determined, the number (negative inspection number) of the target test images with the negative inspection result of the image security inspection result at the manual image judgment end is determined, the number (false inspection number) of the target test images with the false inspection result at the manual image judgment end is determined, and the number (missing inspection number) of the target test images with the missing inspection result at the manual image judgment end is determined;

similarly, determining the positive detection number, the negative detection number, the false detection number and the missed detection number of the intelligent image judging end;

in the embodiment, three test items of accuracy, false alarm rate and false negative rate are provided; wherein:

the accuracy rate = (positive detection number + negative detection number)/(positive detection number + missing detection number + false detection number + negative detection number) = 100%;

the false alarm rate = 1-positive detection number/(positive detection number + false detection number) × 100%;

the false negative rate = 1-positive detection number/(positive detection number + false negative number) × 100%;

calculating a test parameter (recorded as a first accuracy) of the manual graph judging end at the accuracy, a test parameter (recorded as a first false alarm rate) of the manual graph judging end at the false alarm rate, and a test parameter (recorded as a first false alarm rate) of the manual graph judging end at the false alarm rate;

calculating a test parameter (recorded as a second accuracy) of the intelligent graph judging end at the accuracy, a test parameter (recorded as a second false alarm rate) of the intelligent graph judging end at the false alarm rate, and a test parameter (recorded as a second false alarm rate) of the intelligent graph judging end at the false alarm rate;

respectively comparing the first accuracy with the second accuracy, comparing the first false alarm rate with the second false alarm rate, and comparing the first false alarm rate with the second false alarm rate;

if the first accuracy rate is smaller than the second accuracy rate, the score corresponding to the accuracy rate is 7, otherwise, the score corresponding to the accuracy rate is 0;

if the first false alarm rate is larger than the second false alarm rate, the value corresponding to the false alarm rate is 1, otherwise, the value corresponding to the false alarm rate is 0;

if the first missing report rate is larger than the second missing report rate, the score corresponding to the missing report rate is 2, otherwise, the score corresponding to the missing report rate is 0;

and comparing the total score obtained by adding the scores corresponding to the three test items with a preset score, and if the total score is greater than the preset score, determining that the intelligent graph judging end passes the test.

The above test procedures are only exemplary, and the number of test items and how to evaluate each test item are not particularly limited in the present application.

Fig. 4 is a schematic flowchart of a third digital twin-based intelligent security inspection testing method provided in an embodiment of the present application, and as shown in fig. 4, the method includes the following steps:

step S401: and determining the real-time pushing speed in the testing process according to the image acquisition information of at least one target security inspection machine, and acquiring a testing image from the alternative image set based on the real-time pushing speed.

The specific implementation manner of step S401 may refer to the above embodiments, and is not described herein again.

Step S402: and adjusting the pixels of the test images based on the acquisition pixels of the target security inspection machine, and filtering the test images after the pixels are adjusted.

For example, when constructing the candidate image set, a plurality of preset security check machines are used, and in order to make the simulation scene closer to the real security check scene, the target security check machine used in the actual security check adjusts the pixels of each test image based on the collected pixels of the target security check machine;

in addition, noise may be generated when adjusting pixels of an image; based on this, the present embodiment also performs filtering processing on the test image after the pixel adjustment.

Step S403: and pushing the test image subjected to filtering processing to an artificial image judgment end and an intelligent image judgment end respectively based on the real-time pushing speed.

Step S404: and determining whether the intelligent image judging end passes the test or not based on first image judging information of the manual image judging end on the test image and second image judging information of the intelligent image judging end on the test image in the test process.

The specific implementation manner of steps S403-S404 may refer to the above embodiments, and will not be described herein.

According to the scheme, the pixels of the test images are adjusted based on the collected pixels of the target security inspection machine, so that the simulation scene is closer to the real security inspection scene; by performing filtering processing on the test image after the pixels are adjusted, noise generated when the pixels of the image are adjusted is reduced.

In some optional embodiments, if there are a plurality of target security inspection machines, adjusting the pixels of each test image may be implemented by, but not limited to:

1) Aiming at any test image, adjusting the original pixel of the test image to be the acquisition pixel of any target security inspection machine.

Illustratively, there are 2 target security inspection machines, the resolution of the image output by the target security inspection machine 1 is 100 × 100, and the resolution of the image output by the target security inspection machine 2 is 500 × 600; the resolution of the test images was randomly adjusted to 100 x 100 or 500 x 600.

2) And aiming at any test image, selecting a target acquisition pixel closest to an original pixel of the test image from acquisition pixels of all target security inspection machines, and adjusting the original pixel of the test image to the target acquisition pixel.

Illustratively, there are 2 target security inspection machines, the resolution of the image output by the target security inspection machine 1 is 100 × 100, and the resolution of the image output by the target security inspection machine 2 is 500 × 600; the resolution of the test image is 150 × 150, the resolution of the image output by the target security inspection machine 1 is closest to the resolution of the test image, and the resolution of the test image is adjusted to 100 × 100.

Of course, if there is only one target security inspection machine, the original pixels of all the test images can be directly adjusted to the collected pixels of the target security inspection machine.

In this embodiment, the pixel adjustment process is not specifically limited, for example, a pixel adjustment operator in opencv (a visual processing system) is adopted, where the operator code is: rimg = cv2.Resize (img, (100,100)); wherein img is an input test image; (100 ) adjusting the resolution of the test image after the pixels; rimg is the test image after adjusting the pixels.

In some optional embodiments, the filtering process performed on the test image after adjusting the pixels may be implemented by, but not limited to, the following ways:

In this embodiment, the gaussian kernel used in the gaussian filtering method may be selected according to an actual application scenario, for example, the gaussian kernel is 5 × 5,3 × 3, and so on.

In this embodiment, the filtering process is not specifically limited, for example, a gaussian filter operator in opencv is used, where the operator code is: gauss = cv2.Gaussianblur (image, (5, 5), 0); wherein, the Image is a test Image after adjusting the pixels; (5, 5) a Gaussian kernel representing the filtering process; 0 is the default value of the boundary sample; gauss is the filtered test image.

According to the scheme, the Gaussian filtering method is adopted to perform noise reduction processing on the test image generating the noise, and the definition of the subsequently pushed test image is ensured.

As described above, in some alternative embodiments, each test image includes a front view and a side view.

Correspondingly, the embodiment of the present application provides a schematic flow chart of a fourth intelligent security inspection testing method based on digital twins, as shown in fig. 5, including the following steps:

step S501: and determining the real-time pushing speed in the testing process according to the image acquisition information of at least one target security inspection machine, and acquiring a testing image from the alternative image set based on the real-time pushing speed.

Wherein the candidate image set comprises images acquired by a plurality of preset security inspection machines.

The specific implementation manner of step S501 may refer to the above embodiments, and is not described herein again.

Step S502: if the real-time pushing speed is greater than a preset speed, respectively pushing a main view in the test image to the manual image judging end and the intelligent image judging end; otherwise, respectively pushing the front view and the side view in the test image to the manual image judging end and the intelligent image judging end.

As described above, each test image includes a front view and a side view, and a judging mode can be determined according to the real-time pushing speed (the single-view judging mode only pushes the front view, and the double-view judging mode pushes the front view and the side view); in the process of judging the graph, the main view is mainly referred to, but the side view can play a role in assisting the judgment of the graph.

In this embodiment, if the real-time pushing speed is greater than the preset speed, it indicates that the image pushing speed is high, the processing pressures of the manual graph judging end and the intelligent graph judging end are high, and a user related to the manual graph judging end may not be able to synchronously refer to the side view, and based on this, only the main view in the test image is pushed to the manual graph judging end and the intelligent graph judging end;

if the real-time pushing speed is smaller than or equal to the preset speed, the image pushing speed is slow, the processing pressure of the manual graph judging end and the intelligent graph judging end is low, and based on the situation, the main view and the side view in the test image are pushed to the manual graph judging end and the intelligent graph judging end.

Step S503: and determining whether the intelligent image judging end passes the test or not based on first image judging information of the manual image judging end on the test image and second image judging information of the intelligent image judging end on the test image in the test process.

The specific implementation manner of step S503 can refer to the above embodiments, and is not described herein again.

According to the scheme, the real-time pushing speed is compared with the preset speed, so that the image pushing speed is determined, and the image judging mode suitable for each test moment is accurately determined.

Fig. 6 is a schematic flowchart of a fifth intelligent security inspection testing method based on digital twins according to an embodiment of the present application, and as shown in fig. 6, the method includes the following steps:

step S601: and aiming at any test time in the test process, determining the corresponding time of the test time in a preset time period.

In this embodiment, in order to make the simulation scene closer to the real security check scene, the test process is associated with different times of the preset time period, and the image acquisition speed of the target security check machine at the corresponding time is simulated.

Illustratively, the preset period is a day of history, and the test procedure is a day after the preset period. The relative time of the test time in the test process is the same as the relative time of the corresponding time in the preset time period. As 8 in the test procedure: 10, corresponding to 8:10.

step S602: and determining the real-time pushing speed of the test moment based on the image acquisition speed of the at least one target security inspection machine at the corresponding moment in the preset time period.

As described above, this embodiment needs to simulate the image capturing speed of the target security inspection machine at the corresponding time. Exemplarily, if 1 target security check machine exists, determining the image acquisition speed of the target security check machine at the corresponding moment as the real-time pushing speed at the test moment; and if a plurality of target security inspection machines exist, determining the real-time pushing speed of the test time according to the sum of the image acquisition speeds of all the target security inspection machines at the corresponding time (namely the image judging speed in the actual security inspection process).

For example, the real-time push speed at each test time in the test process can be referred to as fig. 7. In the example, the target security inspection machine is arranged at the entrance of the subway entrance, the subway entrance is not opened in 0-4h and 23-24h, and no package passes through the target security inspection machine, so that the image acquisition speed of the target security inspection machine in the two time periods is 0, and correspondingly, the real-time pushing speed is 0;6-9h are 'early peak' time periods, 18-21h are 'late peak' time periods, and the two time periods are wrapped more by the target security inspection machine, so that the image acquisition speed of the target security inspection machine in the two time periods is higher, and the corresponding real-time pushing speed is also higher.

Step S603: and acquiring a test image from the alternative image set based on the real-time pushing speed.

Step S604: and respectively pushing the test image to a manual image judging end and an intelligent image judging end based on the real-time pushing speed.

Step S605: and determining whether the intelligent image judging end passes the test or not based on first image judging information of the manual image judging end on the test image and second image judging information of the intelligent image judging end on the test image in the test process.

The specific implementation manner of steps S603-S605 can refer to the above embodiments, and is not described herein again.

According to the scheme, the test process is associated with different moments of the preset time period, and the image acquisition speed of the target security inspection machine at the corresponding moment is simulated, so that the simulated scene is closer to the real security inspection scene.

Based on the same inventive concept, the embodiment of the present application provides an intelligent security inspection testing device based on digital twins, and referring to fig. 8, the intelligent security inspection testing device 800 includes:

the speed determining module 801 is configured to determine a real-time pushing speed in a testing process according to image acquisition information of at least one target security inspection machine, and acquire a test image from an alternative image set based on the real-time pushing speed; the alternative image set comprises images collected by a plurality of preset security inspection machines;

a pushing module 802, configured to push the test image to a manual image judging end and an intelligent image judging end respectively based on the real-time pushing speed;

the testing module 803 is configured to determine whether the intelligent image judging end passes the test based on first image judging information of the manual image judging end on the test image and second image judging information of the intelligent image judging end on the test image in a testing process.

In some optional embodiments, the testing module 803 is specifically configured to:

selecting a target test image from all test images in the test process based on the image judging parameters of the test images; the image judging parameters of the test images comprise the total number of contraband articles contained in all the test images, the types of the contraband articles contained in all the test images, the pixel parameters of all the test images, image judging modes corresponding to all the test images and part or all of information representing artificial experience corresponding to an artificial image judging end;

aiming at any target test image, determining an image security check result of the target test image; the image security check result is obtained by comparing the judgment image information of the target test image with the actual forbidden information of the target test image; the image judging information comprises first image judging information and second image judging information of the target test image; the image security inspection result comprises part or all of positive inspection, negative inspection, false inspection and missing inspection;

In some optional embodiments, the identification difficulty information includes an identification difficulty level, and the testing module 803 is specifically configured to:

if the identification difficulty level is less than or equal to a preset level, determining all the test images in the test image set as target test images; the identification difficulty level of the test image set is smaller when the identification difficulty of the judgment parameter representation of the test image in the test image set is larger.

In some optional embodiments, the test module 803 is specifically configured to:

In some optional embodiments, the image adjusting module 804 is further included to:

before the pushing module 802 pushes the test images to the manual image judging end and the intelligent image judging end respectively based on the real-time pushing speed, pixels of the test images are adjusted based on the collected pixels of the target security inspection machine, and the test images after the pixels are adjusted are filtered.

In some optional embodiments, if there are multiple target security inspection machines, the image adjusting module 804 is specifically configured to:

aiming at any test image, adjusting an original pixel of the test image to be an acquisition pixel of any target security inspection machine; or

In some optional embodiments, the image adjusting module 804 is specifically configured to:

In some alternative embodiments, each test image includes a front view and a side view; the pushing module 802 is specifically configured to:

if the real-time pushing speed is greater than a preset speed, respectively pushing a main view in the test image to the manual image judging end and the intelligent image judging end;

In some alternative embodiments, the speed determining module 801 is specifically configured to:

Since the apparatus is the apparatus in the method in the embodiment of the present application, and the principle of the apparatus for solving the problem is similar to that of the method, the implementation of the apparatus may refer to the implementation of the method, and repeated details are not repeated.

Based on the same technical concept, an electronic device 900 is further provided in the embodiments of the present application, as shown in fig. 9, and includes at least one processor 901 and a memory 902 connected to the at least one processor, and a specific connection medium between the processor 901 and the memory 902 is not limited in the embodiments of the present application, and the processor 901 and the memory 902 are connected through a bus 903 in fig. 9 as an example. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of illustration, only one thick line is shown in FIG. 9, but this does not indicate only one bus or one type of bus.

The processor 901 is a control center of the electronic device, and may be connected to various parts of the electronic device by using various interfaces and lines, and implement data processing by executing or executing instructions stored in the memory 902 and calling data stored in the memory 902. Optionally, the processor 901 may include one or more processing units, and the processor 901 may integrate an application processor and a modem processor, where the application processor mainly processes an operating system, a user interface, an application program, and the like, and the modem processor mainly processes an issued instruction. It will be appreciated that the modem processor described above may not be integrated into the processor 901. In some embodiments, the processor 901 and the memory 902 may be implemented on the same chip, or in some embodiments, they may be implemented separately on separate chips.

The processor 901 may be a general-purpose processor, such as a Central Processing Unit (CPU), a digital signal processor, an Application Specific Integrated Circuit (ASIC), a field programmable gate array or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or the like, and may implement or perform the methods, steps, and logic blocks disclosed in the embodiments of the present Application. The general purpose processor may be a microprocessor or any conventional processor or the like. The steps of the method disclosed in connection with the embodiment of the intelligent security inspection testing method based on the digital twin can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules in the processor.

The memory 902, which is a non-volatile computer-readable storage medium, may be used to store non-volatile software programs, non-volatile computer-executable programs, and modules. The Memory 902 may include at least one type of storage medium, and may include, for example, a flash Memory, a hard disk, a multimedia card, a card-type Memory, a Random Access Memory (RAM), a Static Random Access Memory (SRAM), a Programmable Read Only Memory (PROM), a Read Only Memory (ROM), a charge Erasable Programmable Read Only Memory (EEPROM), a magnetic Memory, a magnetic disk, an optical disk, and so on. The memory 902 is any other medium that can be used to carry or store desired program code in the form of instructions or data structures and that can be accessed by a computer, but is not limited to such. The memory 902 of the embodiments of the present application may also be circuitry or any other device capable of performing a storage function for storing program instructions and/or data.

In the embodiment of the present application, the memory 902 stores a computer program, and when the program is executed by the processor 901, the processor 901 is caused to execute:

pushing the test image to an artificial image judging end and an intelligent image judging end respectively based on the real-time pushing speed;

In some alternative embodiments, the processor 901 specifically executes:

aiming at any target test image, determining an image security check result of the target test image; the image security check result is obtained by comparing the judgment image information of the target test image with the actual forbidden information of the target test image; the judging image information comprises first judging image information and second judging image information of the target test image; the image security inspection result comprises part or all of positive inspection, negative inspection, false inspection and missing inspection;

In some alternative embodiments, the processor 901 specifically executes:

In some optional embodiments, the identification difficulty information includes an identification difficulty level, and the processor 901 specifically performs:

In some alternative embodiments, the processor 901 specifically executes:

In some optional embodiments, before pushing the test image to the manual graph judging end and the intelligent graph judging end respectively based on the real-time pushing speed, the processor 901 further performs:

In some optional embodiments, if there are multiple target security inspection machines, the processor 901 specifically executes:

In some alternative embodiments, the processor 901 specifically executes:

In some alternative embodiments, each test image includes a front view and a side view; the processor 901 specifically executes:

In some optional embodiments, the processor 901 specifically executes:

Since the electronic device is the electronic device in the method in the embodiment of the present application, and the principle of the electronic device for solving the problem is similar to that of the method, reference may be made to implementation of the method for the electronic device, and repeated details are not described again.

Based on the same technical concept, embodiments of the present application also provide a computer-readable storage medium storing a computer program executable by an electronic device, wherein when the program runs on the electronic device, the electronic device is enabled to execute the steps of the above intelligent security inspection testing method based on digital twin.

As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and so forth) having computer-usable program code embodied therein.

The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

While the preferred embodiments of the present application have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all alterations and modifications as fall within the scope of the application.

It will be apparent to those skilled in the art that various changes and modifications may be made in the present application without departing from the spirit and scope of the application. Thus, if such modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is intended to include such modifications and variations as well.

Claims

1. An intelligent security inspection testing method based on digital twins is characterized by comprising the following steps:

2. The method of claim 1, wherein determining whether the intelligent graph judging end passes the test based on first graph judging information of the manual graph judging end on the test image and second graph judging information of the intelligent graph judging end on the test image in a test process comprises:

3. The method of claim 2, wherein selecting a target test image from all test images of the test procedure based on the mapping parameters of the test image comprises:

aiming at any test image set, determining identification difficulty information of the test image set based on image judging parameters of test images in the test image set;

4. The method of claim 3, wherein the identification difficulty information includes an identification difficulty level, and determining whether a test image of the set of test images is a target test image based on the identification difficulty information of the set of test images comprises:

5. The method of claim 2, wherein determining whether the intelligent image terminal passes the test based on the image security check results of all target test images comprises:

6. The method of claim 1, before pushing the test image to a manual graph judging end and an intelligent graph judging end respectively based on the real-time pushing speed, further comprising:

7. The method of claim 6, wherein if there are multiple target security machines, adjusting pixels of each test image based on collected pixels of the target security machines comprises:

And aiming at any test image, selecting a target acquisition pixel closest to the original pixel of the test image from the acquisition pixels of all target security machines, and adjusting the original pixel of the test image to be the target acquisition pixel.

8. The method of claim 6, wherein filtering the test image after adjusting the pixels comprises:

and filtering the test image with the pixels adjusted by a Gaussian filtering method.

9. The method of claim 1, wherein each test image comprises a front view and a side view; based on the real-time pushing speed, the test image is respectively pushed to an artificial image judging end and an intelligent image judging end, and the method comprises the following steps:

otherwise, respectively pushing the main view and the side view in the test image to the manual image judging end and the intelligent image judging end.

10. The method of any one of claims 1 to 9, wherein determining the real-time pushing speed in the test process according to the image acquisition information of at least one target security inspection machine comprises: