Security inspection method, security inspection apparatus and security inspection system

Abstract

A security inspection method, comprising: on the basis of object features of an object under inspection (3), determining a target category of the object under inspection (3) (101); on the basis of the target category, calling at least one target algorithm in an algorithm set, wherein the target algorithm is determined on the basis of the target category, and the algorithm set is integrated into a same security inspection device (102); and by means of the at least one target algorithm, recognizing a scanned image of the object under inspection (3), so as to generate a recognition result (103). Further provided are a security inspection apparatus and a security inspection system. A security inspection device (2) can inspect various different objects under inspection (3) in a targeted manner, thereby facilitating an improvement in the security inspection efficiency of the security inspection device (2), and reducing the overall space occupied by the security inspection device (2).

Description

Security inspection method, security inspection device and security inspection system

[0001] This application claims priority to Chinese Patent Application No. 202410889083.8, filed on July 3, 2024, the contents of which are incorporated herein by reference in their entirety. TECHNICAL FIELD

[0002] The present application relates to the technical field of detection, and in particular to a security inspection method, a security inspection device and a security inspection system. BACKGROUND

[0003] The security inspection equipment generates a scanning image by scanning luggage, and further identifies the scanning image by a security inspection staff or a picture judging station to realize checking the luggage without opening the bag.

[0004] The luggage is usually placed on a tray and conveyed to the security inspection equipment along with a conveying line. Since the sizes of the luggage have differences, in the prior art, multiple security inspection equipment are usually deployed at the security inspection site, and each security inspection equipment is arranged with an identification algorithm. The luggage is sorted by a sorting system, and different types of luggage, such as large trays, standard trays and naked bags, are sorted to the security inspection equipment installed with the corresponding identification algorithm for security inspection. This security inspection method needs to deploy multiple security inspection equipment, occupies a large area, has high maintenance cost in the later period, and has poor expansibility. For example, if the luggage processing system is upgraded in the later period, new security inspection equipment and the luggage conveying line matched therewith need to be added. Or, a large-size identification algorithm capable of identifying multiple types of luggage is configured to the single security inspection equipment, and the large-size identification algorithm is used to detect any luggage, which causes large server computing pressure and low detection efficiency. SUMMARY

[0005] Therefore, the present application provides a security inspection method, a security inspection device and a security inspection system to solve the problem that the existing security inspection method needs to detect different types of luggage by multiple security inspection equipment or a single security inspection equipment uses the same identification algorithm to detect different types of luggage, which causes large space occupation of the security inspection equipment and poor detection efficiency.

[0006] In a first aspect, an embodiment of the present application provides a security inspection method, comprising:

[0007] determining a target category of the object to be inspected according to an object feature of the object to be inspected;

[0008] calling at least one target algorithm in an algorithm set according to the target category, the at least one target algorithm being determined according to the target category, and the algorithm set being integrated in the same security inspection equipment;

[0009] identifying a scanning image of the object to be inspected by the at least one target algorithm to generate an identification result.

[0010] According to an embodiment of the present application, the object feature of the object under inspection is obtained before determining the target category of the object under inspection according to the object feature of the object under inspection.

[0011] According to an embodiment of the present application, the object feature comprises at least one of a label feature, an image feature, and a size feature.

[0012] According to an embodiment of the present application, the scan image of the object under inspection is obtained before identifying the scan image of the object under inspection by the at least one target algorithm.

[0013] According to an embodiment of the present application, the target category of the object under inspection is determined according to the object feature of the object under inspection before

[0014] A category set is generated, and the category set comprises a plurality of target categories.

[0015] The algorithm set is generated, and the algorithm set comprises a plurality of target algorithms; and

[0016] An association relationship between the plurality of target categories and the plurality of target algorithms is established.

[0017] According to an embodiment of the present application, further comprising: calling at least one target algorithm in the algorithm set according to a preset security inspection instruction and the target category.

[0018] According to an embodiment of the present application, the identification result is generated by identifying the scan image of the object under inspection by the at least one target algorithm, comprising:

[0019] The scan image of the object under inspection is identified by the at least one target algorithm to generate an identification sub-result associated with each target algorithm.

[0020] The identification result is generated according to the identification sub-result.

[0021] According to an embodiment of the present application, each target algorithm in the algorithm set has an independent interactive interface.

[0022] According to an embodiment of the present application, after the identification result is generated, the object under inspection is sorted according to the identification result.

[0023] In a second aspect, an embodiment of the present application provides a security inspection device, comprising:

[0024] A conveying line body for conveying an object under inspection;

[0025] A security inspection device arranged in the conveying direction of the conveying line body, and the security inspection device is configured to execute the security inspection method.

[0026] In a third aspect, the embodiments of the present application provide a security inspection system, comprising the security inspection device.

[0027] According to the security inspection method, the security inspection device and the security inspection system provided by the embodiments of the present application, the following technical effects can be achieved: after the target category of the object is determined according to the object features, at least one of the multiple target algorithms is called by the security inspection equipment according to the target category, the scanning image of the object is identified by the at least one target algorithm, and the identification result is generated, so that one security inspection equipment can detect various different objects in a targeted manner, which is beneficial to improving the security inspection efficiency of the security inspection equipment and reducing the overall space occupied by the security inspection equipment. BRIEF DESCRIPTION OF DRAWINGS

[0028] The above and other objects, features and advantages of the present application will become more apparent from the following description of the embodiments of the present application taken with reference to the accompanying drawings, in which:

[0029] FIG. 1 schematically shows a flowchart of a security inspection method according to an embodiment of the present application;

[0030] FIG. 2 schematically shows a structural diagram of a security inspection device according to an embodiment of the present application;

[0031] Reference signs: 1: conveying line body; 2: security inspection equipment; 3: object to be inspected. DETAILED DESCRIPTION

[0032] In order to make the objects, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the embodiments and the accompanying drawings. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0033] The terms used herein are only used to describe specific embodiments, and are not intended to limit the present application. The terms "comprise", "contain" and the like used herein indicate the existence of the described features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0034] In the description of the present application, it should be noted that unless specifically defined and limited, the terms "mounting", "connection", "connecting" should be understood broadly, for example, it can be fixed connection, or detachable connection, or integrally connected; it can be mechanical connection, or electrical connection; it can be directly connected, or indirectly connected through intermediate medium, or the communication inside two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0035] In the prior art, the luggage of passengers needs to be strictly checked at the security site, especially at the airport. Due to the difference in the size of the luggage, the luggage is usually divided into small items, medium items and large items according to the size of the luggage. The small items can not be placed in the tray, the medium items are placed in the standard size tray, and the large items are placed in the oversized tray. Different identification algorithms are usually used to identify the items to identify whether the items contain explosives, contraband, controlled knives, guns, corrosive liquids and other suspicious substances.

[0036] For different size specifications and different types of items, multiple security devices are usually deployed at the security site, and each security device deploys an identification algorithm. For example, a first security device, a second security device and a third security device are arranged at the security site. A plurality of items conveyed by the conveying line are respectively shunted to the first security device, the second security device and the third security device according to the size specification, type and security requirement of the items. The first security device uses a first identification algorithm to identify small items, the second security device uses a second identification algorithm to identify medium items, and the third security device uses a third identification algorithm to identify large items.

[0037] It can be understood that the operation amount of the first identification algorithm for small items is small, the operation amount of the second identification algorithm is greater than that of the first identification algorithm, and the operation amount of the third identification algorithm is greater than that of the second identification algorithm.

[0038] This security method needs to arrange multiple security devices at the security site, which occupies a large space and has high maintenance and repair costs in the later period. In addition, the expansibility is poor. For example, if the luggage system is upgraded later, the type of tray needs to be increased, which requires new security devices and supporting luggage conveying lines. Or a single security device is configured with a large size identification algorithm capable of identifying multiple types of luggage. Any luggage is detected by using the large size identification algorithm, which causes large server computing pressure and low detection efficiency.

[0039] To solve the above problems, the application provides a security check method, which deploys multiple target algorithms on a security check device, calls at least one target algorithm in the multiple target algorithms according to the target category of the detected object, identifies the scan image of the detected object, generates an identification result, realizes that one security check device detects various different detected objects, is beneficial to improving the security check efficiency of the security check device, and is beneficial to reducing the overall space occupied by the security check device.

[0040] The security check method of the embodiment of the application is described below in combination with FIG. 1.

[0041] As shown in FIG. 1, the security check method provided by the embodiment of the application includes:

[0042] Step 101: determining the target category of the detected object according to the object feature of the detected object;

[0043] Step 102: calling at least one target algorithm in an algorithm set according to the target category, the at least one target algorithm being determined according to the target category, the algorithm set being integrated in the same security check device;

[0044] Step 103: identifying the scan image of the detected object through the at least one target algorithm, and generating an identification result.

[0045] Specifically, before the detected object moves to the security check device along the conveying line, the target category of the detected object is acquired, and the target category of the detected object is determined according to the object feature of the detected object. The object feature includes one or more combinations of a label feature, an image feature, and a size feature.

[0046] For example, the detected object is classified according to the size feature of the detected object. According to the size specification, the to-be-checked object is divided into a first size category, a second size category, and a third size category, etc. The first size category corresponds to a small-size detected object, which can not be placed in a tray. The second size category corresponds to a medium-size detected object, which is placed in a standard-size tray. The size specification of the standard-size tray can be 1100mm*1100mm. The third size category corresponds to a large-size detected object, which is placed in a super-large-size tray. The size specification of the super-large-size tray can be 1100mm*1500mm. It can be understood that the object can also be divided into more than three size categories according to the size specification of the object.

[0047] According to the image features of the detected object, the detected object is classified, for example, according to the geometric features of the objects in the image, the objects are classified, objects similar in shape to powder are classified into a first image category, objects similar in shape to knives are classified into a second image category, and objects similar in shape to guns are classified into a third image category. It can be understood that the objects can also be divided into more than three image categories according to the geometric features of the objects in the image. Objects similar in shape to knives and objects similar in shape to guns can also be classified into the same image category, for example, objects similar in shape to knives and objects similar in shape to guns are classified into the second image category.

[0048] According to the label features of the detected object, the detected object is classified, for example, passengers fill in the object information table according to the types of objects they carry, the types of objects can be daily necessities, cosmetics, electronic devices, etc., a two-dimensional code label is generated from the object information table, and the two-dimensional code label is attached to the outside of the detected object. Cosmetics are classified into a first label category, electronic devices are classified into a second label category, and daily necessities are classified into a third label category. It can be understood that the objects can be divided into more than three label categories according to the types of objects.

[0049] In some embodiments, the object features of the detected object can be label features obtained by radio frequency identification technology, and can be image features and size features obtained by image recognition technology.

[0050] The security inspection device stores a set of algorithms, which includes a plurality of target algorithms, for example, the target algorithms include a first explosive identification algorithm, a second explosive identification algorithm, a third explosive identification algorithm, a contraband identification algorithm, a knife identification algorithm, a gun identification algorithm, and a specified tax-related product identification algorithm. The first explosive identification algorithm is suitable for identifying small-sized detected objects, the second explosive identification algorithm is suitable for identifying medium-sized detected objects, and the third explosive identification algorithm is suitable for identifying large-sized detected objects.

[0051] The plurality of identification algorithms correspond one-to-one to the plurality of target algorithms, for example, the first explosive identification algorithm corresponds to the first target algorithm, the second explosive identification algorithm corresponds to the second target algorithm, the third explosive identification algorithm corresponds to the third target algorithm, the contraband identification algorithm corresponds to the fourth target algorithm, the controlled knife identification algorithm corresponds to the fifth target algorithm, the gun identification algorithm corresponds to the sixth target algorithm, and the specified tax-related product identification algorithm corresponds to the seventh target algorithm. It can be understood that the set of algorithms can also include other suspected object identification algorithms.

[0052] The target category of the detected object is determined according to the object feature of the detected object, the target category includes multiple target categories, and the object feature of each target category includes at least one of the size feature information, the image feature information and the label feature information of the detected object. For example, the target category includes a first target category, a second target category, a third target category, and the like.

[0053] For example, the object feature of the detected object includes a size feature, the detected object is classified into a first size category according to the size of the detected object, the target category of the detected object is determined as the first target category according to the first size category, the detected object is determined as the first target category, and the security inspection equipment calls a first explosive recognition algorithm to recognize the scan image of the detected object.

[0054] The detected object is classified into a second size category according to the size of the detected object, the target category of the detected object is determined as the second target category according to the second size category, the detected object is determined as the second target category, and the security inspection equipment calls a second explosive recognition algorithm to recognize the scan image of the detected object.

[0055] The detected object is classified into a third size category according to the size of the detected object, the target category of the detected object is determined as the third target category according to the third size category, the detected object is determined as the third target category, and the security inspection equipment calls a third explosive recognition algorithm to recognize the scan image of the detected object.

[0056] It can be understood that the larger the size of the detected object is, the larger the operation amount of the corresponding explosive recognition algorithm is, and the longer the corresponding detection time is. The first explosive recognition algorithm corresponds to a first target algorithm, the second explosive recognition algorithm corresponds to a second target algorithm, and the third explosive recognition algorithm corresponds to a third target algorithm. The operation amount of the second target algorithm is greater than that of the first target algorithm, and the operation amount of the third target algorithm is greater than that of the second target algorithm.

[0057] The detected object moves to the security inspection equipment along the conveying belt, and if the detected object is determined as the first target category according to the size feature of the detected object, the security inspection equipment calls the first target algorithm to recognize the scan image of the detected object; if the detected object is determined as the second target category, the security inspection equipment calls the second target algorithm to recognize the scan image of the detected object; and if the detected object is determined as the third target category, the security inspection equipment calls the third target algorithm to recognize the scan image of the detected object.

[0058] According to the target category of the detected object, at least one target algorithm corresponding to the target category in the algorithm set is called to recognize the detected object, which is beneficial to reduce the operation amount of the security inspection equipment and improve the security inspection efficiency of the security inspection equipment.

[0059] For example, the object features of the detected object include size features and image features. On the basis of classifying the detected object according to the size features of the detected object, the detected object can be classified according to the size features and the image features of the detected object together to determine the target category of the detected object.

[0060] According to the size features of the detected object, the detected object is classified into a first size category. According to the image features of the detected object, the detected object is classified into a first image category, that is, the detected object contains an item suspected to be powder. The first size category and the first image category determine that the target category of the detected object is a fourth target category, and the detected object is determined to be of the fourth target category. The security inspection device calls the first explosive recognition algorithm and the contraband recognition algorithm to recognize the scan image of the detected object.

[0061] According to the size features of the detected object, the detected object is classified into a first size category. According to the image features of the detected object, the detected object is classified into a second image category, that is, the detected object contains an item suspected to be a knife.

[0062] The first size category and the second image category determine that the target category of the detected object is a fifth target category, and the detected object is determined to be of the fifth target category. The security inspection device calls the first explosive recognition algorithm and the knife recognition algorithm to recognize the scan image of the detected object.

[0063] According to the size features of the detected object, the detected object is classified into a first size category. According to the image features of the detected object, the detected object is classified into a third image category, that is, the detected object contains an item suspected to be a gun.

[0064] The first size category and the third image category determine that the target category of the detected object is a sixth target category, and the detected object is determined to be of the sixth target category. The security inspection device calls the first explosive recognition algorithm and the gun recognition algorithm to recognize the scan image of the detected object.

[0065] According to the size features of the detected object, the detected object is classified into a second size category. According to the image features of the detected object, the detected object is classified into a first image category, that is, the detected object contains an item suspected to be powder. The second size category and the first image category determine that the target category of the detected object is a seventh target category, and the detected object is determined to be of the seventh target category. The security inspection device calls the second explosive recognition algorithm and the contraband recognition algorithm to recognize the scan image of the detected object.

[0066] According to the size features of the detected object, the detected object is classified into a second size category. According to the image features of the detected object, the detected object is classified into a second image category, that is, the detected object contains an item suspected to be a knife.

[0067] The target category of the object under inspection is determined to be the eighth target category according to the second size category and the second image category, the object under inspection is determined to be the eighth target category, and the security inspection device calls the second explosive recognition algorithm and the cutter recognition algorithm to recognize the scan image of the object under inspection.

[0068] According to the size characteristics of the object under inspection, the object under inspection is classified into the second size category, and according to the image characteristics of the object under inspection, the object under inspection is classified into the third image category, that is, the object under inspection contains an item suspected to be a gun.

[0069] The target category of the object under inspection is determined to be the ninth target category according to the second size category and the third image category, the object under inspection is determined to be the ninth target category, and the security inspection device calls the second explosive recognition algorithm and the gun recognition algorithm to recognize the scan image of the object under inspection.

[0070] According to the size characteristics of the object under inspection, the object under inspection is classified into the third size category, and according to the image characteristics of the object under inspection, the object under inspection is classified into the first image category, that is, the object under inspection contains a powder-like item. The target category of the object under inspection is determined to be the tenth target category according to the third size category and the first image category, the object under inspection is determined to be the tenth target category, and the security inspection device calls the third explosive recognition algorithm and the contraband recognition algorithm to recognize the scan image of the object under inspection.

[0071] After the target category of the object under inspection is determined, the object under inspection is imaged by an imaging system in the security inspection device to generate a scan image of the object under inspection. At least one target algorithm in the algorithm set is called based on the target category, the scan image is input to an interface corresponding to the target algorithm, and a recognition result is generated. The recognition result is obtained by fusing the output results of each target algorithm called based on the target category.

[0072] For example, the scan image of the object under inspection is recognized by the at least one target algorithm to generate a recognition result, the recognition result includes a first recognition result and a second recognition result, the first recognition result indicates that the object under inspection is a normal item, and the second recognition result indicates that the object under inspection is a suspicious item. The object under inspection is determined to be a normal item, and the passenger can take the object under inspection on the conveying line. The object under inspection is determined to be a suspicious item, and the object under inspection is conveyed to a waiting processing area for further unpacking detection processing.

[0073] In the embodiments of the present application, after the target category of the object under inspection is determined according to the object characteristics, at least one of the plurality of target algorithms is called by the security inspection device according to the target category, the scan image of the object under inspection is recognized by the at least one target algorithm, and a recognition result is generated. One security inspection device can detect various different objects, which is beneficial to improve the security inspection efficiency of the security inspection device and reduce the overall space occupied by the security inspection device.

[0074] In an optional embodiment, the object features include at least one of a label feature, an image feature, and a size feature.

[0075] Specifically, the object is classified according to the size feature of the object. The object can be classified into a first size category, a second size category, and a third size category according to the size of the object. The object can be classified into the first size category, the second size category, or the third size category through image size recognition.

[0076] The object is classified according to the image feature of the object. The object can be classified into a first image category, a second image category, a third image category, or other image categories according to the geometric shape of the object. The other image categories indicate that the object does not contain powder-like objects, objects similar to knives, or objects similar to guns, and the object does not need to be identified for contraband, knives, and guns. The object can be photographed by a scanning device, and the object can be classified into the first image category, the second image category, the third image category, or the other image categories according to the gray scale information in the generated image.

[0077] The object is classified according to the label feature of the object. For example, a passenger fills out an object information table according to the types of objects carried by the passenger. The types of objects can be daily necessities, cosmetics, electronic devices, etc. A two-dimensional code label is generated from the object information table and is attached to the outside of the object. Cosmetics are classified into a first label category, electronic devices are classified into a second label category, and daily necessities are classified into a third label category. If the object is classified into the first label category, the object needs to be identified by a liquid identification algorithm to determine whether the object contains corrosive liquid.

[0078] The target category of the object is determined according to one or more combinations of the size feature, the image feature, and the label feature.

[0079] The categories of the size feature include a first size category, a second size category, and a third size category, etc. The categories of the image feature include a first image category, a second image category, and a third image category, etc. The categories of the label feature include a first label category, a second label category, and a third label category, etc.

[0080] For example, the subject is determined as a fourth target category according to the first size category and the first image category; the subject is determined as a fifth target category according to the first size category and the second image category; the subject is determined as a sixth target category according to the first size category and the third image category; the subject is determined as a seventh target category according to the second size category and the first image category; the subject is determined as an eighth target category according to the second size category and the second image category; the subject is determined as a ninth target category according to the second size category and the third image category; the subject is determined as a tenth target category according to the third size category and the first image category; the subject is determined as an eleventh target category according to the third size category and the second image category; and the subject is determined as a twelfth target category according to the third size category and the third image category.

[0081] According to the target category, one or more of the first explosive recognition algorithm, the second explosive recognition algorithm, the third explosive recognition algorithm, the contraband recognition algorithm, the knife recognition algorithm, and the gun recognition algorithm are called to recognize the subject.

[0082] The object features include size information, geometric shape information, and category information of the subject, and can comprehensively reflect the feature information of the subject. According to the object features, the target category of the subject is determined, one or more target algorithms are called in a targeted manner, the scan image of the subject is recognized, and then the security efficiency and the security accuracy are improved.

[0083] In an optional embodiment, the security method further includes: obtaining the scan image of the subject before the scan image of the subject is recognized by the at least one target algorithm.

[0084] Specifically, the detection channel of the security device is provided with a radiation source and a detector. The subject moves to a position corresponding to the radiation source and the detector along a conveying belt. The radiation source radiates rays to the subject. The detector receives the rays passing through the subject and then converts the rays into an electric signal. An image processing unit receives the electric signal from the detector and performs signal processing to generate a scan image.

[0085] In an optional embodiment, the security method further includes: before the target category of the subject is determined according to the object features of the subject,

[0086] A category set is generated, and the category set includes a plurality of target categories;

[0087] An algorithm set is generated, and the algorithm set includes a plurality of target algorithms; and

[0088] An association relationship between the plurality of target categories and the plurality of target algorithms is established.

[0089] Specifically, the object under inspection is classified according to the size feature, the image feature, and the label feature, and the category set includes multiple target categories. For example, the first target category corresponds to the first size category; the second target category corresponds to the second size category; the third target category corresponds to the third size category. The fourth target category corresponds to the first size category and the first image category; the fifth target category corresponds to the first size category and the second image category; the sixth target category corresponds to the first size category and the third image category. The seventh target category corresponds to the second size category and the first image category; the eighth target category corresponds to the second size category and the second image category; the ninth target category corresponds to the second size category and the third image category. The tenth target category corresponds to the third size category and the first image category; the eleventh target category corresponds to the third size category and the second image category; the twelfth target category corresponds to the third size category and the third image category. The category set also includes other target categories, which are not described here.

[0090] The target algorithms include a first explosive recognition algorithm, a second explosive recognition algorithm, a third explosive recognition algorithm, a contraband recognition algorithm, a knife recognition algorithm, and a gun recognition algorithm. The first explosive recognition algorithm corresponds to the first target algorithm, the second explosive recognition algorithm corresponds to the second target algorithm, the third explosive recognition algorithm corresponds to the third target algorithm, the contraband recognition algorithm corresponds to the fourth target algorithm, the knife recognition algorithm corresponds to the fifth target algorithm, and the gun recognition algorithm corresponds to the sixth target algorithm. The algorithm set can also include a liquid recognition algorithm and a specified tax-related item recognition algorithm, which are not described here.

[0091] An association between the multiple target categories and the multiple target algorithms is established. The first target category corresponds to the first target algorithm; the second target category corresponds to the second target algorithm; and the third target category corresponds to the third target algorithm. The fourth target category corresponds to the first target algorithm and the fourth target algorithm; the fifth target category corresponds to the first target algorithm and the fifth target algorithm; and the sixth target category corresponds to the first target algorithm and the sixth target algorithm. The seventh target category corresponds to the second target algorithm and the fourth target algorithm; the eighth target category corresponds to the second target algorithm and the fifth target algorithm; and the ninth target category corresponds to the second target algorithm and the sixth target algorithm. The tenth target category corresponds to the third target algorithm and the fourth target algorithm; the eleventh target category corresponds to the third target algorithm and the fifth target algorithm; and the twelfth target category corresponds to the third target algorithm and the sixth target algorithm.

[0092] The association between the multiple target categories and the multiple target algorithms is established, and after the target category of the object under inspection is determined, one or more target algorithms corresponding to the target category can be quickly and accurately called, which is beneficial to improving the security efficiency and the security accuracy.

[0093] In an optional embodiment, the security inspection method further comprises calling at least one target algorithm in the algorithm set according to the preset security inspection instruction and the target category. That is, when the target category of the inspected object is determined, in order to meet the security inspection requirements of different places, at least one target algorithm in the algorithm set is called based on the preset security inspection instruction and the target category, so as to ensure the security inspection requirements of different places, only the calling of the algorithm needs to be changed, and the security inspection equipment does not need to be adjusted, thereby improving the security inspection efficiency and the adaptability of the security inspection equipment.

[0094] Specifically, the preset security inspection instruction is a security inspection requirement preset for the inspected object. For example, the preset security inspection instruction can be set according to the security inspection regulations of the destination. Different security inspection instructions are generated according to different flight destinations. For example, the destination of a flight is a first destination, the first destination is a country where the trade of prohibited goods is not controlled, the first security inspection instruction indicates that it is necessary to detect whether the inspected object contains prohibited goods, and the security inspection equipment calls the fourth target algorithm to identify the inspected object. For example, the destination of a flight is a second destination, the second destination is a country where guns are not controlled, the second security inspection instruction indicates that it is necessary to detect whether the inspected object contains guns, and the security inspection equipment calls the sixth target algorithm to identify the inspected object. For example, the destination of a flight is a third destination, the third destination is a city in China, and it is only necessary to detect whether the inspected object contains explosives, and the security inspection equipment calls one of the first target algorithm, the second target algorithm and the third target algorithm to identify the inspected object.

[0095] At least one target algorithm is called according to the target category. For example, it is determined that the inspected object is a first target category, it is necessary to detect whether the inspected object contains explosives, and the security inspection equipment calls the first target algorithm to identify the inspected object. For example, it is determined that the inspected object is a second target category, it is necessary to detect whether the inspected object contains explosives, and the security inspection equipment calls the second target algorithm to identify the inspected object. For example, it is determined that the inspected object is a third target category, it is necessary to detect whether the inspected object contains explosives, and the security inspection equipment calls the third target algorithm to identify the inspected object.

[0096] At least one target algorithm is called according to the preset security inspection instruction and the target category. For example, the destination of a flight is a first destination, a first security inspection instruction is generated, and it is necessary to detect whether the inspected object contains prohibited goods; according to the object characteristics of the inspected object, it is determined that the inspected object is a first target category, it is necessary to detect whether the inspected object contains explosives, and the security inspection equipment calls the first target algorithm and the fourth target algorithm to identify the inspected object. For example, the destination of a flight is a second destination, a second security inspection instruction is generated, and it is necessary to detect whether the inspected object contains guns; according to the object characteristics of the inspected object, it is determined that the inspected object is a second target category, it is necessary to detect whether the inspected object contains explosives, and the security inspection equipment calls the second target algorithm and the sixth target algorithm to identify the inspected object.

[0097] According to the preset security inspection instruction and the target category, at least one target algorithm in the algorithm set is called, which is beneficial to expand the detection range, meet the security inspection needs of various scenes, improve the accuracy of the security inspection result, and does not need to adjust the security inspection equipment and the security inspection procedure, thereby improving the security inspection efficiency.

[0098] In an optional embodiment, the scanning image of the inspected object is identified by the at least one target algorithm to generate an identification result, which includes:

[0099] The scanning image of the inspected object is identified by the at least one target algorithm to generate an identification sub-result associated with each target algorithm;

[0100] The identification result is generated according to the identification sub-result.

[0101] Specifically, each target algorithm corresponds to an identification sub-result, and the identification result is generated from one or more identification sub-results. The identification sub-result is represented by yes or no.

[0102] For example, it is determined that the inspected object belongs to the fourth target category, and the security inspection equipment calls the first target algorithm and the fourth target algorithm to identify the scanning image of the inspected object. The first identification sub-result is generated by the first target algorithm, and the first identification sub-result is yes, indicating that the inspected object contains explosives, and the first identification sub-result is no, indicating that the inspected object does not contain explosives. The fourth identification sub-result is generated by the fourth target algorithm, and the fourth identification sub-result is yes, indicating that the inspected object contains prohibited items, and the fourth identification sub-result is no, indicating that the inspected object does not contain prohibited items. The first identification sub-result and the fourth identification sub-result are both no, and the first identification result is generated, indicating that the inspected object is a normal item. At least one of the first identification sub-result and the fourth identification sub-result is yes, and the second identification result is generated, indicating that the inspected object is a suspicious item, and the inspected object needs to be further detected.

[0103] It can be understood that the inspected object can also be determined to be a normal item or a suspicious item according to a single identification sub-result, and the identification sub-result is represented by different colors in the image. For example, it is determined that the inspected object belongs to the fourth target category, and the security inspection equipment calls the first target algorithm and the fourth target algorithm to identify the scanning image of the inspected object. The first identification sub-result is represented by red in the image, and the fourth identification sub-result is represented by yellow in the image. If a red marked area and a yellow marked area appear in the image, or both a red marked area and a yellow marked area appear in the image, it is determined that the inspected object is a suspicious item, and the inspected object needs to be further detected. If neither a red marked area nor a yellow marked area appears in the image, it is determined that the inspected object is a normal item.

[0104] For example, it is determined that the object under inspection is the fifth target category, and the security inspection device calls the first target algorithm and the fifth target algorithm to identify the scanned image of the object under inspection. The first identification sub-result is generated by the first target algorithm, and the first identification sub-result is yes, indicating that it contains explosives, and the first identification sub-result is no, indicating that it does not contain explosives; the fifth identification sub-result is generated by the fifth target algorithm, and the fifth identification sub-result is yes, indicating that it contains a knife, and the fifth identification sub-result is no, indicating that it does not contain a knife. The first identification sub-result and the fifth identification sub-result are both no, and the first identification result is generated, indicating that the object under inspection is a normal item. At least one of the first identification sub-result and the fifth identification sub-result is yes, and the second identification result is generated, indicating that the object under inspection is a suspicious item, and further detection of the object under inspection is required.

[0105] For example, it is determined that the object under inspection is the sixth target category, and the security inspection device calls the first target algorithm and the sixth target algorithm to identify the scanned image of the object under inspection. The first identification sub-result is generated by the first target algorithm, and the first identification sub-result is yes, indicating that it contains explosives, and the first identification sub-result is no, indicating that it does not contain explosives; the sixth identification sub-result is generated by the sixth target algorithm, and the sixth identification sub-result is yes, indicating that it contains a gun, and the sixth identification sub-result is no, indicating that it does not contain a gun. The first identification sub-result and the sixth identification sub-result are both no, and the first identification result is generated, indicating that the object under inspection is a normal item. At least one of the first identification sub-result and the sixth identification sub-result is yes, and the second identification result is generated, indicating that the object under inspection is a suspicious item, and further detection of the object under inspection is required.

[0106] For example, it is determined that the object under inspection is the seventh target category, and the security inspection device calls the second target algorithm and the fourth target algorithm to identify the scanned image of the object under inspection. The second identification sub-result is generated by the second target algorithm, and the second identification sub-result is yes, indicating that it contains explosives, and the second identification sub-result is no, indicating that it does not contain explosives; the fourth identification sub-result is generated by the fourth target algorithm, and the fourth identification sub-result is yes, indicating that it contains prohibited items, and the fourth identification sub-result is no, indicating that it does not contain prohibited items. The second identification sub-result and the fourth identification sub-result are both no, and the first identification result is generated, indicating that the object under inspection is a normal item. At least one of the second identification sub-result and the fourth identification sub-result is yes, and the second identification result is generated, indicating that the object under inspection is a suspicious item, and further detection of the object under inspection is required.

[0107] It can be understood that the object under inspection is determined to be the fifth target category, the sixth target category, or the seventh target category, and the object under inspection can also be determined to be a normal item or a suspicious item according to a single identification sub-result, and the identification sub-result is represented in different colors in the image.

[0108] The object to be detected is identified by at least one target algorithm, each target algorithm corresponding to an identification sub-result associated with the target algorithm, the security inspection device calling multiple target algorithms according to the target category to identify the scanned image of the object to be detected, each target algorithm corresponding to an identification sub-result, multiple identification sub-results being all no, generating a first identification result, one of the multiple identification sub-results being yes, generating a second identification result, multiple categories of prohibited items can be identified in one scan, which is conducive to improving the security inspection efficiency and also conducive to improving the security inspection accuracy.

[0109] In optional embodiments, generating an identification result according to identification sub-results can be weighted voting for each identification sub-result, and the weight of each sub-result is different. Alternatively, multiple sub-results are fused according to a set fusion strategy to generate an identification result. Generating an identification result according to identification sub-results can improve the accuracy of the identification result.

[0110] In optional embodiments, each target algorithm in the algorithm set has an independent interaction interface.

[0111] Specifically, each target algorithm has an independent interaction interface, for example, the first target algorithm corresponds to the first interaction interface, the second target algorithm corresponds to the second interaction interface, the third target algorithm corresponds to the third interaction interface, the fourth target algorithm corresponds to the fourth interaction interface, the fifth target algorithm corresponds to the fifth interaction interface, and the sixth target algorithm corresponds to the sixth interaction interface.

[0112] When the object to be detected is determined to be the first target category, the first target algorithm is called through the first interaction interface; when the object to be detected is determined to be the fourth target category, the first target algorithm and the fourth target algorithm are called through the first interaction interface and the fourth interaction interface; when the object to be detected is determined to be the fifth target category, the first target algorithm and the fifth target algorithm are called through the first interaction interface and the fifth interaction interface; and when the object to be detected is determined to be the sixth target category, the first target algorithm and the sixth target algorithm are called through the first interaction interface and the sixth interaction interface.

[0113] The multiple target categories in the category set are numbered, and the multiple target categories correspond to an array, for example, the first target category, the second target category, the third target category, the fourth target category, the fifth target category, the sixth target category, and the like are sequentially numbered as category 1, category 2, category 3, category 4, category 5, category 6, and the like. The multiple target algorithms in the algorithm set are numbered, and the multiple target algorithms correspond to an array, for example, the first target algorithm, the second target algorithm, the third target algorithm, the fourth target algorithm, the fifth target algorithm, and the sixth target algorithm are sequentially numbered as algorithm 1, algorithm 2, algorithm 3, algorithm 4, algorithm 5, and algorithm 6.

[0114] The fourth target category corresponds to the first size category and the first image category, and if the judgment results of the first size category and the first image category are both true, algorithm 1 and algorithm 4 are called; the fifth target category corresponds to the first size category and the second image category, and if the judgment results of the first size category and the second image category are both true, algorithm 1 and algorithm 5 are called; the sixth target category corresponds to the first size category and the third image category, and if the judgment results of the first size category and the third image category are both true, algorithm 1 and algorithm 6 are called.

[0115] Each target algorithm has an independent interactive interface, which facilitates quick calling of the target algorithm. In addition, each target algorithm can be optimized independently. When a single target algorithm is optimized, it will not affect other target algorithms.

[0116] In the case of coping with diversified security requirements in the later stage, target algorithms can be added. For example, according to the size characteristics of the detected object, the fourth size category and the fifth size category can be added, and the size specifications of the first size category, the second size category, the third size category, the fourth size category and the fifth size category increase in turn. The newly added target algorithm will not affect the existing other target algorithms, thereby the target algorithm library can be continuously enriched, which is conducive to further improving the security accuracy.

[0117] In an optional embodiment, after the identification result is generated, the detected object is sorted according to the identification result.

[0118] Specifically, the identification result includes a first identification result and a second identification result. The first identification result indicates that the detected object is a normal item, and the passenger can directly take the detected object after detection. The second identification result indicates that the detected object is a suspicious item, and the detected object moves to the waiting area along with the conveying line, waiting for the security personnel to further detect the detected object.

[0119] As shown in FIG. 2, the present application also provides a security device, which comprises a conveying line body 1 and a security device 2. The conveying line body 1 is used for conveying a detected object 3. The security device 2 is arranged in the conveying direction of the conveying line body 1, and the security device 2 is configured to execute the above-mentioned security method.

[0120] Specifically, the security device 2 is located in the conveying direction of the conveying line body 1, and the entrance end and the exit end of the security device 2 can be respectively provided with a first photoelectric module and a second photoelectric module. The first photoelectric module and the second photoelectric module are respectively connected with the security system through a first interface and a second interface. The detected object 3 moves to the entrance end of the security device 2 along with the conveying belt, triggers the first photoelectric module, and the security system sends the target category of the detected object 3 to the security device 2 through the first interface.

[0121] The detection object 3 enters the detection channel of the security inspection device 2, the scanning unit scans the detection object 3 and generates a scanning image, the security inspection device 2 calls a target algorithm corresponding to the target category to identify the scanning image, and generates an identification result.

[0122] If it is determined that the category of the detection object 3 is the first target category, the security inspection device 2 calls the first target algorithm to identify the scanning image; if it is determined that the category of the detection object 3 is the second target category, the security inspection device 2 calls the second target algorithm to identify the scanning image; if it is determined that the category of the detection object 3 is the third target category, the security inspection device 2 calls the third target algorithm to identify the scanning image; and if it is determined that the category of the detection object 3 is the fourth target category, the security inspection device 2 calls the first target algorithm and the fourth target algorithm to identify the scanning image.

[0123] The detection object 3 moves to the exit end of the security inspection device 2 along the conveying belt, triggers the second photoelectric module, and the security inspection system sends the target category and the identification result of the detection object 3 to the sorting device through the second interface, so as to sort the detected detection object 3.

[0124] If the identification result is the first identification result, it indicates that the detection object 3 is a normal article, the detection object 3 moves to the pickup area along the first conveying line body, and the passenger takes the detection object 3; if the identification result is the second identification result, it indicates that the detection object 3 is a suspicious article, the detection object 3 moves to the waiting area along the second conveying line body, and waits for further detection.

[0125] According to the object characteristics of the detection object 3, the target category of the detection object 3 is determined, at least one target algorithm in the algorithm set of the security inspection device 2 is called according to the target category, the scanning image of the detection object 3 is identified through the at least one target algorithm, an identification result is generated, and the detection object 3 is further sorted according to the identification result.

[0126] By switching algorithms to call one or more target algorithms, a security inspection device 2 can identify multiple types of suspicious objects after one scan.

[0127] Multiple target algorithms are integrated on one security inspection device 2, each target algorithm has an independent interactive interface, and the addition or optimization of a target algorithm will not affect other target algorithms, which is conducive to the optimization and upgrading of the target algorithm library and meets diversified security inspection needs.

[0128] By calling a target algorithm corresponding to a target category in an algorithm set through one security inspection device 2, scanning and identification of multiple categories of detection objects are realized, the utilization rate of the security inspection device 2 is improved, the number of security inspection devices 2 is reduced, which is conducive to saving the occupied space of the security inspection device and reducing the maintenance cost of the security inspection device, and also conducive to improving the security inspection efficiency and accuracy.

[0129] The application also provides a security inspection system comprising the security inspection device.

[0130] The security inspection system with the security inspection device has the effective effects brought by the security inspection device and the security inspection method, that is, the security inspection system has reduced occupied space and reduced maintenance cost, and is convenient for system upgrading in the future.

[0131] The above merely describes specific embodiments of the application, but the protection scope of the application is not limited thereto. Any changes or replacements made within the spirit and principle of the application should be covered within the protection scope of the application. Therefore, the protection scope of the application should be subject to the protection scope of the claims.

Claims

1. A security inspection method, comprising: determining a target category of a subject to be inspected according to a subject feature of the subject to be inspected; calling at least one target algorithm in a set of algorithms according to the target category, the at least one target algorithm being determined according to the target category, the set of algorithms being integrated in a same security inspection device; performing recognition on a scan image of the subject to be inspected by the at least one target algorithm to generate a recognition result.

2. The security screening method of claim 1, wherein, Further comprising: acquiring the subject feature of the subject to be inspected before determining the target category of the subject to be inspected according to the subject feature of the subject to be inspected.

3. The security inspection method according to claim 1 or 2, wherein: the subject feature comprises at least one of a label feature, an image feature, and a size feature.

4. The security method of claim 1, wherein, Further comprising: acquiring the scan image of the subject to be inspected before performing recognition on the scan image of the subject to be inspected by the at least one target algorithm.

5. The security method of claim 1, wherein, Further comprising: before determining the target category of the subject to be inspected according to the subject feature of the subject to be inspected, generating a category set, the category set comprising a plurality of the target categories; generating the set of algorithms, the set of algorithms comprising a plurality of the target algorithms; and establishing an association between the plurality of the target categories and the plurality of the target algorithms.

6. The security method of claim 1, wherein, Further comprising: calling the at least one target algorithm in the set of algorithms according to a preset security inspection instruction and the target category.

7. The security inspection method according to claim 1, wherein: the performing recognition on the scan image of the subject to be inspected by the at least one target algorithm to generate the recognition result comprises: performing recognition on the scan image of the subject to be inspected by the at least one target algorithm to generate a recognition sub-result associated with each of the target algorithms; and generating the recognition result according to the recognition sub-result.

8. The security inspection method according to claim 1, wherein: each of the target algorithms in the set of algorithms has an independent interactive interface.

9. The security inspection method according to claim 1, further comprising: sorting the subject to be inspected according to the recognition result after generating the recognition result.

10. A security inspection apparatus, comprising: a conveying line body for conveying a subject to be inspected; a security inspection device arranged in a conveying direction of the conveying line body, the security inspection device being configured to perform the method according to any one of claims 1 to 9.

11. A security inspection system, comprising: the security inspection apparatus according to claim 10.

Images