CN112884085A - Method, system and equipment for detecting and identifying contraband based on X-ray image - Google Patents

Abstract

The invention belongs to the field of computer vision target detection and recognition, and in particular relates to a method, system and device for detecting and identifying prohibited items based on X-ray images. The method includes: determining a target image set according to preset refined category information of contraband; performing data source expansion preprocessing on the target image set to obtain a training image set; inputting the training image set into a detection and recognition training model for training In order to obtain the detection and recognition network model; input the X-ray image to be detected into the detection and recognition network model to obtain the confidence detection value of each detection area in the X-ray image to be detected; The detection area is determined to be a contraband area. The invention greatly improves the detection accuracy.

Description

X-ray image-based detection and identification method, system and device for prohibited items

technical field

The invention belongs to the technical field of computer vision target detection and identification, and particularly relates to a method, system and device for detecting and identifying prohibited items based on X-ray images.

Background technique

Luggage security inspection is a very important line of defense for public security. Through perspective scanning of luggage, hidden safety hazards can be discovered in time. For a long time, in the detection and identification of prohibited items such as guns and knives in public places, most of the prohibited items in X-ray security inspection images are detected and identified with the naked eye. Due to the dense arrangement and overlapping of the items in the luggage, it brings certain difficulties to the security inspection work, and the security personnel are prone to wrong inspections and missed inspections in a high-pressure environment for a long time. Even the security inspectors with excellent professional quality will inevitably make some mistakes, which will cause serious hidden safety problems. Therefore, it is necessary to construct a set of intelligent detection and identification system to assist the work of security inspectors and improve work efficiency.

At present, the prior art is generally based on a high-performance image processor, and an artificial intelligence deep learning algorithm is used to intelligently detect X-ray security images.

However, due to the different penetration capabilities of different X-ray machine manufacturers, there will be certain deviations in the expression and feature distribution of the captured X-ray image data. At the same time, contraband items will show different sizes and shapes in different X-ray images, which further increases the difficulty of detection and identification, resulting in low detection and identification accuracy.

SUMMARY OF THE INVENTION

In order to solve the above-mentioned problems in the prior art, that is, the problem of low detection and identification accuracy, in a first aspect, the present invention provides a method for detecting and identifying contraband based on X-ray images, comprising the following steps:

Determine the target image set according to the refined category information of the preset contraband;

Performing data source expansion preprocessing on the target image set to obtain a training image set;

Inputting the training image set to a detection and recognition training model for training to obtain a detection and recognition network model;

Inputting the X-ray image to be detected into the detection and identification network model to obtain the confidence detection value of each detection area in the X-ray image to be detected;

The detection area corresponding to the confidence detection value in the preset confidence interval is determined as the contraband area.

Optionally, the determining of the target image set according to the preset refined category information of contraband includes:

Get a collection of contraband images;

The images that are consistent with the preset refined category information of contraband are selected from the contraband image set as the target image set.

Optionally, the obtaining a set of images of contraband includes:

Get the first set of contraband images;

and/or acquiring a second set of images of contraband; the first set of images of contraband is an X-ray image set containing contraband; the second set of images of contraband is a set of separate images of contraband;

The selection of images consistent with the preset refined category information of contraband in the contraband image set as the target image set includes:

In the first contraband image set, select images consistent with the preset contraband refined category information as the target image set;

And/or selecting images consistent with preset refined category information of contraband in the second contraband image set as a target image set.

Optionally, performing data source expansion preprocessing on the target image set to obtain a training image set includes:

Perform first preprocessing on the target image set screened in the first contraband image set to obtain a training image set;

And/or performing second preprocessing on the target image set screened in the second contraband image set to obtain a training image set.

Optionally, the process of the first preprocessing is:

The hue, brightness and saturation of each target image in the target image set are randomly processed to different degrees to obtain several first intermediate images;

A training image set is obtained by randomly cropping the several first intermediate images respectively.

Optionally, the process of the second preprocessing is:

Performing rotation and affine transformation operations on each target image in the target image set to obtain a second intermediate image;

The second intermediate image is fused with the X-ray image that does not contain contraband according to a preset fusion rule to obtain a training image set.

Optionally, after determining the detection area corresponding to the confidence detection value in the preset confidence interval as the contraband area, the method further includes:

Compare the confidence detection value corresponding to the contraband area with the preset detection value;

Retain the confidence detection value higher than or equal to the preset detection value and the corresponding detection area;

The detection area image corresponding to the confidence detection value lower than the preset detection value is input into the convolutional neural network classification model to further determine the category of contraband; the preset detection value is any value within the preset confidence interval.

Optionally, the construction process of the convolutional neural network classification model is:

Obtaining a third set of images of contraband and X-ray images that do not contain contraband; wherein, the set of images of the third and second images of contraband are the same;

randomly intercepting part of the image of the X-ray image that does not contain contraband;

The third preprocessing is performed on the third contraband image set and the partial area image of the X-ray image that does not contain contraband to obtain a classification training image set;

The classification training image set is input into a classification training model for training to obtain a convolutional neural network classification model.

Another aspect of the present invention provides a system for detecting and identifying contraband based on X-ray images, including:

a first determining unit, configured to determine a target image set according to the preset refined category information of contraband;

a data expansion unit for performing data source expansion preprocessing on the target image set to obtain a training image set;

A training unit for inputting the training image set into a detection and recognition training model for training to obtain a detection and recognition network model;

a detection and identification unit, configured to input the X-ray image to be detected into the detection and identification network model to obtain the confidence detection value of each detection area in the X-ray image to be detected;

The second determination unit is configured to determine the detection area corresponding to the confidence detection value in the preset confidence interval as the contraband area.

In a third aspect of the present invention, a device is provided, comprising:

at least one processor; and

a memory communicatively coupled to at least one of the processors; wherein,

The memory stores instructions executable by the processor, and the instructions are used to be executed by the processor to implement the method for detecting and identifying contraband based on an X-ray image according to any one of the first aspects.

In a fourth aspect of the present invention, a computer-readable storage medium is provided, where the computer-readable storage medium stores computer instructions, and the computer instructions are used to be executed by the computer to implement any one of the first aspects. A method for detection and identification of contraband based on X-ray images.

Beneficial effects of the present invention: the present invention determines the target image set by refining the category information of the preset contraband, and can obtain the target image set with more detailed classification and definition, avoids the influence of the detection and recognition effect due to individual differences in the same contraband, and provides The target image set is subjected to data expansion preprocessing to make the data of the training image set more diverse, thereby effectively improving the detection and recognition accuracy of the trained detection and recognition network model.

Description of drawings

Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings:

1 is a schematic diagram of a method for detecting and identifying contraband based on an X-ray image according to an embodiment of the present application;

Fig. 2 is the X-ray image set containing contraband in the present application and the schematic diagram of the result after the first preprocessing;

Fig. 3 is the individual image set of contraband of the present application and the schematic diagram of the result after the second preprocessing;

4 is a partial detection result of the X-ray image to be detected in the present application after the detection and identification network model;

5 is a schematic diagram of a method for detecting and identifying contraband based on an X-ray image according to another embodiment of the present application;

6 is a schematic structural diagram of the X-ray image-based contraband detection and identification system of the present application;

FIG. 7 is a schematic structural diagram of a computer system for implementing a server according to embodiments of the method, system, and device of the present application.

Detailed ways

The present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the related invention, but not to limit the invention. In addition, it should be noted that, for the convenience of description, only the parts related to the related invention are shown in the drawings.

It should be noted that the embodiments in the present application and the features of the embodiments may be combined with each other in the case of no conflict. The present application will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

The present invention provides a method for detecting and identifying contraband based on X-ray images, the method comprising:

Determine the target image set according to the refined category information of the preset contraband;

Performing data source expansion preprocessing on the target image set to obtain a training image set;

Inputting the training image set to a detection and recognition training model for training to obtain a detection and recognition network model;

Inputting the X-ray image to be detected into the detection and identification network model to obtain the confidence detection value of each detection area in the X-ray image to be detected;

The detection area corresponding to the confidence detection value in the preset confidence interval is determined as the contraband area.

In order to more clearly describe the method of the present invention for detecting and identifying contraband based on X-ray images, each step in the embodiment of the present invention will be described in detail below with reference to FIG. Note that, of course, the method of this embodiment can also be applied to other contrabands whose categories can be directly and clearly determined by their appearance.

The method for detecting and identifying contraband based on X-ray images according to the first embodiment of the present invention includes steps S101 to S105, and each step is described in detail as follows:

S101. Determine a target image set according to preset refined category information of contraband.

In the embodiment of the present application, the preset refined category information of contraband refers to detailed classification information of contraband, for example, including information such as category name, shape, size, color, and the like of contraband. The target image set is a collection of relevant images of contraband that needs to be detected.

Optionally, the determining of the target image set according to the preset refined category information of contraband includes:

Get a collection of contraband images.

In this step, the category of the contraband image set is firstly obtained, and the contraband image set is obtained from the Internet according to the category.

Specifically, first, according to the detection and identification requirements, the detailed categories of contraband to be detected are determined. For example, the detection and identification requirement is to detect firearms and knives, and classify them in detail according to their size and shape. For example, firearms can be divided into pistols, submachine guns and long guns, and knives can be divided into long and short knives and kitchen knives.

By subdividing the classification of contraband, it can effectively avoid the influence of the same category of contraband, such as guns, on the detection and recognition results due to different shapes and sizes.

The images that are consistent with the preset refined category information of contraband are selected from the contraband image set as the target image set.

In the present application, after obtaining the contraband image set, images consistent with the preset refined category of contraband can be manually or automatically screened out as the target image set.

In the case of manual screening, the pre-defined size and shape of firearms and knives need to be compared with the image set of contraband images, and those that do not match are eliminated.

In the case of automatic screening, in an example, the two images can be compared for the degree of coincidence. For example, if the degree of coincidence is greater than the preset degree of coincidence, it is determined to be consistent, which can improve efficiency.

In a specific embodiment, the obtaining a set of images of contraband includes:

Get the first set of contraband images;

and/or acquiring a second set of images of contraband; the first set of images of contraband is an X-ray image set containing contraband; the second set of images of contraband is a set of separate images of contraband.

In this embodiment, the first set of images of contraband is downloaded from a specific professional website, and the second set of images of contraband can be downloaded from the network using a web crawler. In an example, the process may include the following steps:

The first step is to confirm the target address of the crawled image, and use the Java language Gecco web crawler to initiate a crawling request to the target address.

The second step is to return the source code of the web page containing the image link for the crawling request, use the Jsoup parser to analyze the structure of the web page, and parse the link of each image.

The third step is to traverse the link of the image and initiate a request for the link, write the returned result locally, and save it as an image file.

The selection of images consistent with the preset refined category information of contraband in the contraband image set as the target image set includes:

In the first contraband image set, select images consistent with the preset contraband refined category information as the target image set;

And/or selecting images consistent with preset refined category information of contraband in the second contraband image set as a target image set.

In the embodiment of the present application, the contraband image set can be divided into two categories, one is an X-ray image set containing contraband, and the other is a separate image set of contraband. As shown in Figure 2, four sets of images of contraband are given in Figure 2, the top and bottom are a group, the bottom picture is the X-ray image containing contraband, and the top is the image after data source expansion and preprocessing. As shown in Figure 3, the left-most column in Figure 3 is a separate image of guns and knives. Separate images of guns and knives are preprocessed by data source augmentation and integrated into X-ray images without contraband to obtain the final training image set.

By using two types of contraband image sets as the target image set, the style of the target image set can be greatly increased, the data source can be more abundant, and the detection accuracy can be improved.

In order to further improve the generalization and robustness of the detection and recognition effect, the target image set needs to be further enhanced, so as to expand the data source and meet the data requirements of the detection and recognition training model. The specific steps are given below.

S102. Perform data source expansion preprocessing on the target image set to obtain a training image set. According to the above embodiment, the following steps may be specifically included:

A training image set is obtained by performing first preprocessing on the target image set screened in the first contraband image set. The process of the first preprocessing is:

The hue, brightness and saturation of each target image in the target image set are randomly processed to different degrees to obtain several first intermediate images.

A training image set is obtained by randomly cropping the several first intermediate images respectively.

The process of obtaining the first intermediate image by operating the target image screened in the first contraband image set, that is, the X-ray image containing contraband in the HSV space, is as follows:

Use the OPENCV function to convert the target image from RGB space to HSV space;

Adjust the values of hue H, saturation S and lightness V of the target image in HSV space respectively, so that they have obvious changes from the HSV value of the original image;

Use the OPENCV function to convert the image from HSV space to RGB space and store the image.

Referring to FIG. 2 , a schematic diagram of an X-ray image containing contraband after the first preprocessing is given, and the upper four images are the training images after the first preprocessing respectively.

Through the above-mentioned means, the target image can be rendered with different brightness, size and color, so that the training image set can be expanded to meet the requirements of the detection and recognition training model.

And/or performing second preprocessing on the target image set screened in the second contraband image set to obtain a training image set. The process of the second preprocessing is:

As shown in FIG. 3 , operations such as rotation and affine transformation are performed on each target image (a) in the target image set to obtain a second intermediate image (b). The image in the middle column is the second intermediate image.

The second intermediate image is fused with the X-ray image containing no contraband according to a preset fusion rule to obtain a training image set (c). It makes the same gun or knife show different sizes and shapes in different X-ray images, and realizes the diversification of data.

In this step, the preset fusion rules can employ embedded rules to embed guns or knives into X-ray images that do not contain contraband.

In addition, the acquisition of the X-ray images that do not contain contraband may follow the acquisition process of the first set of prohibited images and the second set of prohibited images in the above example.

Referring to Figure 3, the process of integrating guns and knives into an X-ray image without contraband according to preset fusion rules after transformations such as rotation and affine is given.

S103. Input the training image set into a detection and recognition training model for training to obtain a detection and recognition network model.

In this step, the training image set can be divided into five categories for input, and the number of images in each category is about 4000, totaling 20786. As shown in Figure 2 and Figure 3.

Among them, before training, it is necessary to set various parameters of the detection and recognition training model. For example, the size of the parameter Bathsize is set to 64, and the number of training times is set to 10,000 times, that is, 64 pictures are used as a set of training parameters. After training 10,000 times, the parameters of the detection and recognition network model are obtained.

In the embodiment of the present application, the initial parameters of the detection and recognition training model are based on the parameters trained in the 80 categories of the COCO dataset in the prior art. There is a certain difference in size, so it is necessary to re-cluster the size of the contraband in the X-ray image so that the size of the guns and knives matches the size of the preset frame in the detection and recognition training model.

S104 , inputting the X-ray image to be detected into the detection and identification network model to obtain the confidence detection value of each detection area in the X-ray image to be detected.

In one example, the first-order target detection model YOLO can be used as the detection and recognition training model, which can improve the detection and recognition efficiency.

S105. Determine the detection area corresponding to the confidence detection value in the preset confidence interval as the contraband area. As shown in Figure 4, a schematic diagram of the detection results is given. The contraband area in the figure is framed, and the location of the contraband can be visually seen.

In the embodiment of the present application, the confidence is also called reliability, or confidence level, confidence coefficient, that is, when sampling to estimate a population parameter, due to the randomness of the sample, the conclusion is always uncertain. Therefore, a probability statement method is adopted, that is, the interval estimation method in mathematical statistics, that is, the estimated value and the overall parameter are within a certain allowable error range, what is the corresponding probability, and this corresponding probability is called confidence. A confidence interval is a probability interval.

In one example, the detection area can be determined according to the position of the object in the X-ray image to be detected. For example, a short knife, a toothbrush and a paper towel are displayed in the image to be detected. The detection area can be determined according to the position of the three objects, and three detections are obtained. confidence in the area. For example, the preset confidence interval is 0.5-1, the confidence level of the short knife detection area is 0.6, and the confidence levels of the toothbrush and paper towel detection areas are 0.3 and 0.2 respectively, then the short knife detection area is determined as the contraband area.

The detection and recognition network model can also output the location information of contraband, such as spatial coordinate data. According to the confidence and coordinate data, the contraband can be locked and prompted through the box or voice broadcast.

In order to improve the recall rate of contraband, the threshold of the preset confidence interval can be lowered. For example, the standard preset confidence interval is 0.5-1, then it can be adjusted to 0.3-1, so that more detection results can be obtained. But this will treat items that are not contraband, such as toothbrushes, as contraband, which will affect the detection accuracy.

Therefore, it is necessary to further discriminate the detection results of the above embodiments.

As shown in FIG. 5 , after determining the detection area corresponding to the confidence detection value in the preset confidence interval as the contraband area, the method further includes:

S501. Compare the confidence detection value corresponding to the contraband area with the preset detection value.

S502. Retain a confidence detection value higher than or equal to a preset detection value and a corresponding detection area.

S503: Input the detection area image corresponding to the confidence detection value lower than the preset detection value into the convolutional neural network classification model to further determine the category of contraband. The preset detection value is any value within the preset confidence interval.

On the basis of the above example, the preset confidence interval is 0.3-1, the confidence level of the detection area of the toothbrush is 0.3, and the confidence level of the detection area of the short knife is 0.6, confirming that the toothbrush and the short knife are contraband, in one example, Set the default detection value to 0.5, then continue to retain the detection result of the short knife, input the image of the detection area of the toothbrush below 0.5 into the convolutional neural network classification model for category determination, and exclude the detection area if it is identified as impurity , otherwise it is reserved as the final detection result. Of course in this example, according to the classification model, the toothbrush will be excluded from the category of contraband.

Wherein, the construction process of the convolutional neural network classification model is:

A third set of images of contraband and X-ray images without contraband are acquired; wherein, the set of images of third and second images of contraband are the same. They are all images of single contraband such as guns and knives, and their acquisition methods are also obtained by referring to the above examples and using web crawlers.

Partial area images of the X-ray images that do not contain contraband are randomly intercepted.

Random intercepts are stored as impurity categories in order to exclude items that are not contraband.

The third preprocessing is performed on the third contraband image set and the partial area image of the X-ray image that does not contain contraband to obtain a classification training image set.

The third preprocessing can include operations such as random cropping, affine transformation, and color adjustment, which further enhances the diversity of data and enables the convolutional neural network classification model to have better generalization.

The classification training image set is input into a classification training model for training to obtain a convolutional neural network classification model.

In an example, the deep learning network model VGG16 can be used as the backbone network, and the Focal Loss function can be used as the loss function to construct a classification training model.

Based on the same inventive concept, the second embodiment of the present application provides a system for detecting and identifying contraband based on X-ray images, as shown in FIG. 6 , including:

a first determining unit 610, configured to determine a target image set according to preset refined category information of contraband;

a data expansion unit 620, configured to perform data source expansion preprocessing on the target image set to obtain a training image set;

A training unit 630, configured to input the training image set into a detection and recognition training model for training to obtain a detection and recognition network model;

A detection unit 640, configured to input the X-ray image to be detected into the detection and identification network model to obtain the confidence detection value of each detection area in the X-ray image to be detected;

The second determining unit 650 is configured to determine the detection area corresponding to the confidence detection value within the preset confidence interval as the contraband area.

Those skilled in the art can clearly understand that, for the convenience and brevity of description, for the specific working process and related description of the system described above, reference may be made to the corresponding process in the foregoing method embodiments, which will not be repeated here.

It should be noted that, the X-ray image-based contraband detection and identification system provided in the above-mentioned embodiments is only illustrated by the division of the above-mentioned functional modules. module, that is, the modules or steps in the embodiments of the present invention are decomposed or combined. For example, the modules in the above embodiments can be combined into one module, or can be further split into multiple sub-modules to complete all or part of the above description. Function. The names of the modules and steps involved in the embodiments of the present invention are only for distinguishing each module or step, and should not be regarded as an improper limitation of the present invention.

Based on the same inventive concept, a device according to the third embodiment of the present invention includes:

at least one processor; and

a memory communicatively coupled to at least one of the processors; wherein,

The memory stores instructions executable by the processor, and the instructions are used to be executed by the processor to implement the method for detecting and identifying contraband based on an X-ray image described in the first embodiment.

A computer-readable storage medium according to a fourth embodiment of the present invention, the computer-readable storage medium stores computer instructions, and the computer instructions are used to be executed by the computer to implement the X-ray-based method described in the first embodiment. A method for image detection and identification of contraband.

Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process and relevant description of the storage device and processing device described above can refer to the corresponding process in the foregoing method embodiments, which is not repeated here. Repeat.

Referring to FIG. 7 below, it shows a schematic structural diagram of a computer system for implementing a server of the method, system, and device embodiments of the present application. The server shown in FIG. 7 is only an example, and should not impose any limitations on the functions and scope of use of the embodiments of the present application.

As shown in FIG. 7 , the computer system includes a central processing unit (CPU, Central Processing Unit) 701, which can be loaded into a random access memory according to a program stored in a read only memory (ROM, Read Only Memory) 702 or from a storage part 708 A program in (RAM, Random Access Memory) 703 executes various appropriate operations and processes. In the RAM 703, various programs and data necessary for system operation are also stored. The CPU 701 , the ROM 702 and the RAM 703 are connected to each other through a bus 704 . An input/output (I/O, Input/Output) interface 705 is also connected to the bus 704 .

The following components are connected to the I/O interface 705: an input section 706 including a keyboard, a mouse, etc.; an output section 707 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc. ; a storage section 708 including a hard disk and the like; and a communication section 709 including a network interface card such as a LAN (Local Area Network) card, a modem, and the like. The communication section 709 performs communication processing via a network such as the Internet. A drive 710 is also connected to the I/O interface 705 as needed. A removable medium 711, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is mounted on the drive 710 as needed so that a computer program read therefrom is installed into the storage section 708 as needed.

In particular, according to embodiments of the present disclosure, the processes described above with reference to the flowcharts may be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the method illustrated in the flowchart. In such an embodiment, the computer program may be downloaded and installed from the network via the communication portion 709 and/or installed from the removable medium 711 . When the computer program is executed by the central processing unit (CPU) 701, the above-mentioned functions defined in the method of the present application are performed. It should be noted that the computer-readable medium mentioned above in the present application may be a computer-readable signal medium or a computer-readable storage medium, or any combination of the above two. The computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus or device, or a combination of any of the above. More specific examples of computer readable storage media may include, but are not limited to, electrical connections with one or more wires, portable computer disks, hard disks, random access memory (RAM), read only memory (ROM), erasable Programmable read only memory (EPROM or flash memory), fiber optics, portable compact disk read only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing. In this application, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this application, however, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code therein. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the foregoing. A computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device . Program code embodied on a computer readable medium may be transmitted using any suitable medium including, but not limited to, wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.

Computer program code for performing the operations of the present application may be written in one or more programming languages, including object-oriented programming languages—such as Java, Smalltalk, C++, but also conventional Procedural programming language - such as the "C" language or similar programming language. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer, or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any kind of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (eg, using an Internet service provider through Internet connection).

The flowchart and block diagrams in the Figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code that contains one or more logical functions for implementing the specified functions executable instructions. It should also be noted that, in some alternative implementations, the functions noted in the blocks may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It is also noted that each block of the block diagrams and/or flowchart illustrations, and combinations of blocks in the block diagrams and/or flowchart illustrations, can be implemented in dedicated hardware-based systems that perform the specified functions or operations , or can be implemented in a combination of dedicated hardware and computer instructions.

The terms "first," "second," etc. are used to distinguish between similar objects, and are not used to describe or indicate a particular order or sequence.

The term "comprising" or any other similar term is intended to encompass a non-exclusive inclusion such that a process, method, article or device/means comprising a list of elements includes not only those elements but also other elements not expressly listed, or Also included are elements inherent to these processes, methods, articles or devices/devices.

So far, the technical solutions of the present invention have been described with reference to the preferred embodiments shown in the accompanying drawings, however, those skilled in the art can easily understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

Claims (10)

1. A contraband detection and identification method based on X-ray images is characterized by comprising the following steps:

determining a target image set according to preset contraband refinement category information;

performing data source expansion preprocessing on the target image set to obtain a training image set;

inputting the training image set to a detection and recognition training model for training to obtain a detection and recognition network model;

inputting the X-ray image to be detected into the detection and recognition network model to obtain confidence detection values of all detection areas in the X-ray image to be detected;

and determining a detection area corresponding to the confidence coefficient detection value in the preset confidence interval as a contraband area.

2. The method of claim 1, wherein the determining the target set of images according to the preset contraband refinement category information comprises:

acquiring a contraband image set;

and screening images consistent with preset contraband thinning category information in the contraband image set as a target image set.

3. The method of claim 2, wherein the acquiring the set of images of contraband comprises:

acquiring a first contraband image set;

and/or acquiring a second contraband image set; the first contraband image set is an X-ray image set containing contraband; the second contraband image set is a separate contraband image set;

the step of screening the images consistent with the preset contraband thinning category information in the contraband image set as a target image set comprises the following steps:

screening images consistent with preset contraband thinning category information in the first contraband image set as a target image set;

and/or screening images consistent with preset contraband thinning category information in the second contraband image set to serve as a target image set.

4. The method of claim 3, wherein the pre-processing the target image set by data source expansion to obtain a training image set comprises:

performing first preprocessing on a target image set screened in a first contraband image set to obtain a training image set;

and/or performing second preprocessing on the target image set screened in the second contraband image set to obtain a training image set.

5. The method of claim 4, wherein the first preprocessing is performed by:

processing the hue, brightness and saturation of each target image in the target image set at different degrees randomly to obtain a plurality of first intermediate images;

and respectively carrying out random cutting on the plurality of first intermediate images to obtain a training image set.

6. The method of claim 4, wherein the second preprocessing is performed by:

performing rotation and affine transformation operation on each target image in the target image set to obtain a second intermediate image;

and fusing the second intermediate image with the X-ray image without contraband according to a preset fusion rule to obtain a training image set.

7. The method of claim 1, wherein after determining a detection region corresponding to the confidence detection value at a preset confidence interval as a contraband region, the method further comprises:

comparing a confidence coefficient detection value corresponding to the contraband area with a preset detection value;

reserving a confidence detection value higher than or equal to a preset detection value and a corresponding detection area;

inputting a detection area image corresponding to a confidence coefficient detection value lower than a preset detection value into a convolutional neural network classification model so as to further judge the category of the contraband; wherein the preset detection value is any one value in a preset confidence interval.

8. The method of claim 7, wherein the convolutional neural network classification model is constructed by the following process:

acquiring a third contraband image set and an X-ray image without contraband; wherein the third set of contraband images is the same as the second set of contraband images;

randomly intercepting a partial area image of the X-ray image without the contraband;

performing third preprocessing on the third contraband image set and the partial area image of the X-ray image without the contraband to obtain a classification training image set;

and inputting the classification training image set into a classification training model for training to obtain a convolutional neural network classification model.

9. A system for contraband detection and identification based on X-ray images, comprising:

the first determining unit is used for determining a target image set according to preset contraband refinement category information;

the data expansion unit is used for carrying out data source expansion preprocessing on the target image set to obtain a training image set;

the training unit is used for inputting the training image set to a detection and recognition training model for training to obtain a detection and recognition network model;

the detection and identification unit is used for inputting the X-ray image to be detected into the detection and identification network model to obtain confidence detection values of all detection areas in the X-ray image to be detected;

and the second determining unit is used for determining the detection area corresponding to the confidence detection value in the preset confidence interval as the contraband area.

10. An apparatus, comprising:

at least one processor; and

a memory communicatively coupled to at least one of the processors; wherein,

the memory stores instructions executable by the processor for performing the method for contraband detection and identification based on X-ray images of any of claims 1-8.

Images