Image forgery analysis method and system, electronic equipment and storage medium
By text classification, amount area positioning and cutout processing of the bill screenshots uploaded by the driver, and authenticity detection is carried out in combination with the bill category model, the problem of bill forgery analysis is solved, automated review is realized, and efficiency is improved.
Patent Information
- Application Number
- CN202411617456.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-11-13
- Publication Date
- 2025-02-18
AI Technical Summary
In the prior art, the bill screenshots uploaded by drivers are easily forged, resulting in an increase in user complaints, making the review difficult, mainly relying on human review, and inefficient efficiency.
By obtaining the text information of the bill image, classifying text, positioning the amount area, cutting the image, and using the bill classification model corresponding to the bill category to perform authenticity detection, real-fake analysis is realized.
It reduces the difficulty of identifying forged images, narrows the area range for judging forged images, improves compatibility for identification of different billing categories, reduces the time for manual review, and improves audit efficiency.
Smart Images
Figure CN119478982A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of data analysis, and in particular to an image forgery analysis method, system, electronic equipment and storage medium. Background Art
[0002] Bill payment is the act of settling or clearing the amount on the bill through a payment tool or platform. When a consumer or business completes a purchase at a merchant, the merchant will provide a bill that lists the detailed transaction amount and other possible related fees. The process of paying a bill is to settle the amount on the bill to the merchant through a designated payment channel, such as a bank card, a third-party payment platform or other online payment methods.
[0003] In the existing screenshots of bills uploaded by drivers, there are cases where drivers arbitrarily forge amounts. In view of this situation, user complaints will be generated, which will reduce the user experience. In addition, the current audit method for forged cases is mainly through manual audit, which is difficult to audit. Summary of the invention
[0004] In view of the shortcomings of the prior art described above, the purpose of the present invention is to provide an image forgery analysis method, system, electronic device and storage medium, which are used to solve the problem that the driver may arbitrarily forge the amount of the bill screenshots uploaded by the driver in the prior art. In view of this situation, user complaints will be generated, which will reduce the user experience. In addition, the current audit method for forgery is mainly through human audit, which is difficult to audit.
[0005] To achieve the above-mentioned purpose and other related purposes, the present invention provides an image forgery analysis method, comprising the following steps: obtaining text information of a bill image; performing text classification on the text information of the bill image to obtain a bill category corresponding to the bill image; detecting the amount area of the bill image; after locating the amount area in the bill image, performing cutout processing on the amount area to obtain a cutout image; performing authenticity detection on the cutout image according to a bill classification model corresponding to the bill category of the bill image to obtain an authenticity result of the bill image.
[0006] In one embodiment of the present invention, obtaining text information of a bill image includes: extracting text content from the bill image through OCR (optical character recognition) to obtain the text information of the bill image.
[0007] In one embodiment of the present invention, text information of the bill image is subjected to text classification to obtain a bill category corresponding to the bill image, including: classifying the text information of the bill image through a text classification model to obtain a bill category corresponding to the bill image.
[0008] In one embodiment of the present invention, the amount area of the bill image is detected, including: positioning the amount area of the bill image through a detection model to obtain the position of the amount area in the bill image.
[0009] In one embodiment of the present invention, the extracted image is checked for authenticity based on a bill classification model corresponding to the bill category of the bill image to obtain the authenticity result of the bill image, including: searching for the bill classification model corresponding to the bill category based on the bill category corresponding to the bill image; and the extracted image is checked for authenticity through the bill classification model corresponding to the bill category to obtain the authenticity result.
[0010] In one embodiment of the present invention, the authenticity of the extracted image is detected by a bill classification model corresponding to the bill category to obtain an authenticity result, including: inputting the extracted image of the amount area corresponding to the bill image into the bill classification model; in the bill classification model, judging whether the key features of the extracted image of the amount area corresponding to the bill image are consistent with the key features corresponding to the normal amount image; wherein the key features include the shape of the decimal point corresponding to the amount, the font shape of the amount and the font color of the amount; if so, determining that the extracted image of the amount area corresponding to the bill image is in a normal state, and outputting an authenticity result that the bill image is true; if not, determining that the extracted image of the amount area corresponding to the bill image is in a forged state, and outputting an authenticity result that the bill image is false.
[0011] In one embodiment of the present invention, the bill classification model corresponding to each bill category is obtained by separately training the training data set corresponding to the bill category.
[0012] To achieve the above-mentioned purpose and other related purposes, the present invention also provides an image forgery analysis system, including: a text extraction unit, used to obtain text information of a bill image; a bill classification unit, used to perform text classification on the text information of the bill image to obtain a bill category corresponding to the bill image; an area positioning unit, used to detect the amount area of the bill image; an amount extraction unit, used to perform extraction processing on the amount area after locating the amount area in the bill image to obtain an extracted image; a model detection unit, used to perform authenticity detection on the extracted image according to a bill classification model corresponding to the bill category of the bill image to obtain an authenticity result of the bill image.
[0013] To achieve the above-mentioned purpose and other related purposes, the present invention further provides an electronic device, which includes: one or more processors; a storage device for storing one or more programs, and when the one or more programs are executed by one or more processors, the electronic device implements the aforementioned image forgery analysis method.
[0014] To achieve the above-mentioned object and other related objects, the present invention further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor of a computer, enables the computer to execute the aforementioned image forgery analysis method.
[0015] As described above, an image forgery analysis method, system, electronic device and storage medium of the present invention have the following beneficial effects: by adopting a multimodal processing method, text extraction is performed on bill images of multiple categories of bills, and various types of bills are classified by text classification, which can reduce the difficulty of subsequent forged image recognition. Then, by performing regional detection on the amount area in the bill image and performing a cutout process on the amount area, it is possible to reduce the regional range of the forged image during the forgery analysis. Then, by performing forged image recognition on the cutout image through the bill classification model corresponding to the bill category, the compatibility of the recognition process of different bill categories during bill forgery analysis can be effectively increased, so that multiple types of bills can be well classified. Through the recognition and processing of forged images by the bill classification model corresponding to the bill category, it is also possible to better identify the situation where there are slight changes in the text characters under the corresponding bill category, and then effectively identify the forged images existing in each type of bill, which can more conveniently expand the recognition process of slight changes in multiple types of bills, thereby reducing the time of manual review and improving the review efficiency. BRIEF DESCRIPTION OF THE DRAWINGS
[0016] Figure 1 A schematic diagram of a flow chart of an image forgery analysis method provided by an embodiment of the present invention.
[0017] Figure 2 Shown is a schematic flow chart of an image forgery analysis process provided by an embodiment of the present invention.
[0018] Figure 3 Shown is a structural block diagram of an image forgery analysis system provided by an embodiment of the present invention.
[0019] Figure 4 Shown is a structural schematic diagram of an electronic device according to an embodiment of the present invention.
[0020] Component number description
[0021] Electronic device 1; forgery analysis system 11; memory 12; processor 13; text extraction unit 111; bill classification unit 112; area positioning unit 113; amount extraction unit 114; model detection unit 115. DETAILED DESCRIPTION
[0022] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict.
[0023] It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present invention, and thus the drawings only show components related to the present invention rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.
[0024] In the following description, numerous details are discussed to provide a more thorough explanation of the embodiments of the present invention. However, it is obvious to those skilled in the art that the embodiments of the present invention can be implemented without these specific details. In other embodiments, well-known structures and devices are shown in the form of block diagrams rather than in detail to avoid making the embodiments of the present invention difficult to understand.
[0025] See also Figure 1 The present invention provides an image forgery analysis method, comprising the following steps:
[0026] Step S10: Acquire text information of the bill image.
[0027] Step S20: Perform text classification on the text information of the bill image to obtain the bill category corresponding to the bill image.
[0028] Step S30: Detect the amount area of the bill image.
[0029] Step S40: After locating the amount area in the bill image, perform cropping processing on the amount area to obtain a cropped image.
[0030] Step S50: According to the bill classification model corresponding to the bill category of the bill image, the extracted image is checked for authenticity to obtain the authenticity result of the bill image.
[0031] It is not difficult to find through the above steps that in the process of image forgery analysis such as bill images, the classification of bill categories can be achieved based on the text information of the bill image to reduce the difficulty of identifying subsequent forged images. Specifically, by detecting the specific amount area in the bill image, the corresponding amount area in the bill image is determined. The amount area is then cut out, so that only the cutout amount area can be checked for authenticity, so as to narrow the area range for judging forged images. Furthermore, the corresponding bill classification model can be selected for authenticity detection based on the bill category of the bill image, which can effectively increase the compatibility of the recognition process of different bill types.
[0032] Figure 1 The flowchart of the image forgery analysis method in an exemplary embodiment of the present application is shown, including steps S10 to S50. Figure 1 The technical solution of this application will be described in detail.
[0033] First, step S10 is executed to obtain text information of the bill image.
[0034] In this embodiment, the bills corresponding to the bill image include multiple types, such as WeChat bills, Alipay bills, etc. By obtaining the text information of the bill image, it is convenient to further determine the bill type corresponding to the bill image according to the text information, so as to detect the authenticity of the extracted image through the bill classification model corresponding to the bill type, thereby realizing effective recognition of different types of bills.
[0035] In step S10, the text information of the bill image is obtained, including: extracting text content from the bill image through OCR to obtain the text information of the bill image.
[0036] In this embodiment, the bill image can be a document obtained in the form of a screenshot. Specifically, by taking screenshots of multiple types of bills, a bill image of the corresponding bill is obtained. After obtaining the bill image, the bill image can be recognized by OCR (optical character recognition), and then the text information in the bill image can be obtained, so as to further determine the bill type through the text information. Among them, OCR is the process of translating shapes into computer text using character recognition methods; for printed characters in the bill image, an optical method can be used to convert the text in the bill image into a black and white dot matrix image file, and the text in the image can be converted into a text format through recognition software to form text information.
[0037] Next, step S20 is executed to perform text classification on the text information of the bill image to obtain the bill category corresponding to the bill image.
[0038] In this embodiment, after obtaining the text information in the bill image, the bill category corresponding to the text information, that is, the bill category corresponding to the bill image, can be determined based on the recognized text content. This enables subsequent authenticity detection to be performed directly through the bill classification model corresponding to the bill category, thereby reducing the difficulty of subsequent forged image recognition.
[0039] In step S20, text classification is performed on the text information of the bill image to obtain the bill category corresponding to the bill image, including: classifying the text information of the bill image through a text classification model to obtain the bill category corresponding to the bill image.
[0040] In this embodiment, when determining the bill category of the bill image based on the text information of the bill image, the text information is input into the text classification model, and the bill category corresponding to the bill image can be directly determined by the text classification model. Specifically, the text classification model can be a multi-classification model, which can be obtained by inputting text information features and corresponding bill category labels as training data sets for model training. During use, by inputting the text information corresponding to the bill image, the bill category label corresponding to the text information can be obtained. Specifically, the bill category can include WeChat bills, Alipay bills, etc.
[0041] Next, step S30 is executed to detect the amount area of the bill image.
[0042] In this embodiment, when checking the authenticity of the bill image, the detection range of the image area of the bill image can be reduced. That is, only the amount area can be checked for authenticity, and before that, the amount area of the bill image needs to be checked, so that after the location of the amount area is located, the amount area can be processed and the authenticity detection can be performed.
[0043] In step S30, the amount area of the bill image is detected, including: positioning the amount area of the bill image through a detection model to obtain the position of the amount area in the bill image.
[0044] In this embodiment, when detecting the amount area of the bill image, the position of the amount area in the bill image can be located through the detection model to obtain the positioning information of the amount area, thereby facilitating subsequent operations such as amount area clipping and authenticity detection based on the positioning information of the amount area.
[0045] Next, step S40 is executed. After the amount area in the bill image is located, the amount area is cut out to obtain a cut-out image.
[0046] In this embodiment, after the amount area in the bill image is detected by using the detection model, the amount area in the bill image is first located during the cutout process, and after locking the amount area, the amount area is cut out to obtain a cutout image. Thus, the authenticity of the bill image corresponding to the cutout image can be further determined in the bill classification model through the cutout image.
[0047] Next, step S50 is executed to perform authenticity detection on the extracted image according to the bill classification model corresponding to the bill category of the bill image to obtain the authenticity result of the bill image.
[0048] In this embodiment, when determining the authenticity of the bill image, the cutout image obtained by the cutout processing can be input into the bill classification model corresponding to the bill category to further determine the authenticity of the bill image. After classifying multiple types of bills by the text information of the bill image, forged image recognition is performed on each category of bills according to the bill classification model corresponding to the bill category, which can effectively identify the forged images existing in each category of bills and effectively increase the compatibility of the recognition process of different bill types. Moreover, by utilizing the recognition mechanism for multiple types of bills, the time for manual review can be effectively reduced during the forgery analysis of bill images, and the review efficiency can be improved.
[0049] In step S50, according to the bill classification model corresponding to the bill category of the bill image, the extracted image is subjected to authenticity detection to obtain the authenticity result of the bill image, including the following steps:
[0050] Step S501: searching for a bill classification model corresponding to the bill category according to the bill category corresponding to the bill image;
[0051] Step S502: Authenticity detection is performed on the extracted image using the bill classification model corresponding to the bill category to obtain an authenticity result.
[0052] In this embodiment, before performing the authenticity detection of the bill image, it is necessary to first determine the bill classification model to be used. Specifically, the bill category corresponding to the bill image is obtained based on the text information of the bill image. According to the bill category, the bill classification model corresponding to the bill category is further found. Next, the obtained bill classification model is used to perform authenticity detection on the extracted image corresponding to the bill image, so that the authenticity of the bill image can be accurately determined, that is, the authenticity result can be obtained.
[0053] In step S502, the authenticity of the extracted image is detected by using the bill classification model corresponding to the bill category to obtain an authenticity result, including the following steps:
[0054] Step S5021: inputting the cutout image of the amount area corresponding to the bill image into the bill classification model;
[0055] Step S5022: In the bill classification model, determine whether the key features of the clipped image of the amount area corresponding to the bill image are consistent with the key features corresponding to the normal amount image; wherein the key features include the shape of the decimal point corresponding to the amount, the shape of the font of the amount, and the font color of the amount;
[0056] Step S5023: If yes, determine that the cutout image of the amount area corresponding to the bill image is in a normal state, and output a result indicating that the bill image is true;
[0057] Step S5024: If not, determine that the cut-out image of the amount area corresponding to the bill image is in a forged state, and output a authenticity result indicating that the bill image is fake.
[0058] In this embodiment, when the bill classification model is used to detect the authenticity of the extracted image, by inputting the extracted image of the amount area corresponding to the bill image into the bill classification model, the bill classification model can detect the key features of the extracted image, that is, determine whether the key features of the extracted image are consistent with the key features of the normal amount image. If they are consistent, it means that there is no abnormality in the extracted image, proving that the bill image is a normal bill; if they are inconsistent, it means that there is an abnormality in the extracted image, and the authenticity result of the bill image being forged or false is output.
[0059] For example, when judging whether the shape of the decimal point corresponding to the amount in the extracted image is forged, the shape of the decimal point corresponding to the amount in the extracted image is compared with the shape of the decimal point corresponding to the amount in the normal amount image (such as the decimal point shape is divided into square dots and round dots when compared). When the two are consistent, it means that the decimal point shape corresponding to the amount is not forged. Similarly, when judging whether the font shape of the amount in the extracted image is forged, the font shape of the amount in the extracted image is compared with the font shape of the amount in the normal amount image. When the two are consistent, it means that the font shape of the amount is not forged. When judging whether the font color of the amount in the extracted image is forged, the font color of the amount in the extracted image is compared with the font color of the amount in the normal amount image. When the two are consistent, it means that the font color of the amount is not forged.
[0060] In one embodiment of the present invention, the bill classification model corresponding to each bill category is obtained by training the training data set corresponding to the bill category. In the process of training the bill classification model using the training data set, the bill classification model can be obtained by simultaneously inputting the amount area picture of the normal amount image, the amount area picture of the forged amount image, and the authenticity result into the basic model for training. Specifically, the bill classification model can be a classification model using a multi-layer neural network classification method.
[0061] See also Figure 2 In the process of image forgery identification of bill images, multiple categories of bill images (such as WeChat bill images, Alipay bill images, etc.) are input, and the text content in the image is identified and extracted through OCR. The bill image is classified according to the extracted text content through the text classification model. Different bill images belong to different categories. For example, the image of the WeChat bill is classified into the bill category of the WeChat bill, and the screenshot of the Alipay bill is classified into the bill category of the Alipay bill. After classification, it is convenient to further sub-classify the images of each bill category. Specifically, the amount area of the input bill image is detected by the detection model to obtain the positioning information of the amount area. According to the positioning information, the corresponding amount area is located for clipping processing, and the clipped image obtained by the clipping processing is further detected for authenticity through the classification model. And when the clipped image is judged to be forged, a normal result will be output. If the clipped image is judged to be normal, a normal authenticity result will also be output accordingly.
[0062] In a preferred embodiment of the present invention, for example, when there are three bill images, that is, Figure a is an Alipay bill, Figure b is a WeChat bill, and Figure c is a WeChat bill forgery, the three bill images are first obtained, and the text content in the three images is identified by OCR technology. The identified text content is further classified by a text classification model, and the categories of the three pictures can be obtained in sequence: Alipay bill, WeChat bill, WeChat bill. Next, the three bill images are located by the detection model for the amount area, and the extracted pictures of the amount position in each picture are obtained respectively, for example, the extracted pictures of the amounts of "-￥17.94", "-30.00", and "-93.00" are obtained respectively. After obtaining the extracted pictures, the bill classification models of the bill categories corresponding to the extracted pictures are Alipay bill classification model, WeChat bill classification model, and WeChat bill classification model, respectively, and the corresponding authenticity results can be accurately output: normal, normal, and forgery.
[0063] like Figure 3As shown, in one embodiment of the present invention, the present invention also provides an image forgery analysis system 11, including: a text extraction unit 111, used to obtain text information of a bill image; a bill classification unit 112, used to perform text classification on the text information of the bill image to obtain a bill category corresponding to the bill image; an area positioning unit 113, used to detect the amount area of the bill image; an amount extraction unit 114, used to perform extraction processing on the amount area after locating the amount area in the bill image to obtain an extracted image; a model detection unit 115, used to perform authenticity detection on the extracted image according to a bill classification model corresponding to the bill category of the bill image to obtain an authenticity result of the bill image.
[0064] In this embodiment, in the image forgery analysis system 11 of the present invention, the text content extraction in the bill image, such as WeChat bills, Alipay bills, etc., can be realized by the text extraction unit 111. The bill classification unit 112 can further classify the bill category according to the text information in the bill image to obtain the accurate bill category corresponding to the bill image. Then, the area positioning unit 113 can detect the amount area in the bill image, and the area range of the forged image can be narrowed by positioning the area. And after determining the amount area, the amount extraction unit 114 performs a cutout process on the located amount area to obtain a cutout image of the amount area. Further, the model detection unit 115 can realize the input of the cutout image into the bill classification model corresponding to the bill type of the cutout image for authenticity detection, and then accurately determine whether the bill image is forged, which effectively increases the compatibility of the identification process of different bill categories during bill forgery analysis, and the division of bill categories according to text classification can also effectively reduce the difficulty of forged image recognition.
[0065] It should be noted that the image forgery analysis system 11 provided in the above embodiment and the image forgery analysis method provided in the above embodiment belong to the same concept, wherein the specific manner in which each module and unit performs the operation has been described in detail in the method embodiment, and will not be repeated here. In actual application, the image forgery analysis system 11 provided in the above embodiment can allocate the above functions to different functional modules as needed, that is, divide the internal structure of the device into different functional modules to complete all or part of the functions described above, and this is not limited here.
[0066] See also Figure 4 The electronic device 1 may include a memory 12, a processor 13 and a bus, and may also include a computer program stored in the memory 12 and executable on the processor 13, such as an image forgery analysis program.
[0067] Among them, the memory 12 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, mobile hard disk, multimedia card, card-type memory (for example: SD or DX memory, etc.), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 12 can be an internal storage unit of the electronic device 1, such as a mobile hard disk of the electronic device 1. In other embodiments, the memory 12 can also be an external storage device of the electronic device 1, such as a plug-in mobile hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the electronic device 1. Further, the memory 12 can also include both an internal storage unit of the electronic device 1 and an external storage device. The memory 12 can not only be used to store application software and various types of data installed in the electronic device 1, such as code for image forgery analysis, etc., but also can be used to temporarily store data that has been output or is to be output.
[0068] In some embodiments, the processor 13 may be composed of an integrated circuit, for example, a single packaged integrated circuit, or a plurality of packaged integrated circuits with the same or different functions, including one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and combinations of various control chips. The processor 13 is the control core (Control Unit) of the electronic device 1, and uses various interfaces and lines to connect various components of the entire electronic device 1, and executes or executes programs or modules (such as image forgery analysis programs, etc.) stored in the memory 12, and calls data stored in the memory 12 to execute various functions of the electronic device 1 and process data.
[0069] The processor 13 executes the operating system and various installed application programs of the electronic device 1. The processor 13 executes the application programs to implement the steps in the above-mentioned image forgery analysis method.
[0070] Exemplarily, the computer program may be divided into one or more modules, which are stored in the memory 12 and executed by the processor 13 to complete the present application. The one or more modules may be a series of computer program instruction segments capable of completing specific functions, which are used to describe the execution process of the computer program in the electronic device 1. For example, the computer program may be divided into a text extraction unit 111, a bill classification unit 112, a region positioning unit 113, an amount extraction unit 114, and a model detection unit 115.
[0071] The above-mentioned integrated unit implemented in the form of a software function module can be stored in a computer-readable storage medium, and the computer-readable storage medium can be non-volatile or volatile. The above-mentioned software function module is stored in a storage medium, including several instructions for enabling a computer device (which can be a personal computer, a computer device, or a network device, etc.) or a processor to perform part of the functions of the image forgery analysis method described in each embodiment of the present application.
[0072] In summary, the image forgery analysis method, system, electronic device and storage medium disclosed in the present invention can reduce the difficulty of subsequent forged image recognition by adopting a multimodal processing method to extract text from bill images of multiple categories of bills and classify various bills through text classification. Then, by performing regional detection on the amount area in the bill image and performing a cutout process on the amount area, it is possible to reduce the regional range of the forged image during the forgery analysis. Then, by performing forged image recognition on the cutout image through the bill classification model corresponding to the bill category, the compatibility of the recognition process of different bill categories during the bill forgery analysis can be effectively increased, so that multiple types of bills can be well classified. Through the recognition and processing of the forged image by the bill classification model corresponding to the bill category, it is also possible to better identify the situation where there are slight changes in the text characters under the corresponding bill category, and then effectively identify the forged images existing in each type of bill, and it is possible to more conveniently expand the recognition process of slight changes in multiple types of bills, thereby reducing the time of manual review and improving the review efficiency. Therefore, the present invention effectively overcomes the various shortcomings in the prior art and has a high industrial utilization value.
[0073] The above embodiments are merely illustrative of the principles and effects of the present invention, and are not intended to limit the present invention. Anyone familiar with the art may modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by a person of ordinary skill in the art without departing from the spirit and technical concept disclosed by the present invention shall still be covered by the claims of the present invention.
Claims
1. A method for analyzing image forgery, characterized in that: The following steps are involved: Get the text information of the bill image; Performing text classification on the text information of the bill image to obtain the bill category corresponding to the bill image; Detecting the amount area of the bill image; After locating the amount area in the bill image, performing a cutout process on the amount area to obtain a cutout image; According to the bill classification model corresponding to the bill category of the bill image, the extracted image is subjected to authenticity detection to obtain the authenticity result of the bill image.
2. The image forgery analysis method according to claim 1, characterized in that: Get text information of bill image, including: The text content of the bill image is extracted through OCR to obtain the text information of the bill image.
3. The image forgery analysis method according to claim 1, characterized in that: Performing text classification on the text information of the bill image to obtain the bill category corresponding to the bill image includes: The text information is used to classify the bill image through a text classification model to obtain the bill category corresponding to the bill image.
4. The image forgery analysis method according to claim 1, characterized in that: Detecting the amount area of the bill image includes: The amount area of the bill image is located through the detection model to obtain the position of the amount area in the bill image.
5. The image forgery analysis method according to claim 1, characterized in that: According to the bill classification model corresponding to the bill category of the bill image, the extracted image is subjected to authenticity detection to obtain the authenticity result of the bill image, including: According to the bill category corresponding to the bill image, searching for a bill classification model corresponding to the bill category; The authenticity of the extracted image is detected by using the bill classification model corresponding to the bill category to obtain the authenticity result.
6. The image forgery analysis method according to claim 5, characterized in that: The authenticity of the extracted image is detected by using the bill classification model corresponding to the bill category to obtain the authenticity result, including: Inputting the cutout image of the amount area corresponding to the bill image into the bill classification model; In the bill classification model, it is determined whether the key features of the cutout image of the amount area corresponding to the bill image are consistent with the key features corresponding to the normal amount image; wherein the key features include the shape of the decimal point corresponding to the amount, the shape of the font of the amount, and the font color of the amount; If so, determine that the cutout image of the amount area corresponding to the bill image is in a normal state, and output the authenticity result that the bill image is authentic; If not, it is determined that the cut-out image of the amount area corresponding to the bill image is in a forged state, and the authenticity result indicating that the bill image is fake is output.
7. The image forgery analysis method according to claim 1, characterized in that: The bill classification model corresponding to each bill category is obtained by separately training the training data set corresponding to the bill category.
8. An image forgery analysis system, characterized in that: include: A text extraction unit, used for obtaining text information of the bill image; A bill classification unit, used to perform text classification on the text information of the bill image to obtain a bill category corresponding to the bill image; A region positioning unit, used for detecting the amount region of the bill image; An amount extraction unit is used to perform image extraction processing on the amount area in the bill image after locating the amount area in the bill image to obtain an extracted image; The model detection unit is used to perform authenticity detection on the extracted image according to the bill classification model corresponding to the bill category of the bill image to obtain the authenticity result of the bill image.
9. An electronic device, characterized in that: The electronic device comprises: one or more processors; A storage device for storing one or more programs, which, when executed by the one or more processors, enables the electronic device to implement the image forgery analysis method according to any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor of a computer, the computer is caused to execute the image forgery analysis method according to any one of claims 1 to 7.