Automatic audit method, device, equipment and storage medium based on artificial intelligence

Abstract

The present invention relates to the field of artificial intelligence, and discloses an automatic audit method, device, equipment and storage medium based on artificial intelligence. The present invention uses a cluster mode and a preset artificial intelligence recognition model to quickly compare the certificate information with the corresponding information of the information sheet, completes the automatic audit of business data, and improves the accuracy of the automatic audit data and the efficiency of the audit. The method of the present invention includes: obtaining a pre-processed certificate image; segmenting the pre-processed certificate image according to preset rules to obtain a segmented image; comparing the text data of the segmented image with the target data according to the cluster mode to obtain a comparison result, and the target data is the data associated in the preset artificial intelligence recognition model; and automatically feeding back the audit result according to the comparison result.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence, and in particular to an automatic review method, device, equipment and storage medium based on artificial intelligence. Background Art

[0002] With the continuous advancement and development of science and technology, the authenticity of user information has received more and more attention. Generally, users will enter their ID information into a terminal or server, and then the terminal or server will call or process the entered information, which is convenient for the terminal or server to operate the system. Therefore, only by ensuring the authenticity of user information can the operation of the terminal or server be facilitated.

[0003] In the review of information sheets, users need to upload their ID information to compare with the user information sheet that has been entered into the system, compare the same user information, and confirm the authenticity of the information to ensure the security of the user's information. Currently, after the user uploads the ID information, the information sheet will be manually reviewed to complete the proofreading of the user's basic information.

[0004] However, the manual review of information sheets wastes a lot of review time. At the same time, the manual review method may result in incorrect information review to a certain extent, resulting in the information sheet failing the review. Summary of the invention

[0005] The present invention provides an automatic audit method, device, equipment and storage medium based on artificial intelligence, which are used to solve the problem of low efficiency in auditing information sheets and improve the accuracy and efficiency of auditing information sheet detection.

[0006] The first aspect of an embodiment of the present invention provides an automatic review method based on artificial intelligence, comprising: obtaining a preprocessed document image; segmenting the preprocessed document image according to preset rules to obtain a segmented image; comparing the text data of the segmented image with target data according to a clustering mode to obtain a comparison result, wherein the target data is data associated with a preset artificial intelligence recognition model; and automatically feeding back the review result based on the comparison result.

[0007] Optionally, in a first implementation method of the first aspect of the embodiment of the present invention, the segmented image is matched with a preset artificial intelligence recognition model, each preset artificial intelligence recognition model corresponds to a type of the segmented image; the segmented image is transmitted to the matched preset artificial intelligence recognition model; the text data on the segmented image is extracted; in the preset artificial intelligence recognition model, the text data is compared with the target data to obtain a comparison result, and the target data is the associated data in the corresponding preset artificial intelligence recognition model.

[0008] Optionally, in a second implementation method of the first aspect of the embodiment of the present invention, a text area on the segmented image is selected; text lines in the text area are segmented to obtain text characters; a standard template is established based on the text characters; the standard template is matched with a preset template to extract text data, and the text data is the data on the preset template.

[0009] Optionally, in a third implementation manner of the first aspect of the embodiment of the present invention, an uploaded document image is obtained; and the document image is normalized to obtain a preprocessed document image.

[0010] Optionally, in a fourth implementation method of the first aspect of the embodiment of the present invention, the ID image is converted into a grayscale image; the grayscale image is converted into a black and white image; the black and white image is gradient sharpened to obtain a sharpened image; discrete noise is removed from the sharpened image to obtain a processed ID image.

[0011] Optionally, in a fifth implementation of the first aspect of the embodiment of the present invention, feature pixels on the preprocessed document image are extracted; different feature pixels are assigned different feature labels; and the feature pixels with the same feature label are integrated to obtain a segmented image.

[0012] Optionally, in a sixth implementation method of the first aspect of the embodiment of the present invention, when the comparison result is that the text data and the target data are exactly the same, the words "review successful" are automatically fed back; when the comparison result is that the text data and the target data are not exactly the same, the words "review failed" are automatically fed back.

[0013] The second aspect of an embodiment of the present invention provides an automatic review device based on artificial intelligence, including: an acquisition unit, used to acquire a preprocessed document image; a processing unit, used to segment the preprocessed document image according to preset rules to obtain a segmented image; a comparison unit, used to compare the text data of the segmented image with target data according to a cluster mode to obtain a comparison result, wherein the target data is data associated with a preset artificial intelligence recognition model; and a feedback unit, used to automatically feedback the review result based on the comparison result.

[0014] Optionally, in a first implementation method of the second aspect of the embodiment of the present invention, the comparison unit includes: a matching module, used to match the segmented image with a preset artificial intelligence recognition model, each preset artificial intelligence recognition model corresponds to a type of the segmented image; a transmission module, used to transmit the segmented image to the matched preset artificial intelligence recognition model; an extraction module, used to extract text data on the segmented image; a comparison module, used to compare the text data with target data in the preset artificial intelligence recognition model to obtain a comparison result, and the target data is the associated data in the corresponding preset artificial intelligence recognition model.

[0015] Optionally, in a second implementation method of the second aspect of the embodiment of the present invention, the extraction module is specifically used to: select a text area on the segmented image; segment the text lines in the text area to obtain text characters; establish a standard template based on the text characters; match the standard template with a preset template to extract text data, wherein the text data is the data on the preset template.

[0016] Optionally, in a third implementation of the second aspect of the embodiment of the present invention, the acquisition unit includes: an acquisition module, used to acquire an uploaded document image; and a preprocessing module, used to normalize the document image to obtain a preprocessed document image.

[0017] Optionally, in a fourth implementation method of the second aspect of the embodiment of the present invention, the preprocessing module is specifically used to: convert the ID image into a grayscale image; convert the grayscale image into a black and white image; perform gradient sharpening on the black and white image to obtain a sharpened image; remove discrete noise from the sharpened image to obtain a processed ID image.

[0018] Optionally, in a fifth implementation of the second aspect of the embodiment of the present invention, the processing unit is specifically used to: extract feature pixels on the preprocessed document image; assign different feature labels to different feature pixels; and integrate the feature pixels with the same feature label to obtain a segmented image.

[0019] Optionally, in a sixth implementation method of the second aspect of the embodiment of the present invention, the feedback unit is specifically used to: when the comparison result is that the text data and the target data are exactly the same, automatically feedback the words "review successful"; when the comparison result is that the text data and the target data are not exactly the same, automatically feedback the words "review failed".

[0020] The third aspect of an embodiment of the present invention provides an automatic review device based on artificial intelligence, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the automatic review method based on artificial intelligence described in any one of the above-mentioned embodiments is implemented.

[0021] A fourth aspect of an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the method described in the first aspect above.

[0022] It can be seen from the above technical solutions that the embodiments of the present invention have the following advantages:

[0023] The present invention utilizes a cluster mode and a preset artificial intelligence recognition model to quickly compare the certificate information with the corresponding information on the information sheet, completes the automatic review of business data, and improves the accuracy of the automatic review data and the efficiency of the review. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 A schematic diagram of an embodiment of an automatic audit method based on artificial intelligence of the present invention;

[0025] Figure 2 It is a schematic diagram of another embodiment of the automatic audit method based on artificial intelligence of the present invention;

[0026] Figure 3 A schematic diagram of an embodiment of an automatic audit device based on artificial intelligence of the present invention;

[0027] Figure 4 It is a schematic diagram of another embodiment of the automatic audit device based on artificial intelligence of the present invention;

[0028] Figure 5 A schematic diagram of an embodiment of an automatic audit device based on artificial intelligence of the present invention. DETAILED DESCRIPTION

[0029] The present invention provides an automatic audit method, device, equipment and storage medium based on artificial intelligence, which are used to solve the problem of low efficiency in auditing information sheets and improve the accuracy and efficiency of auditing information sheet detection.

[0030] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0031] The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0032] See also Figure 1 In one embodiment of the present invention, an automatic audit method based on artificial intelligence includes:

[0033] 101. Obtain the pre-processed document image.

[0034] The server obtains the pre-processed certificate image. There are many types of certificate images here, such as insurance policies, and there are many types of insurance policies, but there is no limitation on the certificate images obtained here.

[0035] Take the automatic review of insurance policy information as an example: there are many types of insurance policies, such as auto insurance, property insurance, accident insurance, etc. These types of policy models are shared, but different user information will be mounted in the policy model. For example, the policyholder information is mounted in the external insurance policy model. The policyholder information here includes the policyholder's name, age, birthday, and ID number. When verifying the policy information, the policy information must match the ID card information so that the server can complete the review. The vehicle policy model will have vehicle information and owner information, and the accident insurance policy model will have beneficiary information. The policy model obtained in this way must include the information on the driver's license or vehicle registration.

[0036] It is understandable that the execution subject of the present invention may be a terminal or a server, which is not specifically limited herein. The embodiment of the present invention is described by taking a server as the execution subject as an example.

[0037] 102. Segment the preprocessed document image according to preset rules to obtain a segmented image.

[0038] The server segments the preprocessed document image according to preset rules to obtain a segmented image. The server preprocesses the obtained document image. Preprocessing here refers to the operation performed on the image at the lowest abstract level. The purpose of preprocessing is to improve the image data of the document image, suppress unnecessary deformation or enhance certain image features that are important for subsequent processing.

[0039] Based on artificial intelligence image recognition technology, the server uses convolution and upsampling to extract feature pixels on the processed document image, and uses pooling to optimize the feature pixels. The upsampling here is deconvolution. Deconvolution is similar to convolution. Both are multiplication and addition operations. Convolution is one-to-many convolution, while deconvolution is many-to-one convolution. The forward and backward propagation of deconvolution only requires reversing the forward and backward propagation of convolution, and then using pooling to optimize the feature pixels.

[0040] 103. According to the clustering mode, the text data of the segmented image is compared with the target data to obtain a comparison result, and the target data is the data associated with the preset artificial intelligence recognition model.

[0041] According to the cluster mode, the server compares the text data of the segmented image with the target data to obtain a comparison result. The target data is the data associated with the preset artificial intelligence recognition model.

[0042] The server uses load balancing (LB) to distribute text information to different preset target data models. LB is a cluster technology that distributes specific services such as network services and network traffic to multiple network devices such as servers, firewalls, etc. or multiple links, that is, balancing and distributing the load (work tasks) to multiple operation units for execution, thereby improving business processing capabilities and ensuring high business reliability.

[0043] Here, the server uses load balancing to match the segmented image with multiple preset artificial intelligence recognition models. Each preset artificial intelligence recognition model corresponds to a category of segmented images, and there are multiple preset artificial intelligence recognition models of different types. If the segmented image matches the preset artificial intelligence recognition model, the segmented image is transmitted to the preset artificial intelligence recognition model that matches it. In the preset artificial intelligence recognition model, the transmitted segmented image is processed, and the text data on the segmented image is extracted. The extracted text data is then compared with the target data to realize data verification of the certificate information and the information sheet.

[0044] 104. Automatically feedback the audit results based on the comparison results.

[0045] The server automatically feedbacks the audit result based on the comparison result between the text data and the target data. The server automatically feedbacks the audit result based on the judgment result of whether the text data and the target data are the same. When the comparison result is that the text data and the target data are exactly the same, the server automatically feedbacks the words "audit successful"; when the comparison result is that the text data and the target data are not exactly the same, the server automatically feedbacks the words "audit failed".

[0046] The present invention utilizes a cluster mode and a preset artificial intelligence recognition model to quickly compare the certificate information with the corresponding information on the information sheet, completes the automatic review of business data, and improves the accuracy of the automatic review data and the efficiency of the review.

[0047] See also Figure 2 Another embodiment of the automatic audit method based on artificial intelligence in the embodiment of the present invention includes:

[0048] 201. Obtain the uploaded document image.

[0049] The server obtains the uploaded certificate image. There are many types of certificate images here, such as insurance policies, and there are many types of insurance policies, but there is no limitation on the certificate images obtained here.

[0050] Take the automatic review of insurance policy information as an example: there are many types of insurance policies, such as auto insurance, property insurance, accident insurance, etc. These types of policy models are shared, but different user information will be mounted in the policy model. For example, the policyholder information is mounted in the external insurance policy model. The policyholder information here includes the policyholder's name, age, birthday, and ID number. When verifying the policy information, the policy information must match the ID card information so that the server can complete the review. The vehicle policy model will have vehicle information and owner information, and the accident insurance policy model will have beneficiary information. The policy model obtained in this way must include the information on the driver's license or vehicle registration.

[0051] It is understandable that the execution subject of the present invention may be a terminal or a server, which is not specifically limited herein. The embodiment of the present invention is described by taking a server as the execution subject as an example.

[0052] 202. Normalize the document image to obtain a preprocessed document image.

[0053] The server normalizes the document image to obtain a preprocessed document image. Specifically: the server converts the document image into a grayscale image; the server converts the grayscale image into a black-and-white image; the server performs gradient sharpening on the black-and-white image to obtain a sharpened image; the server removes discrete noise from the sharpened image to obtain a processed document image.

[0054] The server will perform normalization processing on the obtained document image. The processing here refers to the operation performed on the image at the lowest abstract level. The purpose of the processing is to improve the image data of the document image, suppress unnecessary deformation or enhance certain image features that are important for subsequent processing, so that the server can better extract information from the document.

[0055] It should be noted that the server converts the document image into a grayscale image in three methods: maximum method, average method and ratio method. Maximum method: The server obtains each pixel value in the document image and obtains the RGB component value of each pixel value, which are: R (red), G (green), B (blue), and then the server takes the largest of the three components as the pixel component value. Average method: The server takes the average of the three components R (red), G (green), and B (blue) as the pixel component value. Ratio method: Assuming that the three components of the current pixel are R (red), G (green), and B (blue), the server uses the following formula to obtain the converted pixel component value: 0.30×R+0.59×G+0.11×B. The method by which the server converts the document image into a grayscale image is not limited here.

[0056] The server uses a threshold conversion algorithm to convert the grayscale image into a black and white image, and performs gradient sharpening on the black and white image to make the image more prominent and easier to analyze. The specific algorithm is as follows: the server calculates the absolute value of the difference between the current pixel value and its next pixel value, plus the absolute value of the difference between the current pixel value and the current pixel value in the next row. If the result is greater than the threshold, the server sets the current pixel value to the calculated result. The server sharpens the black and white image and then removes discrete noise from it. It uses a recursive method to find out whether there are black pixels in the eight directions of the current pixel. The server sets the continuous length of the pixel to 15. If the server uses a recursive method to obtain a continuous pixel value greater than 15, the current pixel is not noise. Otherwise, it is considered to be noise, and the current pixel is set to a white pixel. In this way, the processed document image is obtained.

[0057] 203. Segment the preprocessed document image according to preset rules to obtain a segmented image.

[0058] The server segments the preprocessed document image according to preset rules to obtain a segmented image. Specifically: the server extracts feature pixels on the preprocessed document image; the server assigns different feature labels to different feature pixels; the server integrates feature pixels with the same feature label to obtain a segmented image.

[0059] There are many types of ID images uploaded by users. ID images contain user information. Different user information needs to be classified and processed. The server needs to use image segmentation technology to segment the image. According to the image's color, shape, grayscale, texture and other features, the image is divided into several non-overlapping sub-areas with their own characteristics. Then, these cut sub-areas are analyzed and classified through image classification technology in order to compare with the same type of target data.

[0060] Based on artificial intelligence image recognition technology, the server uses convolution and upsampling to extract feature pixels on the processed document image, and uses pooling to optimize the feature pixels. The upsampling here is deconvolution. Deconvolution is similar to convolution. Both are multiplication and addition operations. Convolution is one-to-many convolution, while deconvolution is many-to-one convolution. The forward and backward propagation of deconvolution only requires reversing the forward and backward propagation of convolution, and then using pooling to optimize the feature pixels.

[0061] For each feature pixel i, it has a category label x i There is also the corresponding observation value y i , the server regards each pixel as a node and the relationship between pixels as an edge, thus forming a conditional random field, and the server observes the variable y i To infer the feature category label x corresponding to pixel i i . The conditional random field here conforms to the Gibbs distribution:

[0062]

[0063] Where E(x|I) is the energy function. For simplicity, the global observation I is omitted, and then:

[0064]

[0065] In the formula:

[0066] Here ψ p (x i ,x j ) To describe the relationship between pixels, similar pixels are encouraged to be assigned the same feature labels, while pixels with large differences are assigned different feature labels. In this way, the server fuses the same feature labels to obtain the segmented image.

[0067] For ease of understanding, taking the segmentation of an ID card image as an example, the server extracts the feature pixels on the ID card, obtains all the pixel features on the ID card, marks the feature pixels on the name, ID number, ethnicity, birthplace and address with feature labels of different categories, and finally integrates the feature pixels with the same feature labels to obtain a segmented image with different text data.

[0068] 204. According to the clustering mode, the text data of the segmented image is compared with the target data to obtain a comparison result, and the target data is the data associated with the preset artificial intelligence recognition model.

[0069] According to the cluster mode, the server compares the text data of the segmented image with the target data to obtain the comparison result, and the target data is the data associated with the preset artificial intelligence recognition model. Specifically: the server matches the segmented image with the preset artificial intelligence recognition model, and each preset artificial intelligence recognition model corresponds to a type of segmented image; the server transmits the segmented image to the matched preset artificial intelligence recognition model; the server extracts the text data on the segmented image; the server compares the text data with the target data in the preset artificial intelligence recognition model to obtain the comparison result, and the target data is the data associated with the corresponding preset artificial intelligence recognition model.

[0070] The server uses load balancing (LB) to distribute text information to different preset target data models. LB is a cluster technology that distributes specific services such as network services and network traffic to multiple network devices such as servers, firewalls, etc. or multiple links, that is, balancing and distributing the load (work tasks) to multiple operation units for execution, thereby improving business processing capabilities and ensuring high business reliability.

[0071] Here, the server uses load balancing to match the segmented image with multiple preset artificial intelligence recognition models. Each preset artificial intelligence recognition model corresponds to a category of segmented images, and there are multiple preset artificial intelligence recognition models of different types. If the segmented image matches the preset artificial intelligence recognition model, the segmented image is transmitted to the preset artificial intelligence recognition model that matches it. In the preset artificial intelligence recognition model, the transmitted segmented image is processed, and the text data on the segmented image is extracted. The extracted text data is then compared with the target data to realize data verification of the certificate information and the information sheet.

[0072] It should be noted that the server extracts text data on the segmented image, specifically including: the server selects the text area on the segmented image; the server segments the text lines in the text area to obtain text characters; the server establishes a standard template based on the text characters; the server matches the standard template with the preset template and extracts text data, which is the data on the preset template.

[0073] When the server extracts text data from the same type of segmented images, it selects the text area on the same type of segmented images, and traverses the segmented images through a sliding window, judges the feature labels, and selects the same feature labels using a rectangular frame to obtain the text area; the server then performs rectangular segmentation on the text area to split it into different text characters, and the server moves a one-dimensional sliding window in the text area, and divides the characters by judging the spacing between characters; the server establishes a standard template for each text character, matches the standard template with a preset template, and further extracts the text data. The preset templates here include graphic templates, stroke templates, and geometric templates. The server matches each standard template with a preset template, and the matched template is the extracted text data.

[0074] The server uses optical character recognition (OCR) to extract text information. OCR refers to the process of analyzing and identifying image files of text materials to obtain text and layout information. It is the process of identifying the text in the image and returning it in the form of text. For a line of text, the server can only determine the content of the final line of text by identifying each character. Therefore, the server can match each possible character through a sliding window. The accuracy of this method depends on the size of the sliding window. If the sliding window size is too large, information will be lost, and if it is too small, the computing power requirement will increase significantly.

[0075] 205. Automatically feedback the audit results based on the comparison results.

[0076] The server automatically feedbacks the audit result based on the comparison result between the text data and the target data. Specifically: when the comparison result is that the text data and the target data are exactly the same, the server automatically feedbacks the words "audit successful"; when the comparison result is that the text data and the target data are not exactly the same, the server automatically feedbacks the words "audit failed".

[0077] Take the automatic review of insurance policy information as an example: if the type of text data is the ID card number of the insured in the insurance policy, the server will use load balancing to transmit the insured's ID card number to different preset recognition data models, such as ID card recognition model, vehicle registration recognition model, driver's license recognition model, etc., and compare the insured's ID card number with the same type of target data in the preset recognition model to determine whether the two ID card numbers are the same. If the two are the same, the server generates the words "review successful". If the two are different, the server generates the words "review failed".

[0078] The present invention utilizes a cluster mode and a preset artificial intelligence recognition model to quickly compare the certificate information with the corresponding information on the information sheet, completes the automatic review of business data, and improves the accuracy of the automatic review data and the efficiency of the review.

[0079] The above describes the automatic audit method based on artificial intelligence in the embodiment of the present invention. The following describes the automatic audit device based on artificial intelligence in the embodiment of the present invention. Figure 3 In one embodiment of the present invention, an automatic audit device based on artificial intelligence includes:

[0080] An acquisition unit 301 is used to acquire a pre-processed document image;

[0081] The processing unit 302 is used to segment the pre-processed document image according to a preset rule to obtain a segmented image;

[0082] A comparison unit 303 is used to compare the text data of the segmented image with target data according to the cluster mode to obtain a comparison result, wherein the target data is data associated with a preset artificial intelligence recognition model;

[0083] The feedback unit 304 is used to automatically feedback the audit result according to the comparison result.

[0084] In an embodiment of the present invention, an acquisition unit 301 acquires a preprocessed document image; a processing unit 302 segments the preprocessed document image according to preset rules to acquire a segmented image; a comparison unit 303 compares text data of the segmented image with target data according to a clustering mode to obtain a comparison result, wherein the target data is data associated with a preset artificial intelligence recognition model; and a feedback unit 304 automatically feeds back the review result based on the comparison result.

[0085] The present invention utilizes a cluster mode and a preset artificial intelligence recognition model to quickly compare the certificate information with the corresponding information on the information sheet, completes the automatic review of business data, and improves the accuracy of the automatic review data and the efficiency of the review.

[0086] See also Figure 4 Another embodiment of the automatic audit device based on artificial intelligence in the embodiment of the present invention includes:

[0087] An acquisition unit 301 is used to acquire a pre-processed document image;

[0088] The processing unit 302 is used to segment the pre-processed document image according to a preset rule to obtain a segmented image;

[0089] A comparison unit 303 is used to compare the text data of the segmented image with target data according to the cluster mode to obtain a comparison result, wherein the target data is data associated with a preset artificial intelligence recognition model;

[0090] The feedback unit 304 is used to automatically feedback the audit result according to the comparison result.

[0091] Optionally, the comparison unit 303 includes:

[0092] A matching module 3031 is used to match the segmented image with a preset artificial intelligence recognition model, each of the preset artificial intelligence recognition models corresponds to a type of the segmented image;

[0093] A transmission module 3032, used for transmitting the segmented image to the matched preset artificial intelligence recognition model;

[0094] An extraction module 3033, used to extract text data on the segmented image;

[0095] The comparison module 3034 is used to compare the text data with the target data in the preset artificial intelligence recognition model to obtain a comparison result, and the target data is the corresponding data associated with the preset artificial intelligence recognition model.

[0096] Optionally, the extraction module 3033 is specifically used for:

[0097] Selecting a text area on the segmented image;

[0098] Segmenting the text lines in the text area to obtain text characters;

[0099] Establishing a standard template based on the text characters;

[0100] The standard template is matched with the preset template to extract text data, where the text data is the data on the preset template.

[0101] Optionally, the acquiring unit 301 includes:

[0102] The acquisition module 3011 is used to acquire the uploaded document image;

[0103] The preprocessing module 3012 is used to normalize the document image to obtain a preprocessed document image.

[0104] Optionally, the preprocessing module 3012 is specifically used for:

[0105] Converting the document image into a grayscale image;

[0106] Converting the grayscale image into a black and white image;

[0107] Performing gradient sharpening on the black-and-white image to obtain a sharpened image;

[0108] The sharpened image is subjected to removal of discrete noise to obtain a processed document image.

[0109] Optionally, the processing unit 302 is specifically configured to:

[0110] Extracting characteristic pixels from the preprocessed document image;

[0111] Assigning different feature labels to different feature pixels;

[0112] The feature pixels with the same feature label are integrated to obtain a segmented image.

[0113] Optionally, the feedback unit 304 is specifically used for:

[0114] When the comparison result shows that the text data is exactly the same as the target data, the word "review success" is automatically fed back;

[0115] When the comparison result shows that the text data is not completely the same as the target data, the word "review failure" is automatically fed back.

[0116] In the embodiment of the present invention, the acquisition unit 301 acquires the preprocessed document image, wherein the acquisition module 3011 is used to acquire the uploaded document image, and the preprocessing module 3012 is used to normalize the document image to obtain the preprocessed document image; the processing unit 302 segments the preprocessed document image according to the preset rules to obtain the segmented image; the comparison unit 303 compares the text data of the segmented image with the target data according to the cluster mode to obtain the comparison result, wherein the target data is the data associated with the preset artificial intelligence recognition model, wherein the matching module 3031 is used to normalize the segmented image Matching with preset artificial intelligence recognition models, each of the preset artificial intelligence recognition models corresponds to a type of the segmented image; a transmission module 3032 is used to transmit the segmented image to the matched preset artificial intelligence recognition model; an extraction module 3033 is used to extract text data on the segmented image; a comparison module 3034 is used to compare the text data with the target data in the preset artificial intelligence recognition model to obtain a comparison result, and the target data is the associated data in the corresponding preset artificial intelligence recognition model; the feedback unit 304 automatically feeds back the review result according to the comparison result.

[0117] The present invention utilizes a cluster mode and a preset artificial intelligence recognition model to quickly compare the certificate information with the corresponding information on the information sheet, completes the automatic review of business data, and improves the accuracy of the automatic review data and the efficiency of the review.

[0118] above Figure 3 to Figure 4 The automatic audit device based on artificial intelligence in the embodiment of the present invention is described in detail from the perspective of modular functional entities. The automatic audit device based on artificial intelligence in the embodiment of the present invention is described in detail from the perspective of hardware processing.

[0119] Combine the following Figure 5 A detailed introduction to the various components of the automatic audit equipment based on artificial intelligence:

[0120] Figure 5 It is a structural schematic diagram of an automatic audit device based on artificial intelligence provided by an embodiment of the present invention. The automatic audit device 500 based on artificial intelligence may have relatively large differences due to different configurations or performances, and may include one or more processors (central processing units, CPU) 501 (for example, one or more processors) and a memory 509, and one or more storage media 508 (for example, one or more mass storage devices) storing application programs 507 or data 506. Among them, the memory 509 and the storage medium 508 can be short-term storage or permanent storage. The program stored in the storage medium 508 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations in the sign-in management device. Furthermore, the processor 501 can be configured to communicate with the storage medium 508, and execute a series of instruction operations in the storage medium 508 on the automatic audit device 500 based on artificial intelligence.

[0121] The artificial intelligence-based automatic audit device 500 may also include one or more power supplies 502, one or more wired or wireless network interfaces 503, one or more input and output interfaces 504, and / or one or more operating systems 505, such as Windows Serve, Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art will appreciate that Figure 5 The structure of the artificial intelligence-based automatic audit device shown in the figure does not constitute a limitation of the artificial intelligence-based automatic audit device, and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.

[0122] Combine the following Figure 5 A detailed introduction to the various components of the automatic audit equipment based on artificial intelligence:

[0123] Processor 501 is the control center of the automatic audit device based on artificial intelligence, and can be processed according to the automatic audit method based on artificial intelligence. Processor 501 uses various interfaces and lines to connect the various parts of the entire automatic audit device based on artificial intelligence, and by running or executing the software program and / or module stored in the memory 509, and calling the data stored in the memory 509, the cluster mode and the preset artificial intelligence recognition model are used to quickly compare the certificate information with the corresponding information of the information sheet, thereby improving the accuracy of the automatic audit data and the efficiency of the audit. Storage medium 508 and memory 509 are both carriers for storing data. In the embodiment of the present invention, storage medium 508 can refer to an internal memory with a small storage capacity but a fast speed, and memory 509 can be an external memory with a large storage capacity but a slow storage speed.

[0124] The memory 509 can be used to store software programs and modules. The processor 501 executes various functional applications and data processing of the automatic audit device 500 based on artificial intelligence by running the software programs and modules stored in the memory 509. The memory 509 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function, etc.; the data storage area may store data created according to the use of the automatic audit device based on artificial intelligence, etc. In addition, the memory 509 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage devices. The automatic audit program based on artificial intelligence provided in the embodiment of the present invention and the received data stream are stored in the memory, and when needed, the processor 501 calls from the memory 509.

[0125] When loading and executing computer program instructions on a computer, a process or function according to an embodiment of the present invention is generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. Computer instructions may be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium, for example, a computer instruction may be transmitted from a website site, a computer, a server, or a data center to another website site, a computer, a server, or a data center by wired (e.g., coaxial cable, optical fiber, twisted pair) or wireless (e.g., infrared, wireless, microwave, etc.) means. Computer-readable storage media may be any available medium that a computer can store or a data storage device such as a server or a data center that includes one or more available media integrations. Available media may be magnetic media, (e.g., floppy disk, hard disk, tape), optical media (e.g., optical disk), or semiconductor media (e.g., solid state disk (SSD)), etc.

[0126] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0127] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0128] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0129] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0130] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0131] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An automatic audit method based on artificial intelligence, characterized in that: include: Obtain the preprocessed document image; Segmenting the preprocessed document image according to preset rules to obtain a segmented image; According to the cluster mode, the text data of the segmented image is compared with the target data to obtain a comparison result, wherein the target data is the data associated with the preset artificial intelligence recognition model; According to the cluster mode, the text data of the segmented image is compared with the target data to obtain a comparison result, and the target data is the data associated with the preset artificial intelligence recognition model, including: Matching the segmented image with a preset artificial intelligence recognition model through load balancing, each of the preset artificial intelligence recognition models corresponding to a type of the segmented image; Transmitting the segmented image to the matched preset artificial intelligence recognition model; Extracting text data on the segmented image; The extracting of text data on the segmented image comprises: Selecting a text area on the segmented image by framing the text area on the segmented image, wherein the selecting the text area on the segmented image comprises: traversing the segmented image by a sliding window to obtain a plurality of feature labels, and judging the plurality of feature labels, selecting the same feature labels by using a rectangular frame to obtain a text area; Segmenting the text lines in the text area to obtain text characters, wherein segmenting the text lines in the text area to obtain text characters comprises: performing a one-dimensional sliding window movement on the text lines in the text area and dividing the text characters based on the spacing between characters; Establishing a standard template based on the text characters; Matching the standard template with a preset template to extract text data, where the text data is data on the preset template, and the preset template includes a graphic template, a stroke template, and a geometric template; In the preset artificial intelligence recognition model, the text data is compared with the target data to obtain a comparison result, and the target data is the associated data in the corresponding preset artificial intelligence recognition model; Based on the comparison results, the audit results are automatically fed back.

2. The automatic audit method based on artificial intelligence according to claim 1 is characterized in that: The obtaining of the preprocessed document image comprises: Get the uploaded document image; The document image is normalized to obtain a preprocessed document image.

3. The automatic audit method based on artificial intelligence according to claim 2 is characterized in that: The normalizing the document image to obtain a preprocessed document image comprises: Converting the document image into a grayscale image; Converting the grayscale image into a black and white image; Performing gradient sharpening on the black-and-white image to obtain a sharpened image; The sharpened image is subjected to a process of removing discrete noise to obtain a processed document image.

4. The automatic audit method based on artificial intelligence according to claim 1 is characterized in that: Segmenting the preprocessed document image according to a preset rule to obtain a segmented image comprises: Extracting characteristic pixels from the preprocessed document image; Assigning different feature labels to different feature pixels; The feature pixels with the same feature label are integrated to obtain a segmented image.

5. The automatic audit method based on artificial intelligence according to any one of claims 1 to 4, characterized in that: The automatic feedback of the audit result according to the comparison result includes: When the comparison result is that the text data is completely identical to the target data, the word "review success" is automatically fed back; When the comparison result is that the text data is not completely the same as the target data, the word "review failure" is automatically fed back.

6. An automatic audit device based on artificial intelligence, executing the automatic audit method based on artificial intelligence as described in any one of claims 1 to 5, characterized in that: include: An acquisition unit, used for acquiring a pre-processed document image; A processing unit, used to segment the preprocessed document image according to a preset rule to obtain a segmented image; A comparison unit, used for comparing the text data of the segmented image with target data according to the cluster mode to obtain a comparison result, wherein the target data is data associated with a preset artificial intelligence recognition model; The feedback unit is used to automatically feedback the audit result according to the comparison result.

7. An automatic audit device based on artificial intelligence, characterized in that: include: A memory and at least one processor, wherein instructions are stored in the memory, and the memory and the at least one processor are interconnected via a line; The at least one processor calls the instructions in the memory so that the artificial intelligence-based automatic audit device executes the method as described in any one of claims 1-5.

8. A computer-readable storage medium, characterized in that: It includes instructions, which, when executed on a computer, enable the computer to execute the steps of the automatic audit method based on artificial intelligence as described in any one of claims 1 to 5.

Images