Insurance claim settlement method and device based on document image, electronic equipment and medium

Abstract

The embodiment of the invention provides an insurance claim settlement method and device based on a document image, electronic equipment and a medium, belongs to the technical field of artificial intelligence, and is suitable for the field of financial science and technology. The method comprises the following steps: acquiring an original image of an original document; performing target identification on the original image to obtain a document region containing the original document, and performing video frame interception on the original image according to the region label of the document region to obtain an initial document image; and screening the initial document images according to the region labels to obtain candidate document images. And performing text generation according to the candidate document image to obtain a prompt text, and sending the prompt text to the target object. And in response to an image acquisition operation performed by the target object according to the prompt text, obtaining a reference image. And determining a target document image according to the candidate document image and the reference image, and performing claim settlement based on the target document image. According to the embodiment of the invention, the processing efficiency of the document image in the claim settlement process can be improved.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and is applicable to the financial technology field, particularly to a method and apparatus for target document image recognition, an electronic device, and a storage medium. Background Technology

[0002] In the insurance claims process, users typically need to take and upload images of relevant documents, such as their ID card, photos reflecting the insured's current status, and expense lists. Claims personnel then review and confirm these images. Because users generally lack professional knowledge about claims, the photos they take often do not meet the review requirements. This necessitates repeated communication, retaking of photos, and the provision of instructions by claims personnel. This not only increases the workload of claims staff but also makes the process cumbersome for users, resulting in low claims processing efficiency. Therefore, improving the efficiency of document image processing in the claims process has become an urgent problem to be solved. Summary of the Invention

[0003] The main objective of this application is to propose an insurance claims method, apparatus, electronic device, and medium based on document images, aiming to improve the processing efficiency of document images during the claims process.

[0004] To achieve the above objectives, a first aspect of this application proposes an insurance claims method based on document images, the method comprising:

[0005] Obtain the original image of the original document; wherein the original document is used by the target object to trigger a claim application;

[0006] Target recognition is performed on the original image to obtain the document region containing the original document, and video frames are extracted from the original image according to the region label of the document region to obtain the initial document image;

[0007] The initial document image is filtered according to the region label to obtain candidate document images;

[0008] Text is generated based on the candidate document image to obtain prompt text, and the prompt text is sent to the target object.

[0009] In response to the image acquisition operation performed by the target object based on the prompt text, a reference image is acquired;

[0010] The target document image is determined based on the candidate document image and the reference image, and the claim is processed based on the target document image.

[0011] In some embodiments, the step of filtering the initial document image based on the region label to obtain candidate document images includes:

[0012] The original image is identified as an insured object image by querying a preset object type mapping table based on the region label.

[0013] Select a first verification condition that matches the insured object from the preset image verification conditions; wherein, the first verification condition includes: requiring a preset number of images of the insured object at different angles;

[0014] Angle recognition is performed on the initial document image to obtain the acquisition angle of the original document in the initial document image;

[0015] The initial document image is filtered according to the first verification condition and the acquisition angle to obtain the candidate document image.

[0016] In some embodiments, the step of filtering the initial document image based on the region label to obtain candidate document images includes:

[0017] The original image is identified as a proof document image by querying a preset object type mapping table based on the region label.

[0018] Select a second verification condition that matches the proof document from the preset image verification conditions; wherein, the second verification condition includes: document text information that requires a preset document type;

[0019] Text extraction is performed on the initial document image to obtain the document text;

[0020] The initial document image is filtered according to the second verification condition and the document text to obtain the candidate document image.

[0021] In some embodiments, generating text based on the candidate document image to obtain prompt text, and sending the prompt text to the target object, includes:

[0022] Select intermediate document types that are different from the document types of the candidate document images from the preset target document types;

[0023] Based on the intermediate document type, text is generated to obtain the prompt text;

[0024] Send the prompt text to the target object.

[0025] In some embodiments, the step of generating text based on the intermediate document type to obtain prompt text includes:

[0026] Based on the timestamp of the initial document image, the motion direction of the initial document image is identified to obtain the shooting motion direction;

[0027] The prompt text is generated based on the shooting direction and the intermediate document type.

[0028] In some embodiments, determining the target document image based on the candidate document image and the reference image includes:

[0029] Based on the intermediate document type and the document type of the reference image, the reference image is filtered to obtain the intermediate image;

[0030] The authenticity of the candidate document images and the intermediate images is evaluated to obtain an authenticity score for each image;

[0031] The candidate document image and the intermediate image are selected based on the authenticity score and the target document type to obtain the target document image.

[0032] In some embodiments, the step of performing target recognition on the original image to obtain a document region containing the original document, and then extracting video frames from the original image based on the region label of the document region to obtain an initial document image, includes:

[0033] Target recognition is performed on the original image to obtain a document region containing the original document;

[0034] The original image is subjected to speech recognition to obtain the original speech text;

[0035] Keyword extraction is performed on the original speech text to obtain speech keywords;

[0036] The original image is processed by extracting video frames based on the region label and the voice keywords to obtain the initial document image.

[0037] To achieve the above objectives, a second aspect of this application provides an insurance claims processing device based on document images, the device comprising:

[0038] The first acquisition module is used to acquire the original image of the original document; wherein, the original document is used by the target object to trigger a claim application;

[0039] The document image cropping module is used to perform target recognition on the original image to obtain a document region containing the original document, and to crop video frames from the original image according to the region label of the document region to obtain an initial document image.

[0040] The document image filtering module is used to filter the initial document image according to the region label to obtain candidate document images;

[0041] The prompt text generation module is used to generate text based on the candidate document image, obtain prompt text, and send the prompt text to the target object;

[0042] The second acquisition module is used to acquire a reference image in response to the image acquisition operation performed by the target object according to the prompt text;

[0043] The claims module is used to determine the target document image based on the candidate document image and the reference image, and to process the claims based on the target document image.

[0044] To achieve the above objectives, a third aspect of this application provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the method described in the first aspect.

[0045] To achieve the above objectives, a fourth aspect of the present application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described in the first aspect.

[0046] This application proposes an insurance claims processing method, apparatus, electronic device, and medium based on document images. It acquires the original image of the original document uploaded by the target object, and uses image target recognition to obtain the document region and its corresponding region label, accurately locating and capturing the initial document image. Subsequently, based on the region label, it filters out candidate document images with clear claims value. Then, it automatically generates prompt text for the candidate document image and sends it to the target object, guiding them to complete the supplementary shooting operation of the reference image according to business requirements. Finally, it matches the reference image with the candidate document images to determine the target document image, thereby achieving efficient and accurate insurance claims processing. The embodiments of this application can effectively reduce repeated communication between the target object and claims personnel, simplify the target object's operating process, and thus improve the processing efficiency of document images in insurance claims. Attached Figure Description

[0047] Figure 1 This is a flowchart of the insurance claims method based on document images provided in the embodiments of this application;

[0048] Figure 2 yes Figure 1 The flowchart of step S102 in the document;

[0049] Figure 3 yes Figure 1The flowchart of step S103 in the process;

[0050] Figure 4 yes Figure 1 Another flowchart of step S103 in the process;

[0051] Figure 5 yes Figure 1 The flowchart of step S105 in the process;

[0052] Figure 6 yes Figure 1 The flowchart of step S106 in the process;

[0053] Figure 7 This is a schematic diagram of the structure of the insurance claims device based on document images provided in the embodiments of this application;

[0054] Figure 8 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0055] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0056] It should be noted that although functional modules are divided in the device schematic diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the device or the order in the flowchart. The terms "first," "second," etc., in the specification, claims, and the aforementioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0057] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0058] First, let's analyze some of the terms used in this application:

[0059] Artificial intelligence (AI) is a new branch of computer science that studies, develops, and applies theories, methods, technologies, and systems to simulate, extend, and expand human intelligence. It aims to understand the essence of intelligence and produce intelligent machines that can react in a way similar to human intelligence. Research in this field includes robotics, speech recognition, image recognition, natural language processing, and expert systems. AI can simulate the information processes of human consciousness and thought. Furthermore, AI utilizes digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceiving the environment, acquiring knowledge, and using that knowledge to achieve optimal results.

[0060] Natural Language Processing (NLP): NLP uses computers to process, understand, and utilize human language (such as Chinese and English). NLP is a branch of artificial intelligence and an interdisciplinary field of computer science and linguistics, often referred to as computational linguistics. NLP includes syntactic analysis, semantic analysis, and discourse understanding. It is commonly used in machine translation, handwritten and printed character recognition, speech recognition and text-to-speech conversion, intent recognition, information extraction and filtering, text classification and clustering, sentiment analysis, and opinion mining. It involves data mining, machine learning, knowledge acquisition, knowledge engineering, artificial intelligence research, and linguistic research related to language computation.

[0061] In the insurance claims process, users are typically required to take and upload images of specific target documents, such as ID cards, driver's licenses, vehicle registration certificates, expense lists, and photos reflecting the current state of the insured or the accident scene. However, in current practice, due to users' general lack of understanding of claims procedures and related professional knowledge, the photos they take often suffer from unclear content, incomplete information, or incorrect angles, making it difficult for the uploaded documents to meet the review requirements of claims personnel. This situation forces claims personnel to repeatedly communicate with users, send relevant operating instructions, and even require users to retake and submit photos multiple times. This repetitive communication and retaking not only places a significant additional workload on claims personnel but also makes the process cumbersome and time-consuming for users, further reducing the overall efficiency of the claims process. Therefore, improving the processing efficiency of target document images in the claims process has become an urgent problem to be solved.

[0062] Based on this, embodiments of this application provide an insurance claims method, apparatus, electronic device, and medium based on document images, aiming to improve the processing efficiency of document images during the claims process.

[0063] The insurance claim method, apparatus, electronic device, and medium based on document images provided in this application are specifically described through the following embodiments. First, the insurance claim method based on document images in this application embodiment is described.

[0064] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0065] Foundational technologies for artificial intelligence generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies mainly encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0066] The insurance claim method based on document images provided in this application relates to the field of artificial intelligence technology. This method can be applied to a terminal, a server, or software running on either a terminal or a server. In some embodiments, the terminal can be a smartphone, tablet, laptop, desktop computer, etc.; the server can be configured as an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN, and big data and artificial intelligence platforms; the software can be an application implementing the insurance claim method based on document images, but is not limited to the above forms.

[0067] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0068] It should be noted that in all specific embodiments of this application, when processing data related to user identity or characteristics, such as user information, user behavior data, user historical data, and user location information, user permission or consent is obtained first. Furthermore, the collection, use, and processing of this data comply with relevant laws, regulations, and standards. In addition, when embodiments of this application require access to sensitive personal information of users, separate permission or consent from the user is obtained through pop-ups or redirection to confirmation pages. Only after obtaining the user's separate permission or consent is the necessary user-related data required for the proper functioning of these embodiments acquired.

[0069] Figure 1 This is an optional flowchart of the insurance claims method based on document images provided in the embodiments of this application. Figure 1 The method may include, but is not limited to, steps S101 to S106.

[0070] Step S101: Obtain the original image of the original document; wherein, the original document is used by the target object to trigger a claim application.

[0071] Step S102: Perform target recognition on the original image to obtain the document region containing the original document, and extract video frames from the original image according to the region label of the document region to obtain the initial document image.

[0072] Step S103: Filter the initial document images according to the region labels to obtain candidate document images.

[0073] Step S104: Generate text based on candidate document images to obtain prompt text, and send the prompt text to the target object.

[0074] Step S105: In response to the image acquisition operation performed by the target object based on the prompt text, a reference image is acquired.

[0075] Step S106: Determine the target document image based on the candidate document image and the reference image, and process the claim based on the target document image.

[0076] Steps S101 to S106 of this embodiment involve acquiring the original image of the original document uploaded by the target object, using image target recognition to obtain the document area and corresponding area tags, accurately locating and capturing the initial document image. Subsequently, candidate document images with clear claims value are selected based on the area tags. Then, a prompt text is automatically generated for each candidate document image and sent to the target object, guiding them to complete the supplementary shooting of the reference image according to business requirements. Finally, the reference image and the candidate document images are matched to determine the target document image, achieving efficient and accurate insurance claims processing. This embodiment effectively reduces repeated communication between the target object and claims personnel, simplifies the target object's operational process, and thus improves the processing efficiency of document images in insurance claims.

[0077] In step S101 of some embodiments, the target object is the policyholder, insured, or their authorized claim applicant under the insurance contract. A claim application refers to the act of the target object submitting a claim for compensation to the insurance company in accordance with the insurance contract due to an insured event. For example, the policyholder applies to the insurance company for vehicle damage compensation due to a vehicle collision. Original documents may include images of the accident scene, images of the injured parts of the policyholder, images of the damaged parts of the vehicle, or images of relevant documents. The original images are the images or videos captured as described above. For example, when a vehicle insurance accident occurs, the target object uses a smartphone to take photos of the accident scene after the collision, driver's license, and accident scene video as the original images of the original documents, and then uploads them to the insurance company's claim review system.

[0078] In step S102 of some embodiments, the initial document image is a key video frame or still image selected from the original image that clearly contains a specific object closely related to the insurance claims business (such as an accident scene, insured vehicle, documents, etc.).

[0079] Specifically, please refer to Figure 2 Step S102 may include, but is not limited to, steps S201 to S204:

[0080] Step S201: Perform target recognition on the original image to obtain the document area containing the original document.

[0081] Step S202: Perform speech recognition on the original image to obtain the original speech text.

[0082] Step S203: Extract keywords from the original speech text to obtain speech keywords.

[0083] Step S204: Extract video frames from the original image based on the region label and voice keywords to obtain the initial document image.

[0084] In step S201 of some embodiments, specific objects or scenes existing in the original image can be automatically detected and classified using image processing algorithms and machine learning models, thereby identifying and marking the area containing the original document information, i.e., the document area. For example, if a target object takes and uploads a video of a traffic accident, the target recognition algorithm (such as the YOLO algorithm) can automatically locate the vehicle damage area, license plate position, and accident background area in the video to obtain a clear document area.

[0085] In step S202 of some embodiments, audio recognition technology is used to analyze and process the speech signal in the original image, automatically converting it into text information, thus obtaining the original speech text. For example, when the policyholder makes a verbal description while filming an accident scene, such as the location, time, and extent of the damage, the system uses speech recognition technology to convert this description into a corresponding text description, forming the original speech text.

[0086] In step S203 of some embodiments, natural language processing technology is used to segment, analyze parts of speech, and match semantics of the original speech text, automatically selecting words with high business relevance to form speech keywords. For example, if the original speech text is "The front bumper of the vehicle is severely deformed, the license plate is A12345, and the accident occurred at a highway intersection," the speech keywords obtained through keyword extraction technology are "front bumper," "severely deformed," "license plate," "A12345," and "highway intersection."

[0087] In step S204 of some embodiments, it should be noted that the target object may capture similar items not belonging to the insured in complex environments. For example, when the target object is filming in an environment with heavy traffic (e.g., on a highway), it is difficult to avoid capturing other vehicles in the video. However, when performing target recognition on the original image, all vehicles in the original image are identified as "vehicles," resulting in the original image containing multiple non-target vehicles. In this case, the corresponding voice occurrence timestamp can be determined based on voice keywords (e.g., the target object describing "this is my car" or "the front bumper is bent," obtaining the keywords "my car" and "front bumper" respectively). Subsequently, the time period range in the original image is determined based on the voice occurrence timestamp. Finally, within this time period range, from the multiple vehicle locations marked by vehicle area labels, the vehicle area corresponding to the video frame closest to the voice keyword timestamp is selected for video frame extraction, resulting in an initial document image centered on the target vehicle. For example, the target subject takes a video at the scene of a highway accident and says "This is my car" at the 12-second mark of the video. The timestamp corresponding to this voice keyword is recorded as 12 seconds. Then, combined with vehicle area tag information, the vehicle that matches the keyword timestamp in one frame or several consecutive frames near the 12-second mark of the video is selected as the target vehicle, thereby accurately extracting the video frame as the initial document image.

[0088] Steps S201 to S204 of this embodiment illustrate the method of accurately obtaining the document area containing the original documents by performing target recognition on the uploaded original image. At the same time, speech recognition is performed on the voice information contained in the original image, and voice keywords related to claims review are further extracted. Then, the document area and voice keywords are combined to accurately capture video frames, thereby effectively generating an initial document image that meets business requirements. This significantly reduces the manual intervention and repetitive verification work of insurance claims review personnel, and improves the accuracy and efficiency of document image review.

[0089] Please see Figure 3 In some embodiments, step S103 may include, but is not limited to, steps S301 to S304:

[0090] Step S301: Query the preset object type mapping table according to the region label to determine that the original image is the insured object image.

[0091] Step S302: Select a first verification condition that matches the insured object from the preset image verification conditions; wherein, the first verification condition includes: needing a preset number of images of the insured object at different angles.

[0092] Step S303: Perform angle recognition on the initial document image to obtain the acquisition angle of the original document in the initial document image.

[0093] Step S304: Filter the initial document images according to the first verification condition and the acquisition angle to obtain candidate document images.

[0094] In step S301 of some embodiments, the object type mapping table stores the correspondence between various document area labels and specific object types during the insurance claim process. The specific claim object category involved in the current document can include, but is not limited to: supporting documents (such as ID cards, driver's licenses, vehicle registration certificates, medical expense receipts, etc.) and insured objects (such as insured vehicles, insured persons, insured property, etc.). For example, the object type mapping table records specific associations such as "front of ID card" as a supporting document type and "vehicle front bumper" as an insured vehicle type.

[0095] In step S302 of some embodiments, after determining the category of the insured object, the corresponding image acquisition requirements are automatically selected from a preset verification condition library based on the category to specify the required number of images and shooting angle requirements. For example, when the insured object is determined to be an insured vehicle, the first verification condition is selected from the preset verification condition library, which requires the insured object to provide three images of vehicle damage taken from the front, side, and rear of the vehicle, respectively.

[0096] In step S303 of some embodiments, target pose estimation technology from computer vision can be used. By detecting the feature key points, contour boundaries, and local details of the insured object in the initial document image, a three-dimensional spatial coordinate system of the insured object is established. Based on the geometric perspective relationship of the insured object in the image, the angle information of the insured object relative to the shooting device is calculated, thereby clarifying the specific acquisition angle of the initial document image. For example, in the process of acquiring vehicle damage images, multiple feature key point information such as headlights, bumpers, and wheels of the insured vehicle can be extracted through a convolutional neural network. Based on the spatial distribution of the key points and comparison with the three-dimensional vehicle model, the acquisition angle of the insured vehicle image can be calculated, such as 0 degrees in front, approximately 45 degrees to the front side, or 90 degrees to the side.

[0097] In step S304 of some embodiments, the acquisition angle of the initial document image is compared and analyzed with the shooting angle specified in the first verification condition, and only the initial document image that matches or is closest to the specified angle is selected as the candidate document image. For example, if the first verification condition requires images from three different angles—the front, side, and rear of the vehicle—then only the initial document images with shooting angles clearly defined as the front, side, and rear of the vehicle are selected as candidate document images.

[0098] Steps S301 to S304 as shown in this embodiment accurately determine the insured object category of the original document by combining the document area label with the object type mapping table, and automatically match the corresponding image verification conditions based on the confirmed insured object category. Then, the initial document image is subjected to angle recognition, and then candidate document images that meet the requirements are strictly screened according to the preset verification conditions. This avoids the tedious process of claims personnel communicating with users multiple times to collect supplementary images and effectively improves the accuracy of document images.

[0099] Please see Figure 4 In some other embodiments, step S103 may include, but is not limited to, steps S401 to S404:

[0100] Step S401: Query the preset object type mapping table according to the region label to determine that the original image is a proof document image.

[0101] Step S402: Select a second verification condition that matches the proof document from the preset image verification conditions; wherein, the second verification condition includes: document text information that requires a preset document type.

[0102] Step S403: Extract text from the initial document image to obtain the document text.

[0103] Step S404: Filter the initial document image according to the second verification condition and the document text to obtain candidate document images.

[0104] In some embodiments, step S401 is the same as the embodiment of step S301, and will not be described again here.

[0105] In step S402 of some embodiments, the system automatically selects the corresponding verification requirements from the image verification condition library using the confirmed document category information to obtain the second verification conditions for subsequent screening. For example, when the document category is determined to be "ID card", the system automatically selects the corresponding second verification condition for the ID card from the image verification condition library, that is, the ID card must contain text information of name, ID number and validity period.

[0106] In step S403 of some embodiments, the text area contained in the initial document image can be scanned using optical character recognition technology to automatically identify the corresponding text content and convert the identified content into standardized document text. For example, when the initial document image is an ID card image, information such as "Zhang San", "ID card number: 110101199001010011" and "validity period: 2020.01.01-2040.01.01" can be automatically extracted from the ID card image using optical character recognition technology to form the corresponding document text.

[0107] In step S404 of some embodiments, the document text is automatically compared to see if it fully contains all the preset text fields or key content information required by the second verification condition. For initial document images that fully meet the verification conditions, they are selected as candidate document images. The second verification condition may include, but is not limited to, the following: Identity documents must clearly display the name, ID number, and validity period. Driver's licenses must clearly include the name, document number, permitted vehicle type, and validity date. Vehicle registration certificates must include the vehicle model, chassis number, and license plate number. For example, in the document text of an ID card image, if the identification document text simultaneously and completely contains the three fields "name," "ID number," and "validity period," then the initial document image is confirmed to meet the second verification condition, and thus the image is selected as a candidate document image for subsequent claims review.

[0108] In some personal accident insurance claims processes, the second verification condition can also be: the medical expense receipts must contain complete information such as the hospital name, patient name, details of the treatment items, the amount charged, and the date of payment. These verification conditions clarify the specific textual content that supporting documents must have during the insurance claim review process, thereby ensuring the completeness and authenticity of the document information during the review process.

[0109] Steps S401 to S404 as shown in this embodiment of the application automatically and accurately identify the original documents as proof documents based on the area label, and further automatically match the text information to be verified as the second verification condition based on the determined proof document type. Then, the text content in the initial document image is automatically extracted, and the extracted document text is compared with the second verification condition, thereby quickly and effectively realizing the automatic screening of document images and avoiding the tedious process of claims personnel manually checking one by one.

[0110] In step S104 of some embodiments,

[0111] Please see Figure 5 In some embodiments, step S104 may also include, but is not limited to, steps S501 to S503:

[0112] Step S501: Select intermediate document types that are different from the document types of the candidate document images from the preset target document types.

[0113] Step S502: Generate text based on intermediate document type to obtain prompt text.

[0114] Step S503: Send a prompt text to the target object.

[0115] In step S501 of some embodiments, the specific document type corresponding to the candidate document image is first determined based on the region label or document text information of the candidate document image. Then, the document type corresponding to the candidate document image is compared with a preset target document type set. The confirmed document types are removed from the set to obtain intermediate document types that have not yet been collected but are necessary for claims review. For example, the preset target document type set includes three document types: front-end damage photos of vehicles, side-end damage photos of vehicles, and ID card photos. If the candidate document image is a front-end damage photo of a vehicle, this type is excluded from the target document type set, and side-end damage photos of vehicles and ID card photos are selected as intermediate document types that need to be collected.

[0116] In step S502 of some embodiments, the intermediate document type can be directly used as a keyword in the prompt text. For example, if the intermediate document type is a vehicle side damage photo, the prompt text could be: "Please upload a vehicle side damage photo".

[0117] In scenarios involving real-time acquisition of document images, the generation of prompt text can be accomplished using the following steps: First, based on the timestamp of the initial document image, the motion direction of the initial document image is identified to determine the shooting direction. Then, prompt text is generated based on the shooting direction and the document type. Specifically, image feature points from consecutive frames in the initial document image sequence can be extracted first, for example, using Scale Invariant Feature Transform (SIFT) or ORB feature description algorithms to identify the positional changes of the same feature points between frames. Next, by matching the feature points between adjacent frames, the Structure from Motion (SfM) algorithm or visual odometry is used to reconstruct the spatial motion trajectory and viewpoint changes of the shooting device over time. Subsequently, by analyzing the obtained spatial motion trajectory, the specific motion direction of the shooting device relative to the insured vehicle is determined, such as moving from the front of the vehicle to the left or from the left to the rear. Finally, based on the determined direction of movement of the shooting equipment and the type of intermediate documents to be supplemented, clear and directional prompt text is automatically generated, such as prompting the target object "Please continue to move to the left side of the vehicle to photograph the damage on the left side of the vehicle", to guide the target object to complete the supplementary collection of subsequent document images.

[0118] In step S503 of some embodiments, a message notification module is integrated into the claims client or application logged into by the target object. Upon generating a prompt text, the application interface is automatically invoked to push the prompt text to the front-end interface of the terminal device in real time, displaying it to the target object as an interactive pop-up window, clearly indicating the document image information that needs to be supplemented. For example, when the target object opens the insurance claims client on its terminal device, a prompt text pop-up window appears in the center of the interface: "Please continue moving to the left side of the vehicle and take pictures of the damage on the left side of the vehicle to complete the supplementary document image collection." After the target object clicks to confirm, it can continue taking pictures as prompted.

[0119] Steps S501 to S503, as illustrated in this embodiment, automatically identify the document types corresponding to the candidate document images and automatically compare and filter the identified document types with a preset set of target document types to accurately confirm missing and necessary intermediate document types. Subsequently, based on the identified intermediate document types, specific and clear prompt text is automatically generated and promptly sent to the target object, thereby providing real-time and effective guidance to the target object. This significantly reduces manual intervention and repetitive communication work by claims review personnel and effectively improves the integrity of document images in insurance claims processing.

[0120] In step S105 of some embodiments, the target object takes a targeted image using a terminal device based on the received prompt information and uploads it to the insurance claims system. The insurance claims system receives and confirms the uploaded image. For example, after receiving the prompt text "Please upload a clear image of the injured person's wound," the target object takes a close-up photo of the injured area using a smartphone and uses that image as reference image 1.

[0121] In step S106 of some embodiments, for a vehicle damage claim, the vehicle damage in the candidate document image is matched and analyzed with the license plate and vehicle details uploaded in the reference image to determine a clear and complete image of the vehicle damage as the target document image, so as to conduct loss assessment and claim amount verification.

[0122] Further, please refer to Figure 6 In some embodiments, step S106 includes, but is not limited to, steps S601 to S603:

[0123] Step S601: Based on the intermediate document type and the document type of the reference image, filter the reference image to obtain the intermediate image.

[0124] Step S602: Evaluate the authenticity of the candidate document images and intermediate images to obtain an authenticity score for each image.

[0125] Step S603: Select candidate document images and intermediate images based on authenticity score and target document type to obtain target document image.

[0126] In step S601 of some embodiments, the specific document type corresponding to each reference image is first identified based on the region label or text content of the reference image. Then, each specific document type is compared one by one with a pre-determined intermediate document type, and a reference image whose document type is completely consistent with the intermediate document type is automatically selected to obtain an intermediate image. For example, if the intermediate document type is a vehicle left-side damage image, and the reference images uploaded by the target object are of three types: vehicle left-side damage image, ID card image, and vehicle front damage image, then the reference image with the document type of vehicle left-side damage image is selected as the intermediate image.

[0127] In step S602 of some embodiments, image analysis models such as convolutional neural networks (e.g., image classification networks based on ResNet or DenseNet architectures) or adversarial neural networks (e.g., anomaly detection networks based on GANs) can be used to automatically extract the texture, detail, and scene consistency features of each image, determine the authenticity of the image, and output a quantified authenticity score.

[0128] In step S603 of some embodiments, for each document type in the target document type, the image with the highest authenticity score is selected from the candidate document images and intermediate images as the target document image.

[0129] Steps S601 to S603 as shown in the embodiments of this application automatically filter out intermediate images that meet the conditions from the reference images based on the intermediate document type, and further use an image analysis network to automatically evaluate the matching degree between the candidate document images and intermediate images and the real claims scenario, obtain the authenticity score of each image, and then accurately select the most reliable target document image that matches the actual situation of the insurance accident based on the authenticity score and the target document type, thereby effectively improving the accuracy of the document images.

[0130] Please see Figure 7 This application also provides an insurance claim device based on document images, which can implement the above-mentioned insurance claim method based on document images. The device includes:

[0131] The first acquisition module 701 is used to acquire the original image of the original document; wherein, the original document is used by the target object to trigger a claim application;

[0132] The document image cropping module 702 is used to perform target recognition on the original image, obtain the document region containing the original document, and crop the video frame of the original image according to the region label of the document region to obtain the initial document image.

[0133] Document image filtering module 703 is used to filter the initial document image according to the region label to obtain candidate document images;

[0134] The prompt text generation module 704 is used to generate text based on the candidate document image, obtain prompt text, and send the prompt text to the target object.

[0135] The second acquisition module 705 is used to acquire a reference image in response to the image acquisition operation performed by the target object according to the prompt text.

[0136] The claims module 706 is used to determine the target document image based on the candidate document image and the reference image, and to process the claims based on the target document image.

[0137] The specific implementation of this document image-based insurance claims device is basically the same as the specific implementation of the document image-based insurance claims method described above, and will not be repeated here.

[0138] This application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the aforementioned insurance claim method based on document images. This electronic device can be any smart terminal, including tablet computers, in-vehicle computers, etc.

[0139] Please see Figure 8 , Figure 8 The hardware structure of an electronic device according to another embodiment is illustrated. The electronic device includes:

[0140] The processor 801 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application.

[0141] The memory 802 can be implemented as a read-only memory (ROM), static storage device, dynamic storage device, or random access memory (RAM). The memory 802 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 802 and is called and executed by the processor 801 to implement the insurance claim method based on document images of this application.

[0142] The 803 input / output interface is used to implement information input and output.

[0143] The communication interface 804 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0144] Bus 805 transmits information between various components of the device (e.g., processor 801, memory 802, input / output interface 803, and communication interface 804);

[0145] The processor 801, memory 802, input / output interface 803, and communication interface 804 are connected to each other within the device via bus 805.

[0146] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described insurance claims method based on document images.

[0147] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0148] The insurance claims method, device, electronic equipment, and storage medium based on document images provided in this application acquire the original image of the original document uploaded by the target object, and use image target recognition to obtain the document area and corresponding area tags, accurately locating and capturing the initial document image. Subsequently, based on the area tags, candidate document images with clear claims value are selected. Then, prompt text is automatically generated for each candidate document image and sent to the target object, guiding the target object to complete the supplementary shooting operation of the reference image according to business requirements. Finally, the reference image and the candidate document images are matched to determine the target document image, thereby achieving efficient and accurate insurance claims processing. This application embodiment can effectively reduce repeated communication between the target object and claims personnel, simplify the target object's operation process, and thus improve the processing efficiency of document images in insurance claims.

[0149] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0150] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0151] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0152] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0153] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0154] It should be understood that in this application, "at least one (item)" means one or more, and "more than one" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0155] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of the units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0156] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0157] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0158] If the integrated unit is implemented as 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 this application, 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. This computer software product is stored in a storage medium and includes multiple instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0159] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. An insurance claims method based on document images, characterized in that, The method includes: Obtain the original image of the original document; wherein the original document is used by the target object to trigger a claim application; Target recognition is performed on the original image to obtain the document region containing the original document, and video frames are extracted from the original image according to the region label of the document region to obtain the initial document image; The initial document image is filtered according to the region label to obtain candidate document images; Text is generated based on the candidate document image to obtain prompt text, and the prompt text is sent to the target object. In response to the image acquisition operation performed by the target object based on the prompt text, a reference image is acquired; The target document image is determined based on the candidate document image and the reference image, and the claim is processed based on the target document image.

2. The method according to claim 1, characterized in that, The step of filtering the initial document image based on the region label to obtain candidate document images includes: The original image is identified as an insured object image by querying a preset object type mapping table based on the region label. Select a first verification condition that matches the insured object from the preset image verification conditions; wherein, the first verification condition includes: requiring a preset number of images of the insured object at different angles; Angle recognition is performed on the initial document image to obtain the acquisition angle of the original document in the initial document image; The initial document image is filtered according to the first verification condition and the acquisition angle to obtain the candidate document image.

3. The method according to claim 1, characterized in that, The step of filtering the initial document image based on the region label to obtain candidate document images includes: The original image is identified as a proof document image by querying a preset object type mapping table based on the region label. Select a second verification condition that matches the proof document from the preset image verification conditions; wherein, the second verification condition includes: document text information that requires a preset document type; Text extraction is performed on the initial document image to obtain the document text; The initial document image is filtered according to the second verification condition and the document text to obtain the candidate document image.

4. The method according to claim 1, characterized in that, The step of generating text based on the candidate document image to obtain prompt text, and sending the prompt text to the target object, includes: Select intermediate document types that are different from the document types of the candidate document images from the preset target document types; Based on the intermediate document type, text is generated to obtain the prompt text; Send the prompt text to the target object.

5. The method according to claim 4, characterized in that, The step of generating text based on the intermediate document type to obtain prompt text includes: Based on the timestamp of the initial document image, the motion direction of the initial document image is identified to obtain the shooting motion direction; The prompt text is generated based on the shooting direction and the intermediate document type.

6. The method according to claim 4, characterized in that, The step of determining the target document image based on the candidate document image and the reference image includes: Based on the intermediate document type and the document type of the reference image, the reference image is filtered to obtain the intermediate image; The authenticity of the candidate document images and the intermediate images is evaluated to obtain an authenticity score for each image; The candidate document image and the intermediate image are selected based on the authenticity score and the target document type to obtain the target document image.

7. The method according to any one of claims 1 to 6, characterized in that, The process of performing target recognition on the original image to obtain a document region containing the original document, and then extracting video frames from the original image based on the region label of the document region to obtain an initial document image, includes: Target recognition is performed on the original image to obtain a document region containing the original document; The original image is subjected to speech recognition to obtain the original speech text; Keyword extraction is performed on the original speech text to obtain speech keywords; The original image is processed by extracting video frames based on the region label and the voice keywords to obtain the initial document image.

8. An insurance claims device based on document images, characterized in that, The device includes: The first acquisition module is used to acquire the original image of the original document; wherein, the original document is used by the target object to trigger a claim application; The document image cropping module is used to perform target recognition on the original image to obtain a document region containing the original document, and to crop video frames from the original image according to the region label of the document region to obtain an initial document image. The document image filtering module is used to filter the initial document image according to the region label to obtain candidate document images; The prompt text generation module is used to generate text based on the candidate document image, obtain prompt text, and send the prompt text to the target object; The second acquisition module is used to acquire a reference image in response to the image acquisition operation performed by the target object according to the prompt text; The claims module is used to determine the target document image based on the candidate document image and the reference image, and to process the claims based on the target document image.

9. An electronic device, characterized in that, The electronic device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1 to 7.

Images