A credit review method, device and storage medium based on video image detection technology

Abstract

The present application discloses a credit review method, device, and storage medium based on video image detection technology. The method includes: upon receiving a credit review request triggered by a user terminal, generating a random privacy sequence corresponding to the credit review request and sending it to the user terminal; encoding the privacy information area in the provident fund image to be verified and the credit application video based on the random privacy sequence; receiving the encoded provident fund image to be verified and the encoded credit application video uploaded by the user terminal, and extracting key frame images from the encoded credit application video; based on the random privacy sequence and a preset similarity judgment rule, determining the similarity between the key frame image and the encoded provident fund image to be verified, and when the similarity is greater than a preset threshold, determining that the provident fund image to be verified originated from the credit application video. Through the above method, the present application realizes automated credit review of provident fund loan applications while protecting user privacy information.

Description

Technical Field

[0001] The present application relates to the field of credit review technology, and in particular to a credit review method, device and storage medium based on video image detection technology. Background Art

[0002] Provident Fund loans refer to housing purchase loans enjoyed by employees who have paid into the Housing Provident Fund. Currently, the process of determining whether a registered user has a risk of fraud is still largely dependent on manual labor and is in a relatively traditional stage. Therefore, the risk control model needs to be upgraded to a certain extent.

[0003] In automated credit review scenarios for provident fund loans, users are required to upload a Provident Fund image captured during the application process (i.e., the Provident Fund image is an image captured from a certain step in the application process when applying online) and a recorded credit application video. Currently, fraud risk is determined by verifying that the Provident Fund image is derived from the credit application video. However, manual review makes this process an easy target for financial fraudsters. Furthermore, directly uploading Provident Fund images and credit application videos exposes personal information to the internet, raising concerns about personal information security. Therefore, how to automate the credit review of Provident Fund loan applications while protecting user privacy has become a pressing technical challenge. Summary of the Invention

[0004] The embodiments of the present application provide a credit review method, device and storage medium based on video image detection technology to solve the following technical problem: how to automatically review provident fund loan applications while protecting user privacy information.

[0005] In a first aspect, an embodiment of the present application provides a credit review method based on video image detection technology, characterized in that the method includes: upon receiving a credit review request triggered by a user terminal, generating a random privacy sequence corresponding to the credit review request, and sending the random privacy sequence to the user terminal; determining the privacy information area in the provident fund image to be verified and the credit application video through a privacy information recognition algorithm preset by the user terminal, and encoding the privacy information area based on the random privacy sequence; receiving the encoded provident fund image to be verified and the encoded credit application video uploaded by the user terminal, and extracting a first number of frame images from the encoded credit application video, and determining the key frame images in the first number of frame images; based on the random privacy sequence and a preset similarity judgment rule, determining the similarity between the key frame image and the encoded provident fund image to be verified, and when the similarity is greater than a preset threshold, determining that the provident fund image to be verified originates from the credit application video.

[0006] In one implementation of the present application, generating a random privacy sequence specifically includes: determining a request value of a credit review request; wherein the request value is used to describe the triggering order of the credit review request; based on the request value, determining the parameter value of the linear congruential recursive formula to be applied to obtain the linear congruential recursive formula to be applied corresponding to the credit review request; based on a preset random privacy sequence length, calculating the random privacy sequence corresponding to the credit review request through the linear congruential recursive formula to be applied.

[0007] In one implementation of the present application, the privacy information area in the provident fund image to be verified and the credit application video is determined by a privacy information identification algorithm preset by the user end, specifically including: based on a preset content type identification model, determining the image content area and the field content area in each frame image of the provident fund image to be verified and the credit application video; performing portrait recognition on the image content area through a portrait recognition algorithm to determine whether the image content of the image content area is a portrait, and when it is determined that the image content area is a portrait, determining that the image content area is a privacy information area; performing content extraction on the field content area to obtain field content, and performing field classification on the field content to determine a number of fields contained in the field content; wherein the type of the field includes any one or more of the following: text field, numeric field, symbol field; performing semantic content type identification on the field to determine whether the semantic content type of the field is a privacy information type, and when it is determined that the semantic content type of the field is a privacy information type, determining the position of the field in the field content area as the privacy information area.

[0008] In one implementation of the present application, encoding a private information region based on a random private sequence specifically includes: processing the random private sequence using a preset chaotic cellular automaton to obtain an initial observation matrix; performing non-negative processing on the initial observation matrix using a preset non-negative function to obtain an observation matrix to be applied; and performing sampling compression on the private information region based on the observation matrix to be applied to achieve encoding of the private information region.

[0009] In one implementation of the present application, before extracting a first number of frame images from the encoded credit application video, the method also includes: performing grayscale processing on the encoded provident fund image to be verified and each frame image of the encoded credit application video, specifically including: determining the first chromaticity value of each pixel point in the encoded provident fund image to be verified and each frame image of the encoded credit application video in the RGB model; based on the first chromaticity value, determining the second chromaticity value corresponding to each pixel point in the Lab model, and based on the second chromaticity value, determining the grayscale value of each pixel point in each frame image of the encoded provident fund image to be verified and the encoded credit application video through a preset brightness compensation function; based on the grayscale value, converting the encoded provident fund image to be verified and each frame image of the encoded credit application video into a grayscale key frame image.

[0010] In one implementation of the present application, a first number of frame images are extracted from an encoded credit application video, and key frame images in the first number of frame images are determined, specifically including: determining an adjacent operation time interval, and setting the adjacent operation time interval as a frame image extraction time interval; based on the frame image extraction time interval, performing frame image extraction on the encoded credit application video to obtain a first number of frame images; performing initial clustering on the first number of frame images through an affinity propagation algorithm to obtain a second number of initial key frame images; wherein the second number is smaller than the first number; performing two-layer clustering on the second number of initial key frame images through an SCC clustering algorithm to obtain determined key frame images.

[0011] In one implementation of the present application, based on the random privacy sequence and the preset similarity judgment rule, the similarity between the key frame image and the encoded provident fund image to be verified is determined, specifically including: comparing the positions of the private information area in the key frame image and the encoded provident fund image to be verified; when it is determined that the private information area is in the same position in the key frame image and the encoded provident fund image to be verified, determining the grayscale value corresponding to each pixel point in the non-private information area in the key frame image and the encoded provident fund image to be verified, and determining the grayscale value corresponding to each pixel point in the non-private information area in the key frame image and the encoded provident fund image to be verified. The absolute value of the difference between the grayscale values ​​of the pixels corresponding to the information area is less than the preset difference value of the first pixel pair; based on the random privacy sequence and the compressed sensing recovery program corresponding to the encoding, the original grayscale values ​​corresponding to the pixels in the private information area of ​​the key frame image and the encoded provident fund image to be verified are determined, and the second pixel pair is determined, the absolute value of the difference between the grayscale values ​​of the pixels corresponding to the private information area of ​​the key frame image and the encoded provident fund image to be verified is less than the preset difference; based on the first pixel pair and the second pixel pair, the similarity between the key frame image and the encoded provident fund image to be verified is determined.

[0012] In one implementation of the present application, the non-negative function is represented by the following formula:

[0013]

[0014] in, is the initial observation matrix, is the observation matrix to be applied.

[0015] In a second aspect, an embodiment of the present application also provides a credit review device based on video image detection technology, characterized in that the device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can: upon receiving a credit review request triggered by a user terminal, generate a random privacy sequence corresponding to the credit review request, and send the random privacy sequence to the user terminal; determine the privacy information area in the provident fund image to be verified and the credit application video through a privacy information recognition algorithm preset by the user terminal, and encode the privacy information area based on the random privacy sequence; receive the encoded provident fund image to be verified and the encoded credit application video uploaded by the user terminal, extract a first number of frame images from the encoded credit application video, and determine the key frame images in the first number of frame images; based on the random privacy sequence and a preset similarity judgment rule, determine the similarity between the key frame image and the encoded provident fund image to be verified, and when the similarity is greater than a preset threshold, determine that the provident fund image to be verified originates from the credit application video.

[0016] In a third aspect, an embodiment of the present application further provides a non-volatile computer storage medium for credit review based on video image detection technology, which stores computer executable instructions, and is characterized in that the computer executable instructions are configured to: upon receiving a credit review request triggered by a user terminal, generate a random privacy sequence corresponding to the credit review request, and send the random privacy sequence to the user terminal; determine the privacy information area in the provident fund image to be verified and the credit application video through a privacy information recognition algorithm preset by the user terminal, and encode the privacy information area based on the random privacy sequence; receive the encoded provident fund image to be verified and the encoded credit application video uploaded by the user terminal, extract a first number of frame images from the encoded credit application video, and determine the key frame images in the first number of frame images; determine the similarity between the key frame image and the encoded provident fund image to be verified based on the random privacy sequence and a preset similarity judgment rule, and determine that the provident fund image to be verified originates from the credit application video when the similarity is greater than a preset threshold.

[0017] The embodiments of this application provide a credit review method, device, and storage medium based on video image detection technology. Using a random privacy sequence generated based on a request value, the method encodes the private information areas in the housing fund image to be verified and the credit application video. This ensures randomness while also facilitating the subsequent accurate recovery of the sampled and compressed encoded areas, thus ensuring the privacy of user information and facilitating the determination of fraud risks. By extracting key frames through two-layer clustering and determining the similarity between the key frame images and the encoded housing fund image to be verified, the method effectively implements automated credit review of housing fund loan applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0019] Figure 1 A flowchart of a credit review method based on video image detection technology provided in an embodiment of the present application;

[0020] Figure 2 A schematic diagram of the internal structure of a credit review device based on video image detection technology provided in an embodiment of the present application. DETAILED DESCRIPTION

[0021] To make the purpose, technical solutions, and advantages of this application more clear, the technical solutions of this application will be clearly and completely described below in conjunction with the specific embodiments of this application and the corresponding drawings. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0022] The embodiments of the present application provide a credit review method, device and storage medium based on video image detection technology to solve the following technical problem: how to automatically review provident fund loan applications while protecting user privacy information.

[0023] The technical solutions proposed in the embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0024] Figure 1 This is a flowchart of a credit review method based on video image detection technology provided in an embodiment of the present application. Figure 1 As shown, the embodiment of the present application provides a credit review method based on video image detection technology, which specifically includes the following steps:

[0025] Step 101: upon receiving a credit review request triggered by a user terminal, generate a random privacy sequence corresponding to the credit review request, and send the random privacy sequence to the user terminal.

[0026] In one embodiment of the present application, when a user uploads a housing fund image and a credit application video to be verified on a user terminal, a credit review request is triggered, and the user terminal uploads the credit review request to a remote credit review server. Upon receiving the credit review request, the remote credit review server generates a random privacy sequence corresponding to the credit review request based on the credit review request.

[0027] Specifically, first determine the request value of the credit review request. It should be noted that since the credit review will be archived and saved, each credit review request will be in the record table, and the record order is the trigger order of the credit review request, and the credit review trigger order is the request value in the embodiment of the present application. It should also be noted that the embodiment of the present application generates a subsequent random privacy sequence through the request value. Due to the fixed nature of the request value, the generated random privacy sequence is actually a fixed value after a certain calculation. By using such a random privacy sequence, the provident fund image to be verified and the credit application video are compressed sensing encoded, which is more conducive to accurately recovering the sampled compressed coding area.

[0028] Furthermore, based on the request value, the parameter value of the linear congruential recursive formula to be applied is determined to obtain the linear congruential recursive formula to be applied corresponding to the credit review request.

[0029] It should be noted that the general formula of the linear congruential formula is as follows:

[0030]

[0031]

[0032] in, is the recursive afterword, is the recursive antecedent, a and b are parameter values, M is the modulus, is a random privacy sequence, , and , is the preset initial value, and n is the length of the random privacy sequence to be generated.

[0033] In one embodiment of the present application, in order to ensure the encryption of the provident fund image to be verified and the credit application video, the present application does not fix the parameter values ​​a and b, and uses the request value as the variable value of the parameter values ​​a and b. With initial value Can be set according to needs.

[0034] Furthermore, based on the preset random privacy sequence length, the random privacy sequence corresponding to the credit review request is calculated by applying the linear congruential recursive formula.

[0035] In one embodiment of the present application, after generating a random privacy sequence, the remote credit review server sends the random privacy sequence to the user terminal.

[0036] Step 102: Determine the private information area in the provident fund image and the credit application video to be verified using a private information identification algorithm preset on the user end, and encode the private information area based on a random private sequence.

[0037] In one embodiment of the present application, when the user terminal triggers a credit review request, the user terminal will also determine the private information area in the provident fund image and credit application video to be verified through the privacy information recognition algorithm preset by the user terminal.

[0038] Specifically, based on a preset content type recognition model, the image content area and the field content area in each frame image of the provident fund image and the credit application video to be verified are determined; the image content area is subjected to portrait recognition by means of a portrait recognition algorithm to determine whether the image content of the image content area is a portrait, and when it is determined that the image content area is a portrait, the image content area is determined to be a privacy information area; content extraction is performed on the field content area to obtain field content, and field classification is performed on the field content to determine a number of fields contained in the field content; wherein the type of the field includes any one or more of the following: text field, numeric field, symbol field; semantic content type recognition is performed on the field to determine whether the semantic content type of the field is a privacy information type, and when it is determined that the semantic content type of the field is a privacy information type, the position of the field in the field content area is determined to be the privacy information area.

[0039] In one embodiment of the present application, after determining the private information region in the provident fund image and the credit application video to be verified, the private information region is encoded based on a random private sequence.

[0040] Specifically, a random privacy sequence is processed using a preset chaotic cellular automaton to obtain an initial observation matrix. It should be noted that this application accurately determines whether the provident fund image to be verified originates from the credit application video. Therefore, considering the performance optimization issue when reconstructing the original signal, a chaotic cellular automaton is selected to generate the initial observation matrix. This has advantages in image quality over other random matrices and shortens encryption and decryption time.

[0041] Furthermore, since a general observation matrix may have negative values, these negative values ​​may cause pixel-to-pixel cancellation during the encoding process, thereby causing feature loss and affecting image reconstruction. Therefore, after generating the initial observation matrix, the present application will also use a preset non-negative function to non-negatively process the initial observation matrix to obtain the observation matrix to be applied.

[0042] In one embodiment of the present application, the non-negative function is represented by the following formula:

[0043]

[0044] in, is the initial observation matrix, is the observation matrix to be applied.

[0045] Furthermore, based on the observation matrix to be applied, the private information region is sampled and compressed to achieve encoding of the private information region.

[0046] Step 103: Receive the encoded provident fund image to be verified and the encoded credit application video uploaded by the user, extract a first number of frame images from the encoded credit application video, and determine key frame images in the first number of frame images.

[0047] In one embodiment of the present application, after the user terminal encodes the private information area in the provident fund image to be verified and the credit application video, the encoded provident fund image to be verified and the encoded credit application video are uploaded to the remote credit review server. After receiving the encoded provident fund image to be verified and the encoded credit application video, the remote credit review server will first grayscale each frame image of the encoded provident fund image to be verified and the encoded credit application video to facilitate subsequent similarity judgment.

[0048] Specifically, the first chromaticity value of each pixel in each frame image of the encoded provident fund image to be verified and the encoded credit application video in the RGB model is determined; based on the first chromaticity value, the second chromaticity value corresponding to each pixel in the Lab model is determined, and based on the second chromaticity value, the grayscale value of each pixel in each frame image of the encoded provident fund image to be verified and the encoded credit application video is determined through a preset brightness compensation function; based on the grayscale value, the encoded provident fund image to be verified and the encoded credit application video are converted into grayscale key frame images.

[0049] It should be noted that the brightness compensation function of this application is expressed by the following formula:

[0050]

[0051] in, is the gray value of the i-th pixel, is the brightness value of the i-th pixel in the Lab model, is the first channel value of the i-th pixel in the Lab model, is the first compensation coefficient corresponding to the first channel value, is the second channel value of the i-th pixel in the Lab model, is the second compensation coefficient corresponding to the second channel value.

[0052] It should be noted that the first compensation coefficient in the brightness compensation function and the second compensation coefficient The compensation coefficients are obtained using a preset compensation coefficient model. Therefore, before determining the grayscale value of each pixel in each frame of the encoded provident fund image to be verified and the encoded credit application video using a preset luminance compensation function based on the second chromaticity value, the first and second compensation coefficients corresponding to each pixel must be determined using the compensation coefficient model. The compensation coefficient determination model is a converged model obtained through training on a number of sample images.

[0053] In one embodiment of the present application, after grayscale processing is performed on each frame image of the encoded provident fund image to be verified and the encoded credit application video, a first number of frame images are extracted from the encoded credit application video, and key frame images among the first number of frame images are determined.

[0054] Specifically, the time interval between adjacent operations is determined and set as the frame image extraction time interval. It is understandable that the time interval between adjacent operations is the shortest time interval between user operations during the credit application process.

[0055] Furthermore, based on the frame image extraction time interval, frame image extraction is performed on the encoded credit application video to obtain a first number of frame images.

[0056] Furthermore, the first number of frame images are initially clustered using an affinity propagation algorithm to obtain a second number of initial key frame images; wherein the second number is smaller than the first number; and the second number of initial key frame images are subjected to two-layer clustering using an SCC clustering algorithm to obtain determined key frame images.

[0057] It should be noted that, since this application cannot determine how many operation interfaces the user has experienced in total when extracting frame images from the encoded credit application video, and thus cannot determine the number of clusters, general clustering algorithms need to determine the number of clusters before clustering in order to classify the data. The affinity propagation algorithm does not need to set relevant parameters in the process of extracting key frames. Therefore, this application uses the affinity propagation algorithm to perform initial clustering on the first number of frame images. Due to the limitations of the clustering effect of the affinity propagation algorithm, in order to avoid repeated frame images, a two-layer clustering method is used to extract key frames. In the second layer of this application, the second number of initial key frame images are clustered by the SCC clustering algorithm to obtain a certain key frame image.

[0058] Step 104: Based on the random privacy sequence and the preset similarity judgment rule, determine the similarity between the key frame image and the encoded provident fund image to be verified, and if the similarity is greater than a preset threshold, determine that the provident fund image to be verified comes from the credit application video.

[0059] In one embodiment of the present application, after determining the key frame images in the first number of frame images, the similarity between the key frame images and the encoded provident fund image to be verified is determined based on the random privacy sequence and a preset similarity judgment rule.

[0060] Specifically, the positions of the private information area in the key frame image and the encoded provident fund image to be verified are compared; when it is determined that the private information area is located at the same position in the key frame image and the encoded provident fund image to be verified, the grayscale value corresponding to each pixel in the non-private information area in the key frame image and the encoded provident fund image to be verified is determined, and the number of first pixel pairs whose absolute value of the difference in grayscale value corresponding to the pixel in the non-private information area in the key frame image and the encoded provident fund image to be verified is determined to be less than a preset difference.

[0061] Furthermore, based on the random privacy sequence and the compressed sensing recovery program corresponding to the encoding, the original grayscale values ​​corresponding to each pixel in the private information area of ​​the key frame image and the encoded provident fund image to be verified are determined, and the second pixel pairwise number whose absolute value of the difference between the grayscale values ​​of the pixels corresponding to the private information area of ​​the key frame image and the encoded provident fund image to be verified is less than a preset difference is determined; based on the first pixel pairwise number and the second pixel pairwise number, the similarity between the key frame image and the encoded provident fund image to be verified is determined.

[0062] It is understandable that although the image is encoded using a random privacy sequence, the parameter values ​​involved in the encoding are all fixed values. Through the inverse encoding process, namely the compressed sensing recovery process, the original chromaticity value corresponding to each pixel in the private information area can be determined, and its corresponding grayscale value can also be determined.

[0063] It can also be understood that the smaller the sum of the first pixel logarithm and the second pixel logarithm, the higher the similarity. Since the provident fund image should be one of the key frame images, except for the unavoidable color value error caused by the screenshot factor, the preset threshold should be small enough or zero.

[0064] The above is an embodiment of the method proposed in this application. Based on the same inventive concept, this application embodiment also provides a credit review device based on video image detection technology, the structure of which is as follows: Figure 2 shown.

[0065] Figure 2 This is a schematic diagram of the internal structure of a credit review device based on video image detection technology provided in an embodiment of the present application. Figure 2 As shown, the equipment includes:

[0066] at least one processor 201;

[0067] and, a memory 202 communicatively coupled to the at least one processor;

[0068] The memory 202 stores instructions that can be executed by at least one processor, and the instructions are executed by the at least one processor 201 to enable the at least one processor 201 to:

[0069] Upon receiving a credit review request triggered by a user terminal, generating a random privacy sequence corresponding to the credit review request and sending the random privacy sequence to the user terminal;

[0070] Using a privacy information recognition algorithm pre-set on the user end, the private information areas in the provident fund image and credit application video to be verified are determined, and the private information areas are encoded based on a random privacy sequence.

[0071] receiving the encoded provident fund image to be verified and the encoded credit application video uploaded by the user, extracting a first number of frame images from the encoded credit application video, and determining key frame images among the first number of frame images;

[0072] Based on the random privacy sequence and the preset similarity judgment rules, the similarity between the key frame image and the encoded provident fund image to be verified is determined, and when the similarity is greater than the preset threshold, it is determined that the provident fund image to be verified comes from the credit application video.

[0073] Some embodiments of the present application provide corresponding Figure 1 A non-volatile computer storage medium based on video image detection technology stores computer executable instructions, wherein the computer executable instructions are set to:

[0074] Upon receiving a credit review request triggered by a user terminal, generating a random privacy sequence corresponding to the credit review request and sending the random privacy sequence to the user terminal;

[0075] Using a privacy information recognition algorithm pre-set on the user end, the private information areas in the provident fund image and credit application video to be verified are determined, and the private information areas are encoded based on a random privacy sequence.

[0076] receiving the encoded provident fund image to be verified and the encoded credit application video uploaded by the user, extracting a first number of frame images from the encoded credit application video, and determining key frame images among the first number of frame images;

[0077] Based on the random privacy sequence and the preset similarity judgment rules, the similarity between the key frame image and the encoded provident fund image to be verified is determined, and when the similarity is greater than the preset threshold, it is determined that the provident fund image to be verified comes from the credit application video.

[0078] The various embodiments in this application are described in a progressive manner. Similar portions between the various embodiments can be referenced to each other. Each embodiment focuses on the differences from the other embodiments. In particular, the IoT device and media embodiments are generally similar to the method embodiments, so their description is relatively simple. For relevant portions, refer to the description of the method embodiments.

[0079] The system and medium provided in the embodiments of the present application correspond one-to-one to the method. Therefore, the system and medium also have similar beneficial technical effects to their corresponding methods. Since the beneficial technical effects of the method have been described in detail above, the beneficial technical effects of the system and medium will not be repeated here.

[0080] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0081] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0082] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0083] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 The steps for the function specified in one or more boxes.

[0084] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0085] Memory may include non-permanent storage in a computer-readable medium, in the form of random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0086] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can be implemented using any method or technology to store information. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change RAM (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media such as modulated data signals and carrier waves.

[0087] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0088] The foregoing is merely an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. A credit review method based on video image detection technology, characterized in that: The method comprises: Upon receiving a credit review request triggered by a user terminal, generating a random privacy sequence corresponding to the credit review request, and sending the random privacy sequence to the user terminal; Determining the private information region in the provident fund image and the credit application video to be verified by using a privacy information recognition algorithm preset by the user terminal, and encoding the private information region based on the random privacy sequence; receiving the encoded provident fund image to be verified and the encoded credit application video uploaded by the user, extracting a first number of frame images from the encoded credit application video, and determining key frame images among the first number of frame images; Based on the random privacy sequence and a preset similarity judgment rule, determining the similarity between the key frame image and the encoded provident fund image to be verified, and if the similarity is greater than a preset threshold, determining that the provident fund image to be verified is from the credit application video; Generate a random privacy sequence, including: Determining a request value of the credit review request; wherein the request value is used to describe a triggering order of the credit review request; Determining parameter values ​​of a linear congruential recursion formula to be applied based on the request value to obtain a linear congruential recursion formula to be applied corresponding to the credit review request; Based on a preset random privacy sequence length, calculating the random privacy sequence corresponding to the credit review request by using the linear congruential recursive formula to be applied; The privacy information identification algorithm preset by the user terminal is used to determine the privacy information area in the provident fund image and the credit application video to be verified, specifically including: Determining, based on a preset content type recognition model, image content areas and field content areas in each frame of the provident fund image to be verified and the credit application video; performing portrait recognition on the image content region using a portrait recognition algorithm to determine whether the image content of the image content region is a portrait, and if it is determined that the image content region is a portrait, determining the image content region as a private information region; Extracting content from the field content area to obtain field content, and classifying the field content to determine a number of fields included in the field content; wherein the field type includes any one or more of the following: a text field, a numeric field, and a symbol field; The field is semantically identified to determine whether the semantic content type of the field is a privacy information type, and if it is determined that the semantic content type of the field is a privacy information type, the position of the field in the field content area is determined to be a privacy information area.

2. The credit review method based on video image detection technology according to claim 1, characterized in that: Encoding the private information area based on the random private sequence specifically includes: Processing the random privacy sequence through a preset chaotic cellular automaton to obtain an initial observation matrix; Performing non-negative processing on the initial measurement matrix by using a preset non-negative function to obtain a measurement matrix to be applied; Based on the observation matrix to be applied, sampling compression is performed on the private information region to achieve encoding of the private information region.

3. The credit review method based on video image detection technology according to claim 2, characterized in that: Before extracting the first number of frame images from the encoded credit application video, the method further includes: Grayscale processing is performed on the encoded provident fund image to be verified and each frame image of the encoded credit application video, specifically including: Determining a first chromaticity value of each pixel in each frame of the encoded provident fund image to be verified and the encoded credit application video in an RGB model; Based on the first chromaticity value, determining a second chromaticity value corresponding to each pixel in the Lab model, and based on the second chromaticity value, determining the grayscale value of each pixel in each frame of the encoded provident fund image to be verified and the encoded credit application video using a preset brightness compensation function; Based on the grayscale value, the encoded provident fund image to be verified and each frame image of the encoded credit application video are converted into grayscale key frame images.

4. The credit review method based on video image detection technology according to claim 3 is characterized in that: Extracting a first number of frame images from the encoded credit application video and determining key frame images among the first number of frame images specifically includes: Determine a time interval between adjacent operations, and set the time interval between adjacent operations as a frame image extraction time interval; performing frame image extraction on the encoded credit application video based on the frame image extraction time interval to obtain a first number of frame images; Performing initial clustering on the first number of frame images using an affinity propagation algorithm to obtain a second number of initial key frame images; wherein the second number is smaller than the first number; Perform two-layer clustering on the second number of initial key frame images using an SCC clustering algorithm to obtain determined key frame images.

5. The credit review method based on video image detection technology according to claim 4, characterized in that: Determining the similarity between the key frame image and the encoded provident fund image to be verified based on the random privacy sequence and a preset similarity judgment rule specifically includes: Comparing the positions of the private information area in the key frame image and the encoded provident fund image to be verified; When it is determined that the private information area is located at the same position in the key frame image and the encoded provident fund image to be verified, determining the grayscale value corresponding to each pixel in the non-private information area in the key frame image and the encoded provident fund image to be verified, and determining the first number of pixel pairs for which the absolute value of the difference in grayscale value corresponding to the pixel in the non-private information area in the key frame image and the encoded provident fund image to be verified is less than a preset difference value; Based on the random privacy sequence and the compressed sensing recovery program corresponding to the encoding, determining the original grayscale value corresponding to each pixel in the private information area of ​​the key frame image and the encoded provident fund image to be verified, and determining the second number of pixel pairs for which the absolute value of the difference between the grayscale values ​​of the pixels corresponding to the private information area of ​​the key frame image and the encoded provident fund image to be verified is less than a preset difference; Based on the first pixel pair number and the second pixel pair number, the similarity between the key frame image and the encoded provident fund image to be verified is determined.

6. The credit review method based on video image detection technology according to claim 2, characterized in that: The non-negative function is expressed by the following formula: in, is the initial observation matrix, is the observation matrix to be applied.

7. A credit review device based on video image detection technology, characterized in that: The device comprises: at least one processor; and, a memory communicatively coupled to the at least one processor; The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to: Upon receiving a credit review request triggered by a user terminal, generating a random privacy sequence corresponding to the credit review request, and sending the random privacy sequence to the user terminal; Determining the private information region in the provident fund image and the credit application video to be verified by using a privacy information recognition algorithm preset by the user terminal, and encoding the private information region based on the random privacy sequence; receiving the encoded provident fund image to be verified and the encoded credit application video uploaded by the user, extracting a first number of frame images from the encoded credit application video, and determining key frame images among the first number of frame images; Based on the random privacy sequence and a preset similarity judgment rule, determining the similarity between the key frame image and the encoded provident fund image to be verified, and if the similarity is greater than a preset threshold, determining that the provident fund image to be verified is from the credit application video; Generate a random privacy sequence, including: Determining a request value of the credit review request; wherein the request value is used to describe a triggering order of the credit review request; Determining parameter values ​​of a linear congruential recursion formula to be applied based on the request value to obtain a linear congruential recursion formula to be applied corresponding to the credit review request; Based on a preset random privacy sequence length, calculating the random privacy sequence corresponding to the credit review request by using the linear congruential recursive formula to be applied; Using a privacy information recognition algorithm preset on the user end, the privacy information region in the provident fund image and the credit application video to be verified is determined, specifically including: Determining, based on a preset content type recognition model, image content areas and field content areas in each frame of the provident fund image to be verified and the credit application video; performing portrait recognition on the image content region using a portrait recognition algorithm to determine whether the image content of the image content region is a portrait, and if it is determined that the image content region is a portrait, determining the image content region as a private information region; Extracting content from the field content area to obtain field content, and classifying the field content to determine a number of fields included in the field content; wherein the field type includes any one or more of the following: a text field, a numeric field, and a symbol field; The field is semantically identified to determine whether the semantic content type of the field is a privacy information type, and if it is determined that the semantic content type of the field is a privacy information type, the position of the field in the field content area is determined to be a privacy information area.

8. A non-volatile computer storage medium based on video image detection technology, storing computer executable instructions, characterized in that: The computer executable instructions are configured to: Upon receiving a credit review request triggered by a user terminal, generating a random privacy sequence corresponding to the credit review request, and sending the random privacy sequence to the user terminal; Determining the private information region in the provident fund image and the credit application video to be verified by using a privacy information recognition algorithm preset by the user terminal, and encoding the private information region based on the random privacy sequence; receiving the encoded provident fund image to be verified and the encoded credit application video uploaded by the user, extracting a first number of frame images from the encoded credit application video, and determining key frame images among the first number of frame images; Based on the random privacy sequence and a preset similarity judgment rule, determining the similarity between the key frame image and the encoded provident fund image to be verified, and if the similarity is greater than a preset threshold, determining that the provident fund image to be verified is from the credit application video; Generate a random privacy sequence, including: Determining a request value of the credit review request; wherein the request value is used to describe a triggering order of the credit review request; Determining parameter values ​​of a linear congruential recursion formula to be applied based on the request value to obtain a linear congruential recursion formula to be applied corresponding to the credit review request; Based on a preset random privacy sequence length, calculating the random privacy sequence corresponding to the credit review request by using the linear congruential recursive formula to be applied; Using a privacy information recognition algorithm preset on the user end, the privacy information region in the provident fund image and the credit application video to be verified is determined, specifically including: Determining, based on a preset content type recognition model, image content areas and field content areas in each frame of the provident fund image to be verified and the credit application video; performing portrait recognition on the image content region using a portrait recognition algorithm to determine whether the image content of the image content region is a portrait, and if it is determined that the image content region is a portrait, determining the image content region as a private information region; Extracting content from the field content area to obtain field content, and classifying the field content to determine a number of fields included in the field content; wherein the field type includes any one or more of the following: a text field, a numeric field, and a symbol field; The field is semantically identified to determine whether the semantic content type of the field is a privacy information type, and if it is determined that the semantic content type of the field is a privacy information type, the position of the field in the field content area is determined to be a privacy information area.

Images