Real estate title certificate recognition and entry method, device and medium based on adaptive algorithm

Abstract

The present invention provides a real estate certificate recognition and entry method, device and medium based on an adaptive algorithm, the method comprising: step S1: obtaining real estate certificate images in different states, constructing a training set and a test set; step S2: constructing an adaptive scene discrimination model, a nonlinear activation free network NAFNet model and an optical character recognition module to realize the recognition of characters in the real estate certificate image; step S3: extracting key information from the recognized characters, encrypting the extracted key information, and automatically entering the encrypted key information into a structured database to realize digital management and storage of real estate certificate data. The present invention has the advantages of high efficiency and good security, can effectively ensure the security and integrity of real estate data, and provides strong technical support for real estate transactions and management.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and in particular to a method, device and medium for identifying and entering a real estate title certificate based on an adaptive algorithm. Background Art

[0002] As the real estate market develops, the need for automatic identification and evaluation of real estate certificates is becoming increasingly urgent. By using image recognition technology, text information can be automatically extracted from scanned certificates; while natural language processing technology can help understand and classify this information. This not only reduces the demand for human resources, but also greatly reduces the risk caused by manual errors. In addition, the automated system can update and store data in real time, providing faster and more convenient access, thereby enhancing market transparency and efficiency.

[0003] However, developing an effective real estate certificate recognition and entry system faces many challenges. The first is the diversity of certificate formats and quality. Real estate certificates from different regions and institutions vary greatly in format and quality, which requires the system to have strong adaptability, be able to handle various types of documents, and accurately extract key information from them. The second is data security and privacy protection. Real estate transactions are highly sensitive and have legal requirements, and ensuring data security and privacy protection is crucial. Summary of the invention

[0004] The technical problem to be solved by the present invention is: in response to the technical problems existing in the prior art, the present invention provides a method, device and medium for identifying and entering real estate title certificates based on an adaptive algorithm with high efficiency and good security, which can effectively ensure the security and integrity of real estate data and provide strong technical support for real estate transactions and management.

[0005] In order to solve the above technical problems, the technical solution proposed by the present invention is:

[0006] A method for identifying and entering a real estate title certificate based on an adaptive algorithm comprises the following steps:

[0007] Step S01: obtaining images of real estate certificates in different states to construct a training set and a test set;

[0008] Step S02: Adaptive algorithm recognition model construction;

[0009] Constructing an adaptive scene discrimination model to determine whether the clarity of the test set image meets the requirements;

[0010] Constructing a nonlinear activation free network NAFNet, and when the adaptive scene discrimination model determines that the clarity of the real estate certificate image does not meet the requirements, using the nonlinear activation free network NAFNet to deblur the real estate certificate image that does not meet the requirements, so as to obtain the real estate certificate image whose clarity meets the preset conditions;

[0011] Construct an optical character recognition module to perform text recognition on images of real estate certificates whose clarity meets preset conditions;

[0012] Step S03: Extraction, encryption and entry of key information;

[0013] Extract key information from the recognized text, encrypt the extracted key information, and automatically enter the encrypted key information into a structured database to achieve the management and storage of real estate certificate data;

[0014] The method for judging the image clarity of a real estate certificate using the adaptive scene discrimination model includes:

[0015] Step S201: inputting the real estate certificate image into a 5×1 convolution layer, a 1×5 convolution layer, and a 2×2 average pooling layer in sequence to implement the first downsampling process of the real estate certificate image;

[0016] Step S202: inputting the first downsampling processing result into a 3×3 convolutional layer and multiple 3×3 dilated convolutional layers with different dilation rates, and performing a first feature fusion on multiple feature maps output by the 3×3 convolutional layer and the 3×3 dilated convolutional layer;

[0017] Step S203: the output obtained by inputting the first feature fusion result into the asymmetric convolution layer is added element by element for the first time with the output result of the 3×3 convolution layer before the first feature fusion, and the first element-by-element addition result is input into the 2×2 average pooling layer for the second downsampling;

[0018] Step S204: input the second downsampling result into a 3×3 convolutional layer and multiple 3×3 hole convolutional layers with different expansion rates to obtain multiple feature maps, and perform a second feature fusion on the multiple feature maps;

[0019] Step S205: the output obtained by inputting the second feature fusion result into the asymmetric convolution layer and the output of the 3×3 convolution layer before the second feature fusion are added element by element for the second time;

[0020] Step S206: After the second element-by-element addition result is input into the 3×3 convolution layer, the global average pooling layer and the fully connected layer in sequence, the clarity category of the real estate certificate image is obtained through the Softmax function, and the clarity category of the real estate certificate image includes clear and blurred.

[0021] As a further improvement of the method of the present invention: in the step S02, the adaptive scene discrimination model includes multiple asymmetric convolutions, multiple hole convolutions with different receptive fields, a feature fusion mechanism and an element-by-element addition mechanism, and the asymmetric convolution means that the length and width of the convolution kernel are different.

[0022] As a further improvement of the method of the present invention: the step S02 also includes adjusting the contrast of the real estate certificate image adjusted using the nonlinear activation free network NAFNet to enhance the readability of the real estate certificate image.

[0023] As a further improvement of the method of the present invention: when the contrast of the real estate certificate image adjusted by the nonlinear activation free network NAFNet is adjusted, the value of each pixel in the image is processed by using the following formula to enhance the contrast of the real estate certificate image:

[0024] ,

[0025] in, is the pixel value in the output image, is the pixel value in the original image, is the adjustment factor, is the average pixel value of the original image.

[0026] As a further improvement of the method of the present invention: in the step S02, the method of using the optical character recognition module to perform text recognition on the image of the real estate certificate whose clarity meets the preset conditions includes:

[0027] Step S211: using an optical character recognition module to detect the image to obtain a text detection frame;

[0028] Step S212: after correcting the text detection frame, recognize the characters in the text detection frame.

[0029] As a further improvement of the method of the present invention: the extraction of key information from the recognized text includes: using a general information extraction unified framework to read a predefined real estate certificate keyword list, and performing targeted information screening and extraction on the text in the recognized text detection box to achieve accurate matching and extraction of key information.

[0030] As a further improvement of the method of the present invention: in the step S03, the encryption processing is to start the encryption program before the data enters the database layer, use the symmetric encryption algorithm to quickly encrypt the data, use the asymmetric encryption algorithm to generate a public key and a private key through a random number generator, store the private key in the hardware security module, and the public key is used to encrypt the symmetric encryption key, and the encrypted data and the encrypted symmetric key are associated and stored or transmitted, thereby realizing encryption of the data.

[0031] The present invention also provides a computer device, comprising a processor and a memory, wherein the memory is used to store a computer program, and the processor is used to execute the computer program to perform the above method.

[0032] The present invention also provides a computer-readable storage medium storing a computer program, and the computer program implements the above method when executed.

[0033] Compared with the prior art, the present invention has the following beneficial effects:

[0034] The present invention forms a solution for real estate property certificate through a closed-loop process of adaptive scene discrimination model, NAFNet deblurring, optical character recognition and encrypted storage, and realizes automatic recognition of the clarity of real estate property certificate image. When the adaptive scene discrimination model determines that the image clarity does not meet the preset conditions, the NAFNet model deblurs the image that does not meet the clarity requirement to obtain an image with clarity that meets the requirement. Through the linkage mechanism of the adaptive scene discrimination model and the NAFNet model, the system triggers the deblurring process only when necessary, effectively avoiding redundant calculations. The optical character recognition module performs text recognition on the clear real estate property certificate image and extracts key information from the recognized text, and encrypts the extracted key information and enters it into a structured database. The present invention realizes fast and accurate extraction and entry of key information, which not only improves the quality of real estate property certificate image processing and the accuracy of text recognition, but also enhances data security and privacy protection, improves the universality and flexibility of the system, and the efficiency and security of real estate business, thereby playing an important role in real estate transactions and management. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 This is a flow chart of a method for identifying and entering a real estate title certificate based on an adaptive algorithm according to an embodiment of the present invention.

[0036] Figure 2 The figure is a schematic diagram of the structure principle of the adaptive scene discrimination model according to an embodiment of the present invention.

[0037] Figure 3 This is a flow chart of a method for identifying and entering a real estate title certificate based on an adaptive algorithm according to a specific application embodiment of the present invention. DETAILED DESCRIPTION

[0038] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0039] In this embodiment, an adaptive scene discrimination model is first constructed to determine whether the image needs to be deblurred. Then, an optical character recognition module is used to recognize the text in the image. Then, a general information extraction unified framework is used to extract the key information of the real estate certificate in the text. Finally, the database logic is used to encrypt the relevant information of the real estate certificate and then structure it for processing and input. The detailed scheme is as follows:

[0040] like Figure 1 As shown, the real estate title certificate recognition and entry method based on the adaptive algorithm in this embodiment includes the following steps:

[0041] Step S01: Obtain real estate title certificate images in different states to construct a training set and a test set.

[0042] Among them, the real estate certificate images in different states include normal images and blurred images.

[0043] Step S02: Adaptive algorithm recognition model construction;

[0044] An adaptive scene discrimination model is constructed to determine whether the clarity of the test set images meets the requirements.

[0045] In this embodiment, the adaptive scene discrimination model includes multiple asymmetric convolutions, multiple hole convolutions with different receptive fields, a feature fusion mechanism, and an element-by-element addition mechanism, wherein the asymmetric convolution refers to the different lengths and widths of the convolution kernels.

[0046] In this embodiment, the hole convolutions of multiple different receptive fields of the adaptive scene discrimination model can effectively improve the model's ability to extract large-size image features. The asymmetric convolution effectively reduces the amount of calculation and improves the model's operating efficiency. The feature fusion and feature addition mechanisms can effectively improve the model's ability to extract image features.

[0047] In this embodiment, the method for the adaptive scene recognition model to determine the image clarity of the real estate certificate includes:

[0048] Step S201: inputting the real estate certificate image into a 5×1 convolution layer, a 1×5 convolution layer, a 1×5 convolution layer, and a 2×2 average pooling layer in sequence to implement the first downsampling process of the real estate certificate image;

[0049] Step S202: inputting the first downsampling processing result into a 3×3 convolutional layer and multiple 3×3 dilated convolutional layers with different dilation rates, and performing a first feature fusion on multiple feature maps output by the 3×3 convolutional layer and the 3×3 dilated convolutional layer;

[0050] Step S203: the output obtained by inputting the first feature fusion result into the asymmetric convolution layer is added element by element for the first time with the output result of the 3×3 convolution layer before the first feature fusion, and the first element-by-element addition result is input into the 2×2 average pooling layer for the second downsampling;

[0051] Step S204: the second downsampling result is input into a 3×3 convolutional layer and multiple 3×3 hole convolutional layers with different expansion rates to obtain multiple feature maps, and the multiple feature maps are subjected to a second feature fusion;

[0052] Step S205: the output obtained by inputting the second feature fusion result into the asymmetric convolution layer and the output of the 3×3 convolution layer before the second feature fusion are added element by element for the second time;

[0053] Step S206: After the second element-by-element addition result is input into the 3×3 convolution layer, the global average pooling layer and the fully connected layer in sequence, the clarity category of the real estate certificate image is obtained through the Softmax function. The clarity categories of the real estate certificate image include clear and blurred.

[0054] The network architecture of the adaptive scene discrimination model constructed in this embodiment includes an asymmetric convolution layer, a hole convolution layer, multi-scale feature fusion and element-by-element addition operation. Among them, the hole convolution layer captures multi-scale context information through different expansion rates, effectively enhancing the sensitivity to blurred areas; the asymmetric convolution layer reduces the number of parameters while retaining key features, improving computational efficiency; the two feature fusions and downsampling optimize the hierarchical nature of feature expression, effectively avoiding information loss, significantly improving the accuracy of clarity discrimination, and reducing the misjudgment rate.

[0055] like Figure 2 As shown, in a specific application embodiment, the adaptive scene discrimination model structure includes:

[0056] (1) The image is input to the 5×1 convolution layer, the output is input to the 1×5 convolution layer, and the output of the 1×5 convolution layer is input to the 2×2 average pooling layer, thereby performing the first downsampling. This can effectively simplify the network calculation complexity, achieve feature compression, and extract the main features.

[0057] (2) The output of the 2×2 average pooling layer is input to a 3×3 convolutional layer and four 3×3 dilated convolutional layers with dilation rates of 3, 6, 9, and 12 respectively. The outputs of the convolutional layer and the dilated convolutional layer are first fused by splicing multiple feature maps in the depth dimension to obtain richer feature expressions. It is understandable that the choice of dilation rate can be changed according to actual project requirements.

[0058] (3) The output of the first feature fusion is input to the 3×1 convolution layer, and the output obtained is input to the 1×3 convolution layer. The output obtained by the 1×3 convolution layer is added element-by-element to the output of the 3×3 convolution layer before the first feature fusion. The output after the first addition is input to the 2×2 average pooling layer for the second downsampling.

[0059] (4) The output of the 2×2 average pooling layer is input into a 3×3 convolutional layer and three 3×3 dilated convolutional layers with dilation rates of 3, 6, and 9, respectively. The outputs of the convolutional layer and the dilated convolutional layer are subjected to a second feature fusion by concatenating multiple feature maps in the depth dimension.

[0060] (5) The output of the second feature fusion is input to the 3×1 convolution layer, and the output obtained is input to the 1×3 convolution layer. The output obtained by the 1×3 convolution layer is added element-by-element to the output of the 3×3 convolution layer before the second feature fusion.

[0061] (6) The output obtained by the second addition is input into a 3×3 convolutional layer. The output of the 3×3 convolutional layer is then input into a global average pooling layer. The output of the global average pooling layer is input into the fully connected layer and after the Softmax function, the category with the highest probability is the category predicted by the network for the input image.

[0062] In this embodiment, a nonlinear activated free network NAFNet model is constructed. When the adaptive scene discrimination model determines that the clarity of the test set image does not meet the preset conditions, the nonlinear activated free network NAFNet is used to deblur the real estate certificate image that does not meet the preset conditions to obtain the real estate certificate image whose clarity meets the preset conditions.

[0063] In this embodiment, contrast adjustment is performed on the real estate certificate image adjusted by using the nonlinear activation free network NAFNet to enhance the readability of the real estate certificate image.

[0064] In this embodiment, when the contrast of the real estate certificate image adjusted by using the nonlinear activation free network NAFNet is adjusted, the value of each pixel in the image is processed by using the following formula to enhance the contrast of the real estate certificate image:

[0065] , (1)

[0066] in, is the pixel value in the output image, is the pixel value in the original image, is the adjustment factor, is the average pixel value of the original image. It can be understood that the adjustment factor C can be selected according to actual application needs. In this embodiment, it is set to 1.5.

[0067] In this embodiment, an optical character recognition module is constructed to perform text recognition on an image of a real estate certificate whose clarity meets a preset condition.

[0068] In this embodiment, the method of using the optical character recognition module to perform text recognition on the image of the real estate certificate whose clarity meets the preset conditions includes:

[0069] Step S211: Use an optical character recognition module to detect the image to obtain a text detection frame; in this embodiment, the PP-OCRv3 module is selected as the optical character recognition module. It can be understood that the specific optical character recognition module can be selected according to actual needs.

[0070] Step S212: After correcting the text detection frame, recognize the text in the text detection frame.

[0071] Step S03: Extraction, encryption and entry of key information;

[0072] The key information in the recognized text is extracted, encrypted, and automatically entered into a structured database to achieve digital management and storage of real estate certificate data.

[0073] In this embodiment, a general information extraction unified framework is used to read a predefined real estate certificate keyword list, and targeted information screening and extraction are performed on the text in the identified text detection box to achieve accurate matching and extraction of key information.

[0074] In this embodiment, the encryption process is to start the encryption program before the data enters the database layer, use a symmetric encryption algorithm to quickly encrypt the data, use an asymmetric encryption algorithm to generate a public key and a private key through a random number generator, store the private key in the hardware security module, and use the public key to encrypt the symmetric encryption key. The encrypted data and the encrypted symmetric key are associated and stored or transmitted, thereby realizing encryption of the data.

[0075] Specifically, in-application encryption involves several key steps that integrate encryption logic into specific parts of the application to ensure that data is encrypted before entering the database. This method allows for fine-grained encryption of data, that is, different encryption strategies can be selected based on the type and sensitivity of the data. The detailed steps are as follows:

[0076] Step 1: Implementation of aspect-oriented encryption (AOE);

[0077] In the application architecture, determine the key path for data to flow from the application layer to the database layer, insert encryption logic on this path, and ensure that the encryption operation is triggered when the data is about to leave the application layer and enter the database layer. This means that the data is encrypted before it leaves the application layer and enters the database layer. For example, in the data processing flow of the application, add the aspect code, and when the data meets the encryption conditions (such as key information extraction is completed and ready to be stored in the database), call the encryption module to encrypt the data.

[0078] Step 2: Selection and implementation of encryption algorithm;

[0079] For the selection of encryption algorithms: Choose the appropriate encryption algorithm according to the type and sensitivity of the data. For example, for large amounts of data encryption, the symmetric encryption algorithm AES (Advanced Encryption Standard) has a higher encryption speed, so AES is selected to encrypt the data. For data that requires higher security (such as encryption keys for critical information), the asymmetric encryption algorithm RSA (Rivest–Shamir–Adleman) can be selected. Among them, the AES algorithm selects the appropriate key length based on security requirements, and the common ones are 128 bits, 192 bits, or 256 bits. For the RSA algorithm, it is used to generate a public key and a private key pair, the public key is used for encryption, and the private key is used for decryption.

[0080] For the implementation of the encryption algorithm: When using the AES algorithm, a random AES key is generated using a random number generator, the key information to be stored is divided into blocks according to the selected AES mode, the plaintext information is divided into blocks of fixed length (for example, 128 bits), and then each block is encrypted using the AES algorithm and the generated key.

[0081] When the RSA algorithm is used, the generated public key is used to encrypt sensitive information. For example, for keys that need long-term protection, the RSA public key can be used to convert them into ciphertext.

[0082] In this embodiment, after a pair of public and private keys are generated by using the RSA algorithm, a large amount of data is encrypted using AES to generate AES encrypted data, and then the AES key is encrypted using the RSA public key to obtain the encrypted AES key. Finally, the encrypted AES key and the AES encrypted data are stored together in a database. This method ensures both the efficiency of data encryption and the security of the AES key. By combining symmetric and asymmetric encryption technologies, the efficiency and security of the encryption process are effectively improved.

[0083] Step 3: Operation of the Key Management System (KMS);

[0084] The key management system is used to generate, store, distribute and periodically replace keys. In addition, the key management system is also responsible for monitoring and logging all key-related activities to ensure the integrity and security of encryption.

[0085] In the database storage of this embodiment, the database designs a structured database to optimize data storage, retrieval and maintenance. The database schema design should fully consider the type, linkage relationship and query efficiency of the real estate certificate data. Secondly, in terms of secure data storage, the encrypted data is stored in the database to ensure the security and integrity of the data.

[0086] The present invention is further described below by taking the above method for real estate title certificate recognition and entry based on adaptive algorithm in a specific application embodiment as an example. Figure 3 As shown, the detailed steps are:

[0087] Step 1: Obtain images of real estate certificates in different states and construct training sets and test sets;

[0088] Step 2: Build an adaptive scene discrimination model and train the model using the training set;

[0089] Step 3: Save the trained adaptive scene discrimination model and use the model to determine whether the test set image is a blurred image;

[0090] Step 4: Build the NAFNet model and automatically decide whether to use the NAFNet model to process the image based on the image discrimination result of the adaptive scene discrimination model;

[0091] Step 5: Adjust the contrast of the image after step 4;

[0092] Step 6: Build a PP-OCRv3 system and use PP-OCRv3 to recognize the text in the image after step S5;

[0093] Step 7: Use the general information extraction unified framework to extract key information related to the real estate certificate from the text extracted in step S6.

[0094] Step 8: The key information generated in step 7 is encrypted within the AOE application, and the core data is encrypted by combining symmetric and asymmetric encryption technologies before data storage.

[0095] Step 9: The encrypted data in step 8 is automatically entered into the structured database as a module of real estate rights data elements to establish a digital asset library for real estate certificates.

[0096] This embodiment also provides a computer device, including a processor and a memory, the memory is used to store computer programs, and the processor is used to execute a real estate certificate identification and entry method based on an adaptive algorithm.

[0097] It is understandable that the above method of this embodiment can be executed by a single device, such as a computer or server, etc., and can also be applied to a distributed scenario and completed by multiple devices in cooperation with each other. In the case of a distributed scenario, one of the multiple devices can only execute one or more steps in the above method of this embodiment, and multiple devices interact to complete the above method. The processor can be implemented in the form of a general-purpose CPU, a microprocessor, an application-specific integrated circuit, or one or more integrated circuits, etc., for executing related programs to implement the above method of this embodiment. The memory can be implemented in the form of a read-only memory ROM, a random access memory RAM, a static storage device, and a dynamic storage device. The memory can store an operating system and other applications. When the above method of this embodiment is implemented by software or firmware, the relevant program code is stored in the memory and called and executed by the processor.

[0098] This embodiment also provides a computer-readable storage medium storing a computer program, which, when executed, implements a method for identifying and entering a real estate title certificate based on an adaptive algorithm.

[0099] Those skilled in the art should understand that the above-mentioned embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-readable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The present invention 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 invention. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes 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 functions in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 These computer program instructions can also be stored in a computer-readable memory that can guide a computer or other programmable data processing device to work in a specific way, so that the instructions stored in the computer-readable memory produce a product including an instruction device, which implements the functions specified in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to produce a computer-implemented process, so that the instructions executed on the computer or other programmable device provide for implementing the process in the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The above are only preferred embodiments of the present invention and are not intended to limit the present invention in any form. Although the present invention has been disclosed as above with preferred embodiments, it is not intended to limit the present invention. Therefore, any simple modification, equivalent change and modification made to the above embodiments according to the technical essence of the present invention without departing from the content of the technical solution of the present invention shall fall within the scope of protection of the technical solution of the present invention.

[0100] The above is only a preferred embodiment of the present invention, and does not limit the present invention in any form. Although the present invention has been disclosed as a preferred embodiment, it is not intended to limit the present invention. Therefore, any simple modification, equivalent change and modification made to the above embodiment according to the technical essence of the present invention without departing from the content of the technical solution of the present invention shall fall within the scope of protection of the technical solution of the present invention.

Claims

1. A method for identifying and entering a real estate property certificate based on an adaptive algorithm, characterized in that: The following steps are involved: Step S01: obtaining images of real estate certificates in different states to construct a training set and a test set; Step S02: Adaptive algorithm recognition model construction: Build an adaptive scene discrimination model to determine whether the clarity of the test set images meets the requirements; Constructing a nonlinear activation free network NAFNet, and when the adaptive scene discrimination model determines that the clarity of the real estate certificate image does not meet the requirements, using the nonlinear activation free network NAFNet to deblur the real estate certificate image that does not meet the requirements, so as to obtain the real estate certificate image whose clarity meets the preset conditions; Construct an optical character recognition module to perform text recognition on images of real estate certificates whose clarity meets preset conditions; Step S03: Extraction, encryption and entry of key information: Extract key information from the recognized text, encrypt the extracted key information, and automatically enter the encrypted key information into a structured database to achieve the management and storage of real estate certificate data; The method for judging the image clarity of a real estate certificate using the adaptive scene discrimination model includes: Step S201: inputting the real estate certificate image into a 5×1 convolution layer, a 1×5 convolution layer, and a 2×2 average pooling layer in sequence to implement the first downsampling process of the real estate certificate image; Step S202: inputting the first downsampling processing result into a 3×3 convolutional layer and multiple 3×3 dilated convolutional layers with different dilation rates, and performing a first feature fusion on multiple feature maps output by the 3×3 convolutional layer and the 3×3 dilated convolutional layer; Step S203: the output obtained by inputting the first feature fusion result into the asymmetric convolution layer is added element by element for the first time with the output result of the 3×3 convolution layer before the first feature fusion, and the first element-by-element addition result is input into the 2×2 average pooling layer for the second downsampling; Step S204: input the second downsampling result into a 3×3 convolutional layer and multiple 3×3 hole convolutional layers with different expansion rates to obtain multiple feature maps, and perform a second feature fusion on the multiple feature maps; Step S205: the output obtained by inputting the second feature fusion result into the asymmetric convolution layer and the output of the 3×3 convolution layer before the second feature fusion are added element by element for the second time; Step S206: After the second element-by-element addition result is input into the 3×3 convolution layer, the global average pooling layer and the fully connected layer in sequence, the clarity category of the real estate certificate image is obtained through the Softmax function, and the clarity category of the real estate certificate image includes clear and blurred.

2. The method for identifying and entering real estate ownership certificates based on an adaptive algorithm according to claim 1 is characterized in that: In step S02, the adaptive scene discrimination model includes multiple asymmetric convolutions, multiple hole convolutions with different receptive fields, a feature fusion mechanism, and an element-by-element addition mechanism. The asymmetric convolution means that the length and width of the convolution kernel are different.

3. The method for identifying and entering real estate ownership certificates based on an adaptive algorithm according to claim 1, characterized in that: The step S02 also includes adjusting the contrast of the real estate certificate image adjusted by using the nonlinear activation free network NAFNet to enhance the readability of the real estate certificate image.

4. The method for identifying and entering real estate ownership certificates based on an adaptive algorithm according to claim 3 is characterized in that: When the contrast of the real estate certificate image adjusted by the nonlinear activation free network NAFNet is adjusted, the value of each pixel in the image is processed by using the following formula to enhance the contrast of the real estate certificate image: ， in, is the pixel value in the output image, is the pixel value in the original image, is the adjustment factor, is the average pixel value of the original image.

5. The method for identifying and entering real estate ownership certificates based on an adaptive algorithm according to claim 1 is characterized in that: In step S02, the method of using the optical character recognition module to perform text recognition on the image of the real estate certificate whose clarity meets the preset conditions includes: Step S211: using an optical character recognition module to detect the image to obtain a text detection frame; Step S212: After correcting the text detection frame, recognize the text in the text detection frame.

6. The method for identifying and entering real estate ownership certificates based on an adaptive algorithm according to claim 5 is characterized in that: The extraction of key information from the recognized text includes: using a general information extraction unified framework to read a predefined real estate certificate keyword list, and performing targeted information screening and extraction on the text in the recognized text detection box to achieve accurate matching and extraction of key information.

7. The method for identifying and entering real estate ownership certificates based on an adaptive algorithm according to claim 1, characterized in that: In step S03, the encryption process is to start the encryption program before the data enters the database layer, use the symmetric encryption algorithm to quickly encrypt the data, use the asymmetric encryption algorithm to generate a public key and a private key through a random number generator, store the private key in the hardware security module, and use the public key to encrypt the symmetric encryption key. The encrypted data and the encrypted symmetric key are associated and stored or transmitted, thereby realizing encryption of the data.

8. A computer device comprising a processor and a memory, wherein the memory is used to store a computer program, wherein: The processor is configured to execute the computer program to perform the method according to any one of claims 1 to 7.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed, the method according to any one of claims 1 to 7 is implemented.

Images