Hierarchical encryption method for oblique photography three-dimensional model data

Abstract

The invention relates to the field of three-dimensional geographic space data security, and discloses a hierarchical encryption method for oblique photography three-dimensional model data to solve the problems of low encryption efficiency, weak encryption security, functional limitation, data precision loss and the like existing in an existing oblique photography three-dimensional model encryption method. In combination with dynamic key binding, homomorphic encryption and parallel processing technologies, efficient security protection of the oblique photography three-dimensional model is realized, and the method comprises the following specific steps: converting an original oblique photography three-dimensional model into standard 3D Tiles format data, generating a tile set containing tile files and JSON metadata, and storing the tile set in a database; dividing a high-density region and a low-density region according to the geometric density threshold and the LOD hierarchy; and generating a unique key seed through a bijection hash function based on the center coordinates of the tile bounding box, storing the key seed and encryption parameters in a distributed key management system in a fragmented manner, and periodically updating the key by adopting a dynamic rotation strategy.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of three-dimensional geographic space data security, in particular to a tilt photography three-dimensional model data hierarchical encryption method. BACKGROUND

[0002] As a kind of live-action three-dimensional modeling technology, tilt photography three-dimensional model is widely used in geographic information and surveying and mapping field, and shows important value in smart city, land management, cultural heritage protection and architectural design. With the popularity of unmanned aerial vehicle technology and cloud computing, the tilt photography modeling mode of "high precision, low cost and high efficiency" is becoming the core support of city digital development. Therefore, efficient processing and security protection of three-dimensional model data is the key to realize live-action three-dimensional global coverage and intelligent application.

[0003] In the prior art, the encryption scheme of tilt photography model depends on traditional symmetric encryption (such as AES) or simple vertex position confusion technology. Although this method can realize basic data protection, it has the following defects: the encryption process is easy to cause slight distortion of key geometric information such as model vertex coordinates and face index, especially in dense areas (such as building facade and complex terrain), slight deviation may cause structure misplacement or hollow problem during three-dimensional model rendering; and the encrypted data cannot directly support spatial calculation (such as view frustum clipping and LOD dynamic loading), and needs to be completely decrypted before operation, which increases the risk of data exposure. SUMMARY

[0004] The present application aims to solve the problems of low encryption efficiency, weak encryption security, functional limitation and data precision loss in the existing tilt photography three-dimensional model encryption method, and provides a tilt photography three-dimensional model data hierarchical encryption method, which realizes efficient and secure protection of tilt photography three-dimensional model by combining dynamic key binding, homomorphic encryption and parallel processing technology through hierarchical encryption strategy.

[0005] The present application is realized by the following technical scheme: a tilt photography three-dimensional model data hierarchical encryption method, comprising the following specific steps: 1) converting the original tilt photography three-dimensional model into standard 3D Tiles format data, generating a tile set containing tile files and JSON metadata, and dividing high-density areas and low-density areas according to geometric density threshold and LOD level; 2) generating a unique key seed based on tile bounding box center coordinates through bijective hash function, and storing the key seed and encryption parameters in distributed key management system (KMS) by sharding, and periodically updating the key by using dynamic rotation strategy; 3) Perform vertex coordinate nonlinear perturbation, model topology obfuscation, and CKKS homomorphic encryption perturbation on the tile files of high-density regions in sequence. The model topology obfuscation process is as follows: generate an obfuscation rule mapping table by exchanging the vertex connection relationship of adjacent faces according to the parity of the face index, and encrypt the obfuscation rule mapping table with a key seed and embed a custom field (ConfusionMap) of JSON metadata. The process for perturbing data using CKKS homomorphic encryption includes the following steps: First, the vertex coordinate vector after confusing the nonlinear perturbation of vertex coordinates with the model topology. m The encoding is in polynomial form; Then, the CKKS algorithm is used to generate ciphertext pairs: c 0 = Δ m ( X )+ a ( X )• s ( X )+ e 0( X ); c 1=- a ( X )+ e 1( X ); Where Δ is the amplification factor, s ( X The private key is obtained from KMS. a ( X () is a random polynomial. e 0( X )and e 1( X The noise term (which follows a Gaussian distribution, i.e., standard deviation) is... σ =3.2).

[0006] That is, based on the encryption strength factor, a multi-stage encryption method of "vertex coordinate nonlinear perturbation - model topology obfuscation - perturbation data CKKS homomorphic encryption" is used for tile files in high-density areas to enhance security; For tile files in low-density areas, AES-GCM mode encryption is used: a 256-bit key derived from the key seed and an independent initialization vector IV are obtained from KMS. The 256-bit key and the independent initialization vector IV are used to perform AES symmetric encryption on the tile file of the low-density area to be encrypted, generating the encrypted tile file of the low-density area.

[0007] 4) Perform AES-RSA double encryption on the JSON metadata, construct a dynamic Merkle tree based on the BLAKE3 algorithm to verify the integrity of the tile file and JSON metadata, and add a digital signature to the encrypted JSON metadata file; 5) Utilize GPU parallel computing and Web Worker technology to achieve efficient decryption of tile files, and authorize users to access complete data through multi-factor authentication; when authorizing users to access complete data through multi-factor authentication, control access permissions based on whether the user is an authorized user; 6) Encapsulate the decrypted data into glTF format and package it into a .b3dm file to generate a tileset.json file describing the hierarchical structure.

[0008] To further improve the implementation of the oblique photogrammetry 3D model data hierarchical encryption method of the present invention, the following setting method is specifically adopted: In step 2), the key seed is generated by the following formula: Seed=Keccak-512 x 3 + y 5 + z 7 mod2 64 ;in,( x , y , z ) represents the center coordinates of the tile bounding box, and the key is stored in fragments on at least 3 distributed nodes.

[0009] To further improve the implementation of the oblique photogrammetry 3D model data hierarchical encryption method described in this invention, the following setting is specifically adopted: In step 3), when performing nonlinear perturbation of vertex coordinates, dynamic parameters bound to spatial position are used, and this is achieved through the following formula: ;in, v i Let i be the spatial coordinate vector of the i-th vertex in the original 3D model. Let i be the new spatial coordinate vector of the i-th vertex after nonlinear perturbation of the vertex coordinates. β It is a dynamic parameter bound to spatial location, also known as a nonlinear perturbation parameter, and , k For scene coefficients; It is also a dynamic parameter bound to spatial location, and is a random phase parameter generated by a secure random number generator.

[0010] To further improve the implementation of the oblique photogrammetry three-dimensional model data hierarchical encryption method of the present invention, the following settings are specifically adopted: Step 4) specifically involves: using the AES symmetric encryption algorithm to encrypt sensitive fields and protecting the AES key with RSA asymmetric encryption; then digitally signing the encrypted JSON metadata file and constructing a Merkle tree to record the hash value of each tile.

[0011] Compared with the prior art, the present invention has the following advantages and beneficial effects: This invention is based on homomorphic encryption: high-density areas use CKKS homomorphic encryption, which supports spatial computation on encrypted data.

[0012] This invention achieves dynamic key binding: a unique key seed is generated based on spatial coordinates to ensure a strong correlation between the key and the tile position.

[0013] This invention features a layered security strategy: low-density areas employ lightweight AES-GCM encryption, balancing security and performance.

[0014] This invention provides computational security: homomorphic encryption ensures that data in high-density areas can still participate in computation even when encrypted.

[0015] This invention is resistant to reverse attacks: it combines nonlinear perturbation of vertex coordinates with homomorphic encryption, providing dual protection to increase the difficulty of data cracking.

[0016] This invention is standardized and compatible: the encrypted 3DTiles data is fully compatible with rendering engines such as Cesium. Attached Figure Description

[0017] Figure 1 This is the overall flowchart of the present invention.

[0018] Figure 2 This is a schematic diagram of the hierarchical encryption process of the present invention.

[0019] Figure 3 This is a schematic diagram of the decryption process of the present invention. Detailed Implementation

[0020] The present invention will be further described in detail below with reference to embodiments, but the implementation of the present invention is not limited thereto.

[0021] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to represent selected embodiments of the invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0023] Example 1: This invention designs a hierarchical encryption method for oblique photogrammetry 3D model data. Addressing the problems of low encryption efficiency, weak encryption security, functional limitations, and data accuracy loss in existing oblique photogrammetry 3D model encryption methods, this invention achieves efficient and secure protection of oblique photogrammetry 3D models through a hierarchical encryption strategy combined with dynamic key binding, homomorphic encryption, and parallel processing techniques. The specific steps include the following: 1) Convert the original oblique photogrammetry 3D model into standard 3D Tiles format data, generate a tile set containing tile files and JSON metadata, and divide high-density and low-density areas according to geometric density thresholds and LOD levels; 2) Based on the center coordinates of the tile bounding box, a unique key seed is generated using a bijective hash function, and the key seed and encryption parameters are stored in a distributed key management system (KMS) in fragments. At the same time, a dynamic rotation strategy is used to periodically update the key. 3) Perform vertex coordinate nonlinear perturbation, model topology obfuscation, and CKKS homomorphic encryption perturbation on the tile files of high-density regions in sequence; Specifically, when performing nonlinear perturbation of vertex coordinates, dynamic parameters bound to spatial position are used, and this is achieved through the following formula: ;in, v i Let i be the spatial coordinate vector of the i-th vertex in the original 3D model. Let i be the new spatial coordinate vector of the i-th vertex after nonlinear perturbation of the vertex coordinates. β It is a dynamic parameter bound to spatial location, and , k For scene coefficients; It is also a dynamic parameter bound to spatial location, and is a random phase parameter generated by a secure random number generator; The process of model topology confusion is as follows: the vertex connection relationship of adjacent faces is swapped according to the parity of the face index to generate a confusion rule mapping table, and the confusion rule mapping table is encrypted with a key seed and then embedded into a custom field (ConfusionMap) of JSON metadata. The process for perturbing data using CKKS homomorphic encryption includes the following steps: First, the vertex coordinate vector after confusing the nonlinear perturbation of vertex coordinates with the model topology. m The encoding is in polynomial form; Then, the CKKS algorithm is used to generate ciphertext pairs: c 0 = Δ m ( X )+ a ( X )• s ( X )+ e 0( X ); c 1=- a ( X )+ e 1( X ); Where Δ is the amplification factor, s ( X The private key is obtained from KMS. a ( X () is a random polynomial. e 0( X )and e 1( X The noise term (which follows a Gaussian distribution, i.e., standard deviation) is... σ =3.2).

[0024] That is, based on the encryption strength factor, a multi-stage encryption method of "vertex coordinate nonlinear perturbation - model topology obfuscation - perturbation data CKKS homomorphic encryption" is used for tile files in high-density areas to enhance security; For tile files in low-density areas, AES-GCM mode is used for encryption: a 256-bit key derived from the key seed and an independent initialization vector IV are obtained from KMS. The 256-bit key and the independent initialization vector IV are used to perform AES symmetric encryption on the tile file of the low-density area to be encrypted, generating the encrypted tile file of the low-density area. 4) Perform AES-RSA double-layer encryption on the JSON metadata, construct a dynamic Merkle tree based on the BLAKE3 algorithm to verify the integrity of the tile file and JSON metadata, and add a digital signature to the encrypted JSON metadata file; that is, the key seed, nonlinear perturbation parameters, etc., in steps 2) and 3) are used. β Random phase parameters Sensitive parameters such as these are protected twice, and a digital signature is added to the encrypted JSON metadata file to achieve metadata security and integrity verification. 5) Utilize GPU parallel computing and Web Worker technology to achieve efficient decryption of tile files, and authorize users to access complete data through multi-factor authentication; when authorizing users to access complete data through multi-factor authentication, control access permissions based on whether the user is an authorized user; 6) Encapsulate the decrypted data into glTF format and package it into a .b3dm file to generate a tileset.json file describing the hierarchical structure.

[0025] Example 2: This embodiment is a further optimization based on the above embodiments. The parts identical to the aforementioned technical solutions will not be repeated here. Furthermore, to better implement the hierarchical encryption method for oblique photogrammetry 3D model data described in this invention, the following setting is specifically adopted: In step 2), the key seed is generated using the following formula: Seed=Keccak-512 x 3 + y 5 + z 7 mod2 64 ;in,( x , y , z ) represents the center coordinates of the tile bounding box, and the key is stored in fragments on at least 3 distributed nodes.

[0026] Example 3: This embodiment is a further optimization based on any of the above embodiments. The parts that are the same as those in the foregoing technical solutions will not be repeated here. In order to better realize the oblique photogrammetry three-dimensional model data hierarchical encryption method of the present invention, the following setting method is adopted: Step 4) is specifically: use the AES symmetric encryption algorithm to encrypt sensitive fields and protect the AES key with RSA asymmetric encryption; then digitally sign the encrypted JSON metadata file and construct a Merkle tree to record the hash value of each tile.

[0027] Example 4: A hierarchical encryption method for oblique photogrammetry 3D model data is proposed. By employing a hierarchical encryption strategy combined with dynamic key binding, homomorphic encryption, and parallel processing techniques, this method achieves efficient and secure protection for oblique photogrammetry 3D models.

[0028] Combination Figure 1 As shown, step (1): data preprocessing and tile grading is the initial stage of the process. First, the original oblique photogrammetry 3D model is converted into a format and divided into layers. Specifically, it includes step (11) data conversion and step (12) layer division: Step (11): Data Conversion: Original Model File → 3D Tiles Format File. The original oblique photogrammetry 3D model (such as OSGB format) is converted to the standard 3D Tiles format using open-source tools (Cesium ion, FME), generating a tile set containing the following files: ① Binary tile files: including `.b3dm` (batch model tiles), `.i3dm` (instantiated model tiles), etc.; ② JSON metadata file: including `tileset.json` which describes the tile hierarchy, bounding box (`boundingVolume`), geometric error (`geometricError`), and spatial index.

[0029] Parse geometric properties: bounding box, vertex density; that is, during the conversion process, this method automatically parses the geometric properties of each tile, including the bounding volume, geometric error, and density information of vertices and faces.

[0030] Step (12) Hierarchical division: High / low density areas: This method divides the tiles into two levels: high-density areas and low-density areas.

[0031] Judgment criteria: When the geometric density of the tile (i.e., the ratio of the number of vertices to the volume of the bounding box) is greater than or equal to the preset threshold T, it is identified as a high-density area; otherwise, it is identified as a low-density area.

[0032] Specifically, by calculating the ratio of the number of vertices of each tile to the volume of the bounding box, areas with high detail (such as building facades and complex terrain) are marked as high-density areas, while areas with low detail (such as open ground and simplified structures) are classified as low-density areas.

[0033] The tile geometric density is calculated by extracting the `boundingVolume` (bounding box type `box` or `region`) and `geometricError` values ​​for each tile, and by counting the number of vertices (`VertexCount`) and faces (`FaceCount`) for each tile. ; Set a geometric density threshold T (high-density areas ≥ 100 vertices / cubic meter, low-density areas < 100 vertices / cubic meter). By combining the LOD level (high LOD corresponds to high detail, and low LOD corresponds to low detail), the tiles are marked as high-density areas or low-density areas.

[0034] This tiered strategy ensures that high-value data receives more stringent protection, laying the foundation for subsequent layered encryption.

[0035] Step (2): Dynamic key generation and secure storage.

[0036] After data classification is completed, a dynamically bound encryption key is generated for each tile. Specifically, this includes: Step (21): Obtain the center coordinates of the bounding box (i.e., based on the center coordinates of the tile bounding box).

[0037] Extract the center coordinates of the tile bounding box from the JSON metadata. For bounding boxes of type `box`, directly read the first three values ​​as the center coordinates. x , y , z ); For `region` type bounding boxes, the center coordinates are derived by calculating the average values ​​of the west, east, south, and north boundaries and combining them with the height range. x , y , z ): .

[0038] Step (22): Generate a key seed (Keccak-512 hash), that is, use a bijective hash function to convert spatial coordinates into a unique key seed.

[0039] Input center coordinates ( x , y , zThe unique key seed is generated using the following formula: Seed=Keccak-512 x 3 + y 5 + z 7 )mod N ; Among them, SHA-3 (Keccak algorithm) is used to ensure that the key seed is strongly bound to the tile position and has high collision resistance, while the modulo operation ( N =2 64 Limiting the size of the key space avoids numerical overflow while reducing the probability of duplicate seeds.

[0040] Step (23): Key Derivation and Storage.

[0041] The encryption strength factor is determined based on the LOD level and geometric density of the tile (1 for high-density areas and 0.5 for low-density areas); the seed is stored in the key management system (KMS), and the operation keys (vertex perturbation parameters, CKKS key, AES-GCM key, RSA key, etc.) are derived in real time from the key seed and are not stored.

[0042] A fragmented storage and dynamic rotation strategy is adopted: the key seed is divided into 5 fragments and stored on distributed KMS nodes. At least 3 fragments are required to reconstruct the complete key, reducing the risk of key seed leakage. A new key seed is automatically generated every 24 hours. The old key seed is retained for decrypting historical data, and the updated key seed is used to encrypt new data, preventing the risk of leakage of long-term key seeds.

[0043] Step (3): Layered encryption processing, specifically including (31) high-density tile encryption processing method (high-density area encryption) and (32) low-density tile encryption processing method (low-density area encryption). Combined with Figure 2 The diagram shows a hierarchical encryption process.

[0044] Step (31): For high-density areas, the method employs multi-stage encryption to enhance security. This process consists of three stages.

[0045] Phase 1 (Vertex Nonlinear Perturbation): First, the vertex coordinates are nonlinearly perturbed, based on the key seed and nonlinear perturbation parameters. β The transformation is performed, and the formula is: ; v i Let i be the spatial coordinate vector of the i-th vertex in the original 3D model. Let be the new coordinate vector of the i-th vertex after nonlinear perturbation of vertex coordinates, and let be the nonlinear perturbation parameters. β Adjust according to the bounding box size and scene type (building / terrain), specifically taking... , k The scene coefficients are set to 0.02 for buildings and 0.005 for terrain to ensure that the disturbance amplitude adapts to different geometric features; The random phase parameters (0~2) generated by the secure random number generator π ), and the key seed are stored together in KMS.

[0046] Phase 2 (Topology Obfuscation (Parity Index Swap + Virtual Patch)): Obfuscate the topology of the model: ① Iterate through the vertex index array of each face. If the face position is odd (e.g., the 1st, 3rd, 5th... face), swap its first and third vertex indices (...). v 1, v 2, v 3) becomes ( v 3, v 2, v 1) to disrupt local connections; ② Insert a virtual face (index 0,0,0) for every 10 faces, and filter it automatically during decryption; ③ The obfuscation rule sequence is encrypted with AES and then embedded in the `encryptionParams` field of the JSON metadata to prevent the rules from being reverse-engineered and to provide a basis for subsequent decryption.

[0047] Phase 3 (CKKS Homomorphic Encryption (Polynomial Ring N=8192)): CKKS homomorphic encryption is applied to the perturbed data. Vertex coordinates are scaled and encoded into integer polynomials, and ciphertext pairs are generated using an encryption algorithm. The specific process includes the following steps: ① Vertex coordinate encoding: Encoding the perturbed vertex coordinate vector m Scale and quantize to integers : And Δ is the amplification factor (taken as 2). 20 ); ② Polynomial mapping: [This refers to a mapping technique where polynomials are mapped to polynomials.] Mapping to the polynomial ring Generating polynomials m ( X The polynomial ring dimension N=8192, supporting higher precision calculations; ③ Encrypt and generate ciphertext pairs: c 0 = Δ m ( X )+a ( X )• s ( X )+ e 0( X ); c 1=- a ( X )+ e 1( X ); Where Δ is the amplification factor, s ( X The private key is obtained from KMS. a ( X () is a random polynomial. e 0( X )and e 1( X The noise term (which follows a Gaussian distribution, i.e., standard deviation) is... σ =3.2).

[0048] ③ Store ciphertext: c 0 and c 1. Write the encrypted data segment to the `.b3dm` tile file.

[0049] The encrypted data supports limited computations (such as spatial range queries) in the encrypted state, meeting the needs of more business scenarios.

[0050] Step (32): For low-density tile encryption, lightweight AES symmetric encryption is used to improve efficiency, including the following process (AES-GCM encryption): 1. Generate symmetric key: Obtain the AES key (256 bits) and initialization vector (IV) from KMS; 2. Encrypted data blocks: The tile binary data is encrypted using AES-GCM mode, which adds data integrity verification. 3. Replace the original data: Write the encrypted data to the `.b3dm` file and record the IV to JSON metadata.

[0051] The encrypted data can directly replace the original file, ensuring that low-density areas reduce computational overhead while maintaining basic security.

[0052] Step (4): Metadata security and integrity verification, namely metadata embedding, digital signature, and sensitive data encryption. After encryption, sensitive parameters in the JSON metadata (such as nonlinear perturbation parameters) are verified. β A secondary protection mechanism is used (using obfuscation rule indexes). First, sensitive fields are encrypted with AES, and then the AES key is protected with RSA asymmetric encryption, forming a double layer of protection.

[0053] Add a digital signature to the JSON metadata file. Add the following field to the JSON metadata file (example):

[0054] The hash value of the entire JSON metadata file (including the `encryptionParams` field) is calculated using SHA-256, and then signed using an RSA private key to ensure that the metadata has not been tampered with during transmission. The signing process is as follows: 1. Calculate the hash value of the JSON file using SHA256; 2. Use the private key to perform an RSA signature on the hash value, write the signature value into the _signature field in the file header, and decrypt the _signature with the public key during verification and compare it with the recalculated hash value; In addition, a Blake3 hash value is generated for each tile file and a Merkle tree is constructed, embedding the root hash into the metadata. Any data modification will cause a root hash mismatch, resulting in Merkle tree verification failure, thus quickly locating anomalous data.

[0055] Step (5): Parallel encryption / decryption and dynamic access control.

[0056] Figure 3 This is a schematic diagram of the decryption process of the present invention, and the specific implementation method is as follows: Step (51): Tile decryption process (user obtains operation key): First, the decryption method for high-density tiles (high-density area recovery): 1. Key Recovery: After the user is authorized, KMS returns the private key. s ( X and nonlinear perturbation parameters; that is, obtaining the private key from KMS. s ( X and nonlinear disturbance parameters; 2. Homomorphic Decryption (CKKS Homomorphic Decryption (Calculation)) ): Calculate the approximate value: ; decoding: ; 3. Inverse perturbation and topology recovery (inverse vertex perturbation (least squares fitting)): according to β and Seed The least squares method is used to fit the vertex coordinates before the perturbation, and the error introduced by noise is reduced through iterative optimization. The formula is as follows: ; Topology restoration (filtering virtual facets + index restoration) means restoring the facet index order according to the obfuscation rules of the metadata records.

[0057] Low-density area recovery → AES-GCM decryption (IV verification + HMAC verification); that is, the decryption method for low-density tiles is AES-GCM decryption, which includes IV verification and HMAC verification.

[0058] Step (52): Dynamic access control: On the user side, efficient encryption and decryption are achieved through GPU parallel computing and Web Worker technology. Homomorphic decryption tasks in high-density regions are divided into multiple subtasks, which are processed in parallel by the GPU to avoid blocking the main thread.

[0059] Regarding access control, by default, unauthorized users can only load preview data for low-density areas (after AES decryption). When a user enters a valid authorization code, the private key and parameters for high-density areas are dynamically obtained from KMS. Homomorphic decryption, reverse perturbation, and topology recovery operations are then performed step by step, ultimately merging all tile data to generate a complete model (i.e., merging high / low-density tile data). All access behaviors are logged, supporting subsequent auditing and traceability.

[0060] Step (6): Generate 3DTiles data files; In the final stage, the decrypted data is repackaged into a file conforming to the 3DTiles standard, i.e., a standard 3DTiles file is generated. The specific method is as follows: Generate glTF model: 1. Construct the glTF model: Convert the decrypted vertex and face data into glTF format; 2. Package the b3dm file: Combine the glTF data with encrypted attributes (such as `_BATCHID`) into a binary `.b3dm` file; 3. Generate Tileset.json: Describes the tile hierarchy and URL path according to the spatial index structure for loading by the Cesium engine. Generate a `tileset.json` file based on the spatial index structure of the tiles, describing the hierarchy and loading path; Loaded for rendering in the Cesium engine: The encrypted dataset can be directly imported into engines such as Cesium for secure rendering and interaction.

[0061] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Any simple modifications or equivalent changes made to the above embodiments based on the technical essence of the present invention shall fall within the protection scope of the present invention.

Claims

1. A hierarchical encryption method for oblique photogrammetry 3D model data, characterized in that: The specific steps include the following: 1) Convert the original oblique photogrammetry 3D model into standard 3D Tiles format data, generate a tile set containing tile files and JSON metadata, and divide high-density and low-density areas according to geometric density thresholds and LOD levels; 2) Based on the center coordinates of the tile bounding box, a unique key seed is generated by a bijective hash function, and the key seed and encryption parameters are stored in a distributed key management system in fragments. At the same time, a dynamic rotation strategy is used to periodically update the key. 3) For tile files in high-density regions, perform vertex coordinate nonlinear perturbation, model topology obfuscation, and CKKS homomorphic encryption perturbation sequentially; for tile files in low-density regions, use AES-GCM encryption mode. 4) Perform AES-RSA double encryption on the JSON metadata and add a digital signature to the encrypted JSON metadata file; 5) Utilize GPU parallel computing and Web Worker technology to achieve efficient decryption of tile files, and authorize users to access complete data through multi-factor authentication; 6) Encapsulate the decrypted data into glTF format and package it into a .b3dm file to generate a tileset.json file describing the hierarchical structure.

2. The method for hierarchical encryption of oblique photogrammetry 3D model data according to claim 1, characterized in that: In step 2), the key seed is generated using the following formula: Seed=Keccak-512 x 3 + y 5 + z 7 mod2 64 ;in,( x , y , z ) represents the center coordinates of the tile bounding box, and the key is stored in fragments on at least 3 distributed nodes.

3. The method for hierarchical encryption of oblique photogrammetry 3D model data according to claim 1, characterized in that: In step 3), when performing nonlinear perturbation of vertex coordinates, dynamic parameters bound to spatial position are used, and this is achieved through the following formula: ;in, v i Let i be the spatial coordinate vector of the i-th vertex in the original 3D model. Let i be the new spatial coordinate vector of the i-th vertex after nonlinear perturbation of the vertex coordinates. β It is a dynamic parameter bound to spatial location, and , k For scene coefficients; It is also a dynamic parameter bound to spatial location, and is a random phase parameter generated by a secure random number generator.

4. The method for hierarchical encryption of oblique photogrammetry 3D model data according to claim 1, characterized in that: The process of confusing the model topology is as follows: the vertex connection relationship of adjacent faces is swapped according to the parity of the face index to generate a confusion rule mapping table, and the confusion rule mapping table is encrypted with a key seed and then embedded with a custom field of JSON metadata.

5. The method for hierarchical encryption of oblique photogrammetry 3D model data according to claim 1, characterized in that: Step 3) involves the following process when perturbing the data using CKKS homomorphic encryption: First, the vertex coordinate vector after confusing the nonlinear perturbation of vertex coordinates with the model topology. m The encoding is in polynomial form; Then, the CKKS algorithm is used to generate ciphertext pairs: c 0=Δ• m ( X )+ a ( X )• s ( X )+ e 0( X ); c 1=- a ( X )+ e 1( X )； Where Δ is the amplification factor, s ( X ) is the private key. a ( X () is a random polynomial. e 0( X )and e 1( X () represents the noise item.

6. The method for hierarchical encryption of oblique photogrammetry 3D model data according to claim 1, characterized in that: Step 4) specifically involves: encrypting sensitive fields using the AES symmetric encryption algorithm and protecting the AES key using RSA asymmetric encryption; then digitally signing the encrypted JSON metadata file and constructing a Merkle tree to record the hash values ​​of each tile.

Images