Computer-implemented method and computer programme for the recording, validation and traceability of digital visual content

Abstract

The invention relates to a computer-implemented method and computer programme for the recording, validation and traceability of digital visual content. The invention involves performing a perceptual analysis associated with the functional workflow for content acquisition, linking an identity prior to creating an independent operational file. It integrates a unique hash generation module (2) based on the Discrete Cosine Transform (DCT) to generate a representation that is robust against technical transformations; an immunological validation module (3) for detecting structural alterations by means of machine learning; and a blockchain recording module (4) to ensure immutability and provenance. In preferred embodiments, it includes an intelligent compression module (5) based on neural networks and a decentralised deletion module (7) managed by consensus and smart contracts.

Description

[0001] DESCRIPTION

[0002] Method implemented on a computer and software program for the registration, validation and traceability of digital visual content.

[0003] Technology sector

[0004] The technical sector in which the invention is framed is that of the software, computing and calculation industry; more specifically, the invention falls within the fields of cybersecurity and blockchain and artificial intelligence technologies applied to the management of digital files.

[0005] Its purpose is to provide a robust method for registering, validating, and protecting visual content by generating a perceptual identity associated with the content acquisition or capture process itself. The invention enables the authentication and traceability of images, ensuring they can be verified immutably without relying on a centralized entity, thus guaranteeing the integrity of the file from its technical origin, even before it adopts an independent operational form or is capable of autonomous storage.

[0006] Its application covers multiple sectors, such as digital security, copyright protection, social media content verification, and forensic image control, providing a comprehensive solution for the protection of digital files in highly vulnerable environments.

[0007] Background of the invention.

[0008] Currently, the protection of digital images faces multiple challenges in terms of security, integrity, and authenticity. The evolution of digital technologies, and in particular the rise of generative artificial intelligence, has facilitated the proliferation of manipulated images, illegal copies, and the creation of synthetic content that is difficult to distinguish from real captures. This seriously undermines trust in digital media and affects critical sectors such as intellectual property, journalism, legal certainty, and forensic evidence. Conventional methods exist for verifying file integrity based on traditional cryptographic hash functions (e.g., the SHA or MD5 families). However, these methods have a fundamental technical limitation: their extreme sensitivity.Any minimal modification to the data at the bit level—even those that don't alter the visual perception of the content, such as slight compression, a change in metadata, or a format re-encoding—results in a completely different hash value. This makes the system unable to recognize the image's identity if it has undergone minor technical transformations, which are common in digital distribution.

[0009] On the other hand, solutions based on watermarking or the use of standard metadata (such as EXIF) have known vulnerabilities, since they can be easily removed, overwritten or corrupted during image processing, thus losing traceability of the origin.

[0010] A critical problem not satisfactorily addressed by current technology is the so-called "initial vulnerability window." The vast majority of existing solutions—including blockchain logs and timestamping systems—operate on files that have already been generated, processed, and stored autonomously by the operating system of the capture device. Therefore, there is a temporal and operational gap between the physical acquisition of the content by the sensor and the creation of the operational file, during which the content is susceptible to interception or manipulation without leaving any trace of the alteration.

[0011] Furthermore, although validation systems using neural networks or distributed consensus mechanisms have been explored, these usually require high computational capacity or depend on central entities acting as certification authorities, which introduces single points of failure and scalability problems.

[0012] The present invention comprehensively resolves these limitations by combining a robust perceptual identity analysis with an immutable record. The novelty lies in the fact that the generation of this identity does not occur retrospectively, but is inextricably linked to the functional flow of content acquisition itself. By generating a compact numerical representation resistant to technical transformations before the file adopts an independently accessible or storable form, the integrity of the content is guaranteed "from birth," allowing for subsequent decentralized and precise validation against malicious alterations. Explanation of the invention.

[0013] Based on the prior art, an objective of the present invention is to provide a computer-implemented method for recording, validating, and protecting images, having the features of claim 1.

[0014] The present invention describes a method implemented on a computer and software program for the registration, validation, and traceability of digital visual content. The system is based on the generation of a unique and robust identity, its validation through structural recognition mechanisms, and its registration in a distributed and immutable database (blockchain).

[0015] 1. Identity Generation in the Acquisition Flow. As a fundamental technical characteristic, the method begins the moment a visual capture device acquires the content. The system executes a perceptual analysis process associated with the content generation flow itself. In this way, the perceptual identity is linked to the content before it adopts an independent, persistent, or autonomously stored operational form in the device's file system.

[0016] This early integration ensures that any subsequent manipulation, even before the user "saves" the image, can be detected, eliminating the window of vulnerability of conventional systems.

[0017] 2. Unique Hash Register Module (Perceptual Identity). Once the acquisition flow is initiated, the register module generates a compact numerical representation (perceptual hash) by applying mathematical transforms and computational models. Unlike traditional cryptographic hashes, these characteristics are selected for their stability against common technical transformations, such as:

[0018] - Format compression and recoding.

[0019] - Changes in resolution and scale.

[0020] - Photometric adjustments (brightness, contrast).

[0021] - Recapture processes. The resulting representation constitutes a robust perceptual identity that allows subsequent verifications regardless of minor technical modifications that the file may have undergone.

[0022] 3. Immunological Validation Module. This identity is subjected to a validation module inspired by artificial immune systems. The system implements recognition mechanisms to detect alterations in the image structure. Using machine learning algorithms, the module distinguishes between "benign" transformations (such as resizing for social media) and "malignant" alterations (such as the insertion or deletion of objects using generative AI). The system generates an "immunological signature" that accompanies the image identifier.

[0023] 4. Blockchain Registration and Cryptographic Association. The generated and validated perceptual identity is associated with a verifiable cryptographic identifier. This identifier is registered on a blockchain network, creating an immutable timestamp and integrity record. The registration includes:

[0024] - The perceptual hash of the image.

[0025] - The original technical metadata.

[0026] - The immunological validation signature.

[0027] Because it is distributed, the system allows any interested third party to verify the authenticity of an image by comparing its current state with the original record on the blockchain, without needing to consult a central authority.

[0028] 5. Additional Optimization and Management Modules. In preferred embodiments, the invention incorporates:

[0029] - An Intelligent Compression Module: Uses neural networks to reduce file size while maintaining intact the structural characteristics that make up its perceptual identity.

[0030] - A Decentralized Deletion Module: Enables the management of sensitive or illicit content. Upon receiving a deletion request validated by consensus on the network, the system records the content's removal on the blockchain, ensuring traceability even during the deletion process.

[0031] This comprehensive approach not only ensures the authenticity of the content but also guarantees that its protection begins at the very moment of its technical creation, providing a level of security unavailable in current state-of-the-art solutions. To complement the description of the method of the present invention, a more detailed explanation of its various phases follows, along with a software tool developed according to said method to enable the authentication and traceability of images, ensuring that they can be verified immutably without relying on a centralized entity. To facilitate understanding of the invention's features, a set of drawings is included with this specification, illustrating, but not limiting, the following:

[0032] Figure 1 represents a diagram of the computer application architecture, showing the interconnections between the different modules that make it up and the data flows between them.

[0033] Figure 2 shows the workflow for validating and registering an image on the blockchain.

[0034] Figure 3 represents a diagram of the integrity verification mechanisms in the blockchain.

[0035] Realization of the invention

[0036] Based on the prior art, an objective of the present invention is to provide a computer-implemented method and software program for the registration, validation, and traceability of digital visual content, having the features of claim 1. The system is based on generating a unique and robust identity, validating it through structural recognition mechanisms, and registering it in a distributed and immutable database (blockchain).

[0037] 1. Integration of identity into the technical acquisition flow

[0038] As a key technical feature, the method begins the moment a visual capture device acquires the content. Unlike prior art systems that operate on finished files, the present invention performs a perceptual analysis process associated with the content generation workflow itself.

[0039] In this way, the perceptual identity is linked to the content before it takes on an independent, persistent, or autonomously stored form in the file system. This "acquisition-time" execution ensures that the structural identifier is generated before any possibility of interception or manipulation of the digital file at the operating system layers, guaranteeing integrity from the absolute technical origin.

[0040] 2. Unique Hash Registration Module (Perceptual Identity)

[0041] The system comprises a unique hash register module (2) responsible for converting the captured data into a compact, tamper-resistant numerical representation. This process is not based on a conventional binary cryptographic hash, but rather on the extraction of structural features from the image.

[0042] In a preferred embodiment, the method uses mathematical transforms, specifically the Discrete Cosine Transform (DCT), to decompose the image into its spatial frequencies. The system selects the low- and mid-frequency coefficients to generate a feature vector that defines the "visual essence" of the object. These features are selected for their stability under common technical transformations, including, but not limited to:

[0043] - Compression and format recoding (e.g., converting from RAW to JPEG).

[0044] - Changes in resolution, scale, and aspect ratio.

[0045] - Photometric adjustments such as brightness, contrast, and gamma correction.

[0046] - Processes of recapturing or digitizing physical media.

[0047] The resulting representation constitutes a compact perceptual identity, robust enough to allow subsequent verifications regardless of the exact moment when the file generation process is completed.

[0048] 3. Immunological Validation and Structural Signature Module

[0049] The generated identity is subjected to an immunological validation module (3) that implements recognition mechanisms inspired by biological systems to detect alterations. This module works by analyzing the structural coherence of the image against a pre-trained machine learning model.

[0050] The system generates an "immunological signature" that accompanies the identifier. This signature not only confirms authenticity but is also capable of distinguishing between "benign" transformations (necessary for the file's functionality) and "malignant" alterations (semantic modifications, insertion of elements using AI, etc.). Validation is performed by comparing the current structure with the recorded integrity patterns, enabling forensic traceability of the content.

[0051] 4. Cryptographic Association and Blockchain Registry

[0052] Perceptual identity and its associated signature are linked to a verifiable cryptographic identifier. The blockchain ledger module (4) manages the issuance of a transaction on a distributed and immutable network. This ledger acts as a timestamp and integrity seal that is independent of any central authority.

[0053] The blockchain record allows any node on the network to verify the correspondence between a circulating image and its original record. Because the identity was generated during the acquisition process (point 1), the blockchain record conclusively proves that the visual content existed in its original form from the moment of its technical capture, closing the vulnerability described in the background section.

[0054] 5. Intelligent Compression Module using Neural Networks

[0055] To optimize content management without compromising validation, the invention includes an intelligent compression module (5). This module uses a neural compression process based on autoencoder architectures or convolutional networks, which allows reducing the file size while maintaining structural control points intact.

[0056] The system identifies the regions of greatest perceptual importance and applies a differentiated level of compression, ensuring that the perceptual identity generated in step (2) remains unaltered and validatable even in low bandwidth or limited storage environments.

[0057] 6. Decentralized Disposal and Content Management Module

[0058] Finally, the invention incorporates a decentralized removal module (7). Upon detection of sensitive or illicit content, or at the request of the rights holder, the system manages a reporting and consensus validation process.

[0059] If the request is approved by the network nodes, the module generates a transaction on the blockchain that records the deletion decision and revokes the validity of the associated identifier. This mechanism ensures that, even if the image physically persists in some local storage, its "validated content" status is permanently and publicly annulled in the distributed ledger. The software of the present invention is structured in a modular architecture where each component interacts in a coordinated manner to guarantee the traceability of the content from its origin.

[0060] 1. Image Acquisition and Preprocessing Module (1). This module does not act as a simple file receiver, but is integrated into the device's data pipeline. Its main function is to intercept the sensor signals or raw data streams. During this phase, the program executes normalization algorithms, including conversion to a uniform color space (grayscale) to eliminate non-essential chromatic variations.

[0061] It also applies smoothing filters (such as Gaussian and median filters) specifically designed to reduce thermal noise from the sensor without degrading critical edges. This phase is crucial because it ensures that the "raw material" on which the identity will be generated is consistent, regardless of the hardware used for capture.

[0062] 2. Unique Hash Generation Module (2). Once the signal has been preprocessed, this module implements the mathematical logic for creating the fingerprint. The program uses an implementation of the Discrete Cosine Transform (DCT) that divides the image into 8x8 pixel blocks. The algorithm is configured to extract the low-frequency coefficients (specifically the first 64), which contain the most relevant structural and energetic information.

[0063] Unlike a binary hash, the program calculates the average of these coefficients and generates a bit vector based on the relationship of each coefficient to that average. This process guarantees that, if the image is compressed or its resolution changes, the resulting hash will be identical or maintain a minimum Hamming distance, allowing for perceptual identification.

[0064] 3. Immunological Validation Module (3). This module functions as an advanced forensic analysis engine. It uses machine learning libraries to compare image structure with bio-inspired patterns. The program decomposes the image using wavelet transforms to analyze irregularities at different spatial scales.

[0065] The code is designed to generate a matrix of structural descriptors (HOG and SIFT) that encapsulate the geometry of the objects present. If the program detects that an area of ​​the image has been altered using inpainting techniques or synthetic generation, the immunological signature will show a mathematical discrepancy with respect to the original record, issuing an automatic diagnosis of "manipulation."

[0066] 4. Blockchain Registration Module (4). This component is responsible for communication with the distributed network. The program takes the generated hash and the immunological signature, packages them into a data structure compatible with Merkle trees, and manages the cryptographic signature using the device's or user's private key. The program includes the necessary logic to monitor transaction confirmation on the network, ensuring that the "seal of origin" is indelibly recorded.

[0067] Support Modules (API Interface, Compression and Deletion):

[0068] - API Interface (6): Facilitates integration with external services, allowing third-party applications to check the authenticity of a file using REST / JSON calls.

[0069] - Intelligent Compression (5): Implements convolutional neural networks (CNNs) that act as an autoencoder. The program learns to discard redundant data while maintaining the structural control points necessary for Module (2) to continue recognizing the image.

[0070] - Decentralized Deletion (7): Manages the consensus protocol. In case of denunciation, the program executes a smart contract that issues temporary cryptographic keys (E-Keys). These keys work by invalidating the ability of network nodes to serve or validate the original content, effectively effecting a "logical deletion" across the entire distributed system.

Claims

CLAIMS 1. A method implemented on a computer for the registration, validation, and traceability of digital visual content, comprising the following phases: a) generate a perceptual identity linked to the visual content during its functional acquisition flow, so that said perceptual identity is obtained prior to the generation of an operational and independent digital file; b) associating said perceptual identity to an immutable record in a distributed network, which includes integrity information and a timestamp; c) verify the authenticity of visual content by comparing its current state with the identity registered on the distributed network; and d) manage the availability or access to the content by executing programmed instructions on the distributed network (smart contracts) based on the result of the verification or integrity notifications.

2. Method, according to claim 1, wherein the generation of the perceptual identity of phase (a) comprises the use of the Discrete Cosine Transform (DCT) to convert the visual content into a numerical representation or hash.

3. Method, according to claim 2, wherein the application of the DCT comprises: compute frequency coefficients to capture high and low frequency patterns; select the first 64 low-frequency coefficients as a representation of the overall structure; Generate a binary signature by comparing the coefficients with the mean of the values ​​obtained.

4. Method, according to claim 1, wherein phase (b) comprises generating a hybrid perceptual signature that combines perceptual identity with a machine learning model trained to detect structural alterations.

5. Method according to claim 1, wherein the immutable record of phase (b) uses a Merkle tree to link successive versions of the content and associates identity to a verifiable decentralized identifier (DID).

6. Method, according to claim 1, wherein the availability management phase (d) comprises the execution of smart contracts that issue temporary cryptographic keys (E-Keys) to invalidate access or redistribution in case of tampering detection.

7. Method, according to claim 1, further comprising an intelligent compression phase using convolutional neural networks and autoencoders prior to registration in the distributed network.

8. Method, according to claim 7, wherein the compression comprises identifying regions of perceptual redundancy and applying adaptive quantification according to the structural importance of each region.

9. Method, according to claim 1, further comprising a preprocessing phase during acquisition consisting of conversion to grayscale, luminance normalization and noise filtering (median or Gaussian filters).

10. Method, according to claims 1 and 4, wherein the integrity verification comprises extracting structural and textural features, applying edge detection and calculating the similarity distance (Euclidean or Cosine) between the current and registered signature.

11. Method, according to claim 1, wherein the registration on the distributed network comprises converting the hash to hexadecimal format, signing the transaction with a private key and sending it to a public or permissioned blockchain.

12. Method, according to claim 1, comprising a decentralized removal process triggered by user complaints, validated by a consensus mechanism and recorded as a revocation transaction on the distributed network.

13. Computer program for the registration, validation and traceability of digital visual content, comprising: - a technical acquisition module configured to generate a perceptual identity of the visual content during its generation flow, prior to the creation of a standalone operational file; - a cryptographic link module to immutably register said identity on a distributed network; - a verification module to compare current content with historical records on the distributed network; and - a governance module to execute programmed instructions that control access to content based on its integrity status. 14.- Computer program, according to claim 13, wherein the acquisition module integrates the Discrete Cosine Transform (DCT) for the extraction of structural features stable against compressions, resolution changes or photometric adjustments. 15.- Computer program, according to claim 13, which includes an immunological validation module that employs machine learning models and frequency transforms (Fourier / Wavelet) to generate structural signatures based on HOG or SIFT descriptors. 16.- Computer program, according to claim 13, which includes a neural compression module configured as an autoencoder to represent the images in a lower dimension latent space preserving perceptual identity. 17.- Computer program, according to claim 13, comprising an API Interface and a Verification Platform to provide decentralized access to traceability functionality through web services. 18.- Computer program, according to claim 13, wherein the governance module manages a decentralized elimination protocol by means of consensus voting and issuance of abrogation transactions on the distributed network.

Images