A system and method for full-vision data matching, right confirmation and tracing

Abstract

This invention discloses a system and method for full-visual data matching, ownership confirmation, and traceability. The system includes: an encoding unit for performing lightweight fusion encoding on visual data to generate a uniquely corresponding lightweight encoded index, with an encoding response time ≤100ms, accuracy ≥99.9%, and index storage ≤0.03% of the original image; a matching driving unit for setting the matching result as the sole prerequisite for triggering all subsequent functions, with a matching response time ≤1 second; and a matching extension processing unit for executing three operation modes—intelligent matching, generation, and storage—based on the matching result, prohibiting the generation of new indexes throughout the process, with intelligent matching latency ≤150ms, generation latency ≤100ms, and computing power consumption ≤0.03GFLOPS. Non-transactional data is automatically destroyed after 7-15 days. The system is forcibly bound to digital RMB for transaction settlement, achieving value distribution between practitioners and consumers through three-level visual processing. This invention solves the technical problems of fragmented ownership and low circulation efficiency of visual data through a lightweight encoded index and matching driving architecture.

Description

Technical Field

[0001] This invention relates to the field of data ownership confirmation and visual processing technology, specifically to a system and method for full visual data matching, ownership confirmation, and traceability. This invention constructs a visual data processing system with matching as its core driving logic through a lightweight coded index-based unique identification mechanism, used for the definition of visual data ownership, value transfer, and personalized intelligent services. Background Technology

[0002] 2.1 Current Status and Fundamental Dilemmas of the Industry

[0003] Visual data is one of the core production factors of the digital economy. However, in existing technologies, the ownership of visual data is vaguely defined, and its cross-platform circulation is restricted, resulting in a lack of value returns for creators. At the same time, consumer behavior data is occupied by platforms without compensation, and a value distribution mechanism is lacking. Existing technological approaches, including traditional blockchain notarization, centralized databases, and mainstream AIGC generation models, have all failed to solve the technical problems of broken ownership of visual data during circulation and the inability to automatically link value distribution with the original creator.

[0004] 2.2 Limitations of Existing Technological Approaches

[0005] Current mainstream technologies focus on improving computing power, expanding model parameters, increasing the number of templates, and optimizing generation effects, aiming to solve the problem of visual content generation efficiency. However, they have the following technical defects: the ownership relationship between the generated content and the original subject is broken during the generation process, resulting in the generation results being unable to be linked to the original creator; editing, adaptation, and other operations are not bound to the original index, forming new ownerless data; individual AI systems are controlled by the platform, and users cannot have independent data sovereignty and value benefits.

[0006] 2.3 Initial R&D Objectives

[0007] In 2018, during the process of managing drawings for engineering projects, the inventors discovered that the root cause of the problem lay in the broken matching relationship between visual data and the creative subject, rather than insufficient drawing efficiency. Based on this, a team was assembled starting in 2019, and after years of research and engineering verification, the technical solution described in this invention was developed. As of the date of this application, the encoding and ownership confirmation of 1.33 million visual data images have been completed, and 120 visual data asset certificates have been listed on the Guangzhou Data Exchange.

[0008] 2.4 Technical Background Verification

[0009] Through a search and analysis of existing technologies, no technical solution has yet been found that possesses the technical characteristics of "encoding equals rights confirmation and full traceability," implements a lightweight encoding index-driven three-level matching system, and completes the engineering verification of compliant listing of visual data assets on data exchanges. This invention achieves integrated encoding and rights confirmation through a self-developed AIUGC large model. The encoding process and ownership metadata are completed natively within the model, making encoding equal to rights confirmation and index equal to identity. Summary of the Invention

[0010] 3.1 Core Logic of the Invention

[0011] This invention belongs to the field of computer system architecture technology, specifically relating to a full-visual data matching, rights confirmation, and traceability system that uses hard-coded constraints at the system's underlying level as a technical means. Unlike existing technologies that treat "matching" as an optional function, this invention sets the matching result as the only legitimate entry point at the system's underlying level. This entry point is enforced by a unified permission verification module through code logic: any visual data that has not been matched and has not output a matching result, regardless of how it attempts to enter the subsequent processing stage (including but not limited to direct calls from internal system modules, imports from external interfaces, data format conversions, identifier re-encoding, indirect input through third-party systems, etc.), will be automatically identified and blocked by the system's underlying level.

[0012] The above-mentioned technical means solve the technical problems of ownerless data flow caused by ownership breakage, system performance degradation caused by index explosion, and value allocation failure to be automatically executed in the existing technology. It also produces technical effects such as coding accuracy ≥99.9%, index storage compression 3000 times, and generation computing power consumption reduced to less than one ten-thousandth of the existing technology.

[0013] In this invention:

[0014] • Matching layer: Used to establish the connection between visual data and requirements;

[0015] • Intelligent Matching Layer: Based on matching, the visual data is adaptively adjusted, and the adjustment results are bound to the original index without generating new indexes;

[0016] • Generation layer: Based on matching, multiple visual data are combined to generate new content. The new content is bound to all original indexes and no new indexes are generated.

[0017] Matching supply and demand is the starting point of the technical architecture of this invention. As a carrier of value, the digital yuan's transactions are automatically executed by the underlying smart contract of the system when matching is triggered, realizing the direct flow of value from consumers to creators without the need for third-party intermediaries to intercept it.

[0018] 3.2 Definition of core terms

[0019] 3.2.1 Full Vision: Refers to the sum of all visual information directly or indirectly perceived by the human eye, any acquisition device, imaging device, AI vision system, and smart terminal. In the technical system of this invention, "full vision" encompasses the entire chain of visual data forms from acquisition, encoding, rights confirmation, matching, intelligent allocation, generation to transaction and profit sharing, using a lightweight encoding index as a unique identifier to complete a closed-loop circulation on the value carrier of digital RMB.

[0020] 3.2.2 Lightweight Fusion Encoding (corresponding to claims 1 and 6): This integrates a unified encoding algorithm chain encompassing multi-dimensional feature extraction, unified color space mapping, keypoint detection, dimensionality reduction, and sparsification. It compresses visual data into high-dimensional sparse vectors and natively incorporates the author's ownership information during the encoding process. Encoding response time ≤ 100ms (single 2K image), encoding accuracy ≥ 99.9%, and generated index storage size ≤ 0.03% of the original image.

[0021] 3.2.3 Lightweight Encoding Index (corresponding to claims 1 and 6): A unique identifier for visual digital assets generated by lightweight fusion encoding, with a storage size ≤ 0.03% of the original image, and possessing core functions such as identity verification, AI indexing, data reuse, and consumption decision-driven processing.

[0022] 3.2.4 Mirror Space (corresponding to claims 1 and 6, it is the application of lightweight coding index in the set dimension): refers to the digital visual space constructed by the technical system of this invention. This space itself is identified by a unique "mirror space index," which is the application of lightweight coding index (claims 1 and 6) in the set dimension, binding the ownership information of the space creator (subject).

[0023] Within this space, every piece of visual data, every individual AI, and every matching and transaction is uniquely identified and authenticated through its corresponding lightweight coded index. Each piece of visual data, each individual AI, and each matching and transaction is simultaneously bound to its individual index and spatial index. The individual index identifies "who owns it," while the spatial index identifies "where it is located," together forming a complete ownership traceability chain.

[0024] The Mirror Space is not a virtual platform, but a collection of digital assets that is sovereign, traceable, and transferable, composed of countless individual indexes, asset indexes, behavior indexes, and a master space index.

[0025] 3.2.5 Individual AI Database (corresponding to claims 4 and 8): This refers to a dedicated database built based on users' historical matching behavior data, using a lightweight coded index as the unique access key. This database stores structured data such as users' operational preferences and transaction records to support accurate matching for personalized services. Non-transactional data storage has a storage period of 7-15 days and is automatically and irreversibly destroyed upon expiration, except in judicial evidence collection scenarios.

[0026] 3.2.6 Visual Data Visualization: The technical parameters set in this invention (encoding response ≤ 100ms, index storage ≤ 0.03% of the original image, generation computing power ≤ 0.03GFLOPS, matching response ≤ 1 second) are not arbitrary settings for commercial selection, nor are they incremental improvements that can be obtained through conventional optimization. Instead, they are based on the comprehensive calculation results of the following three technical hard constraints:

[0027] (1) Limits of human visual perception: The persistence of vision in the human eye is approximately 100-400 milliseconds. If the end-to-end latency exceeds 100ms, users will clearly perceive a "wait" and will be unable to achieve an "instant visual" experience. This invention compresses the core operations of encoding, matching, intelligent matching, and generation to the 100ms level, so that users are unaware of the existence of data processing throughout the entire interaction process. This is the technical upper limit for achieving a "what you see is what you get, and what you get is what you belong to" experience.

[0028] (2) Physical Constraints of Lightweight Terminals: The NPU computing power of next-generation terminals such as smart glasses and AR / VR headsets is typically less than 1 GFLOPS, and the available memory is typically less than 512 MB. Existing AIGC modes consume 300 GFLOPS of computing power and occupy GB-level video memory per generation, making them physically unsuitable for deployment on such terminals. This invention compresses the generation computing power consumption to 0.03 GFLOPS (less than one ten-thousandth of existing technologies) and the index storage to 0.03% of the original image (3000 times compression), enabling the technical solution of this invention to run natively within the physical constraints of lightweight terminals. This is a necessary technical condition for realizing "data visualization" on the terminal side, not an option.

[0029] (3) System performance inflection point constraint: Through engineering verification (actual measurement with 1.33 million visual data images), for every 10% increase in the size of the index database, the retrieval response time increases by approximately 3%, and the storage cost increases by approximately 15%. When the index storage exceeds 0.03% of the original image, the retrieval response time of the million-level index database will exceed 1 second, which cannot meet the requirements of instant visualization. Therefore, a storage compression ratio of ≤0.03% is the maximum compression ratio achievable under the premise of ensuring retrieval performance, and a temporary storage period of 7-15 days is the optimal balance between storage cost and user experience.

[0030] The aforementioned three constraints together constitute a technical necessity—if any parameter deviates from the threshold set by this invention, the technical requirements of "instantaneous visual experience," "lightweight terminal adaptation," and "system performance scalability" cannot be simultaneously met. Therefore, the parameter settings of this invention do not belong to "conventional optimization," but are necessary technical features for achieving specific technical effects.

[0031] 3.3 Three-level visual processing technology system

[0032] 3.3.1 Primary matching (corresponding to the "matching driving unit / step" in claims 1 and 6)

[0033] The essence of primary matching is feature vector retrieval based on lightweight coded indexes and multi-entity source binding. It's not a simple keyword search, but rather performs similarity calculations in a high-dimensional feature space and permanently binds the matching results to the original index simultaneously. The matching similarity threshold is ≥80% (industry-specific fine-tuning: 85% for design industries, 90% for the medical industry), outputting the Top-K most similar lightweight coded indexes within 1 second (K is configurable from 100-500 sets by default), and the matching response time is ≤1 second (for a million-level index database).

[0034] 3.3.2 Intermediate intelligent matching (corresponding to the "first processing mode" in claims 1 and 6)

[0035] The essence of intermediate-level intelligent matching is to perform cross-scene intelligent adaptation and personalized DIY adjustments on existing visual data while preserving the original ownership. All adjustment parameters are permanently bound to the original index as auxiliary metadata, without generating new indexes. The system's underlying hard-coded code prohibits index generation function calls, and the index library is automatically verified after each intelligent matching operation to ensure that no new index records are added. The processing latency is ≤150ms (single 2K image), and the cross-industry scene adaptation success rate is ≥98%.

[0036] 3.3.3 Advanced Generation (corresponding to the "Second Processing Mode" in claims 1 and 6)

[0037] Advanced generation essentially involves creatively combining existing lightweight encoded indexes with established ownership to generate entirely new visual content. All generation parameters are permanently bound to the original indexes as supplementary metadata, without generating new indexes. All input features required for the generation process originate from successful retrieval of multiple original indexes during the matching process, and the generated result itself does not possess independent asset attributes. Generation latency is ≤100ms (for a single 2K image), computational power consumption is ≤0.03GFLOPS (less than one ten-thousandth of existing AIGC models), and 100 sets of solutions can be output within 1 second.

[0038] 3.3.4 The progressive relationship of the three-level processing

[0039] The three-tiered visual processing technology system is not a simple superposition of three independent functions, but an inseparable value creation chain with "original lightweight coded index" as its core DNA: Primary matching brings existing data to life, marking the starting point of value discovery; intermediate intelligent matching allows the activated data to increase its value across scenarios, marking the starting point of value enhancement; advanced generation enables the creative combination of value-added data, marking the starting point of value creation; individual AI sedimentation allows the combined results to be sedimented into an individual's exclusive memory, with the AI ​​itself acting as a dynamic "individual intelligent index," marking the starting point of value reuse; mirror world construction: when countless individual AIs carrying their owners' lightweight coded indexes connect, interact, and co-create in a dedicated mirror space also identified by a "spatial index," a living, sovereign digital world naturally forms—the endpoint of value aggregation. Without matching, there is no intelligent matching; without matching, there is no generation; both are indispensable. Without matching, there is no individual AI. Without indexing, there is no mirror world.

[0040] 3.4 Beneficial Effects

[0041] 3.4.1 Beneficial effects at the technical level (corresponding to the technical features of the third processing mode in claims 4 and 6)

[0042]

[0043] Through a lightweight, integrated coding mechanism, a permanent binding between visual data and the creator is achieved, enabling seamless and automatic completion of the rights confirmation process with near-zero costs. A progressive "matching-driven" processing architecture sets the matching result as the sole prerequisite for all subsequent operations, eliminating the flow and processing of ownerless data from the system's core. A modular design with three processing modes enables the editing, generation, and user data accumulation of visual data, while prohibiting the generation of new indexes throughout the process, ensuring the uniqueness and traceability of ownership. Through engineering verification, 1.33 million visual data entries have been coded and rights confirmed, and 120 asset certificates have been listed on the Guangzhou Data Exchange, with all core technical indicators meeting the standards.

[0044] In this invention, the storage period for non-transactional data is set at 7-15 days. This is not an arbitrary commercial choice, but rather a hard constraint based on three technical factors: First, compliance with privacy protection regulations. Visual data is considered temporary behavioral data before a transaction is confirmed, and storing it beyond the legally stipulated period would constitute a potential infringement on user privacy rights. Second, calculations from a storage cost optimization model show that for every 10% increase in the index size, retrieval response time increases by approximately 3%, and storage costs increase by approximately 15%. The 7-15 day window achieves optimal allocation of storage resources while ensuring user experience. Third, the system's automatic cleanup cycle is designed to be synchronized with index increments, defragmentation, and backup windows. 7-15 days is the optimal window, verified through engineering practice (based on actual testing with 1.33 million data points). These three factors together constitute a technical necessity, rather than an arbitrary choice.

[0045] 3.4.2 Beneficial effects at the industry level

[0046] • It provides practitioners (creators) with protection and monetization channels for their original value. Every matching, editing, and generation can be traced back to the original creator, protecting their right to income.

[0047] • It provides consumers (users) with a mechanism for quantifying and rewarding the value of their behavior. Every browsing, DIY, and transaction can be accumulated into personal data assets and can participate in secondary profit sharing.

[0048] • It provides a technical foundation for the compliant circulation of data elements in the market, and through the hard binding of lightweight coding index with digital RMB, it realizes full-chain traceability of transactions and automatic tax verification;

[0049] • LBS near-field matching (accuracy ±10 meters) supports localized consumption decision-making scenarios.

[0050] 3.4.3 Visual Data Visualization Industry Effects

[0051] This invention achieves an "instant visualization" effect for visual data through a combination of technologies including "lightweight fusion encoding + matching-driven architecture + three-level visual processing," resulting in beneficial effects in user experience and industrial applications that cannot be replicated by existing technologies.

[0052] Verifiability of technical effects

[0053] Through engineering verification (encoding and confirming the rights to 1.33 million visual data points, and listing 120 asset certificates on the Guangzhou Data Exchange), this invention achieves a generational advantage over existing technologies in the following indicators:

[0054]

[0055] The data above shows that visual data visualization is not only an improvement in user experience, but also a beneficial effect that is quantifiable and verifiable based on hard technical indicators.

[0056] 3.5 The irreplaceable nature of core technological features

[0057] 3.5.1 Definition of the technical attributes of "matching-driven"

[0058] The "matching-driven" technology described in this invention is essentially a mandatory dependency relationship implemented through hard coding at the system architecture level, which is different from the optional search or recommendation functions in existing technologies.

[0059] (1) The matching result serves as the only legal entry point for all subsequent functional modules. This constraint is enforced by the unified permission verification module at the bottom of the system through code logic. It cannot be bypassed or removed by modifying the configuration file, switching the function switch, replacing the module, encapsulating the interface, adapting the middleware, etc. It is a pure technical architecture design.

[0060] (2) The lightweight encoded index on which the matching process depends is generated by binding the author's ownership metadata, hardware identifier, and algorithm feature triple verification information. The index itself is a technical carrier. Any identifier information that fails the triple verification is automatically determined to be an invalid index by the system's underlying layer and cannot trigger the matching driving unit. It is a pure technical verification mechanism.

[0061] (3) Once the matching result is output, the system bottom layer automatically binds the result with the subject index that initiated the matching, the visual data index that was matched, the matching timestamp, and the matching parameters in a permanent encrypted manner to form an immutable technical traceability chain. This binding process is automatically executed by the system bottom layer without any manual intervention and is a pure technical binding mechanism.

[0062] The "matching driver" described in this invention is a technical solution at the computer system architecture level, with a clear technical implementation path and hardware adaptation requirements.

[0063] 3.5.2 Technological Innovation of "Matching as the Sole Precondition"

[0064] In existing technologies, the output of "retrieval" or "recommendation" is usually an optional reference. Users can choose to adopt, ignore, or bypass it. There is no mandatory dependency between it and subsequent operations, and no hard-coded constraints are set at the system architecture level.

[0065] The "matching-driven" approach described in this invention establishes a mandatory dependency relationship between its output and subsequent operations at the system's underlying level. The specific differences are as follows:

[0066] The above differences indicate that the "matching-driven" approach described in this invention represents a fundamental restructuring at the system architecture level and possesses significant technological innovation.

[0067] 3.5.3 The Technical Necessity of "Prohibiting the Generation of New Indexes"

[0068] The core technical purpose of prohibiting the generation of new indexes is to ensure the uniqueness and traceability of ownership. The specific technical logic is as follows:

[0069] (1) If new independent indexes are allowed to be generated during the editing and generation process, the newly generated visual content will form a break in ownership with the original content, and it will be impossible to accurately trace back to the original creator during the subsequent circulation process. This invention achieves full ownership traceability of the editing and generation results through the technical means of "permanently binding the original index with the auxiliary metadata".

[0070] (2) This mechanism avoids both the storage expansion and the decline in retrieval performance caused by the index explosion. Through engineering verification, it can keep the index storage volume within 0.03% of the original image and the retrieval response time within 1 second (million-level index database). It is a necessary technology for ownership protection and system performance optimization.

[0071] (3) The “subsidiary index” and “new index” described in this invention have an essential technical difference: the subsidiary index does not have independent circulation attributes, and any use or transaction of it requires simultaneous verification of the original index on which it depends; while the new index has independent circulation attributes and can be traded after being separated from the original ownership. The two have completely different technical architectures, data models and circulation logics.

[0072] 3.5.4 The Technical Necessity of "Automatic Destruction of Non-Transaction Data After 7-15 Days"

[0073] The storage period is set based on hard constraints from three technical factors:

[0074] (1) Compliance requirements of privacy protection regulations: Visual data is temporary behavioral data before the transaction is confirmed. It is necessary to ensure user privacy compliance through time limit control, which is a compliance technical constraint.

[0075] (2) Technical calculation of storage cost optimization model: For every 10% increase in the size of the index database, the retrieval response time increases by about 3% and the storage cost increases by about 15%. The 7-15 day window period is the optimal balance point between retrieval performance and storage cost, which has been verified by engineering of 1.33 million visual data.

[0076] (3) Technical design of the system's automatic cleanup cycle: The cleanup cycle is synchronized with the index database increment, defragmentation, and backup window. 7-15 days is the optimal window verified by engineering, which can ensure the balance between system operation and maintenance efficiency and user experience. Attached Figure Description

[0077] Appendix Figure 1 Note: This diagram shows the top-level core architecture of the present invention (overall architecture of the full-vision data matching and rights confirmation traceability system). Data flows in a unidirectional arrow pattern to form a seamless technical closed loop, with lightweight coding indexes serving as the sole core carrier throughout the entire process. The system is divided into four main modules, with each module's sub-items and core functions as follows. The component and module numbers in the diagram are as follows:

[0078] • User-side terminal (100): includes mobile phone (101), computer (102), smart wearable (103), and professional data collection equipment (104); its core functions are full-vision data collection, privacy pre-inspection, demand release / transaction payment, providing de-identified basic data for subsequent coding, but without coding generation capability;

[0079] • Core processing unit (200): Includes optical vector encoding server (201), lightweight encoding server (202), AIUGC large model server (203), traceability matching server (204), compliant transaction docking hardware (205), and privacy desensitization hardware (206); its core functions are to complete the integrated native rights confirmation of optical vector / lightweight fusion encoding and AIUGC large model encoding, output a unique legal lightweight encoding index, and realize industry differentiation, traceability matching basic verification, and hardware docking adaptation;

[0080] • Application Service Layer (300): Includes a lightweight fusion coding module (301), a native rights confirmation module (302), a supply and demand matching module (303), a privacy desensitization module (304), an interface access module (305), a user memory module (306), a three-level visual processing module (307), a user revenue distribution module (308), a compliant transaction module (309), and a secondary creation auxiliary index generation module (310); the core functions are full-link traceability query, three-level visual processing without new index generation, compliant transaction support, and two-way revenue distribution;

[0081] • Encrypted Data Layer (400): Includes a lightweight coded index library (401), a visual digital asset library (402), an individual AI database (403), encrypted rights confirmation storage (404), and a personal exclusive temporary database (405); its core function is to encrypt and store indexes and related data, provide users with a 7-15 day temporary cache for non-self-created content (automatically destroyed if no transaction occurs), use the lightweight coded index as the core retrieval / management identifier, and only accept legal indexes and related data generated by this system.

[0082] This image is attached. Figure 2-6 Based on the technology, the data flow direction is: user-side terminal (100) → core processing terminal (200) → application service layer (300) → encrypted data layer (400).

[0083] Appendix Figure 2 Note: This diagram illustrates the entire process from the consumer's perspective. The steps in the diagram are numbered as follows:

[0084] Consumer behavior input (S201) → Privacy pre-check (S202) → Contribution coin generation and recording (S203) → Lightweight coding index feature matching (S204, similarity ≥80%, industry-adjustable) → Dual-role index identity confirmation (S205) → System compliance verification passed (S206) → Supply and demand matching successful (S207) → Two-way revenue settlement (S208) → Consumer-exclusive digital space (S209) → [Non-self-created content] Personal exclusive temporary database (S210, automatically destroyed after 7-15 days without transactions, only viewable) / Collect, cannot be downloaded / exported, corresponding to the third processing mode in claims 4 and 6) / [Self-created / Transactional Content] Permanent exclusive database (S211, corresponding to claims 4 and 8) → Exclusive individual AI call and run (S212) → [Secondary Creation] Secondary creation auxiliary index generation module (S213) → Generate auxiliary index based on original index (S214, corresponding to claims 3 and 7) — Secondary creation content circulation / transaction (S215, dual index verification + bidirectional distribution of revenue) → 100% data sovereignty belongs to the consumer (S216).

[0085] Appendix Figure 3 Note: This diagram illustrates the entire process from the practitioner's perspective. The steps in the diagram are numbered as follows:

[0086] Visual data upload by practitioners (S301) → Privacy pre-check (S302) → Light vector / lightweight fusion encoding (S303) → Native ownership confirmation of AIUGC large model (S304) → Generation of unique lightweight encoding index (S305) → Visual digital asset listing (S306) → Lightweight encoding index feature matching (S307, similarity ≥80%, industry-adjustable) → Dual role index identity confirmation (S308) → System compliance verification passed (S309) → Successful supply and demand matching (S310) → Two-way revenue settlement (S311) → Practitioner's exclusive digital space (S312) → Full lifecycle management of visual digital assets (S313) → [User secondary creation] Binding of original index to auxiliary index (S314, corresponding to claims 3 and 7) → User secondary creation content transaction (S315) → Obtaining basic revenue sharing of secondary transactions based on original index (S316) → 100% data sovereignty belongs to practitioners (S317).

[0087] Appendix Figure 4 Note: This diagram is the core flowchart of the three-level visual processing. The steps in the diagram are numbered as follows:

[0088] Lightweight encoding index trigger (S401) → Primary matching (S402, existing data retrieval + multi-content subject tracing, result bound to the original index, corresponding to the matching driving unit / step in claims 1 and 6) → Intermediate intelligent matching (S403, personalized DIY adjustment + cross-scene adaptation, adjustment parameters as auxiliary metadata bound to the original index, corresponding to claims 1.3 (first processing mode) and 6.3.2 (second processing mode)) → Advanced generation (S404, customized visual content generation, generation parameters as auxiliary metadata bound to the original index, corresponding to the second processing mode in claims 1 and 6) → [Non-secondary creation] Output to matching / transaction stage (S405) / [Secondary creation] Trigger auxiliary index generation module (S406) → Generate auxiliary index based on the original index (S407, not a new index, only an extended identifier, corresponding to claims 3 and 7) → Secondary creation content storage (S408, synchronously bound to the original index + auxiliary index) → No new index generation throughout the process (S409, corresponding to claims 1, 6, and 11).

[0089] Appendix Figure 5 Note: This diagram is a flowchart of compliant transactions and profit distribution. The steps in the diagram are numbered as follows:

[0090] Lightweight coding index legality verification (S501) → Supply and demand index feature matching meets standards (S502, ≥80%) → Dual role index identity confirmation (S503) → System full-dimensional compliance verification passed (S504) → Supply and demand matching successful (S505) → Generate matching success identifier and bind it to both indexes (S506) → Receive platform transaction confirmation instruction (S507) → Digital RMB hard-binding transaction (S508, corresponding to step 4 of claim 6) → Transaction full-link traceability registration (S509) → Tax payment certificate generation (S510) → [First transaction] Two-way revenue settlement (S511, practitioner + consumer) → [Secondary creation transaction] Original index + auxiliary index dual verification (S512, corresponding to claims 3, 7) → Secondary transaction revenue two-way distribution (S513, practitioner basic share + user main revenue) → All transaction / distribution records permanently bound to the corresponding lightweight coding index (S514) → Transaction records synchronously stored in the encrypted data layer (S515).

[0091] Appendix Figure 6 Note: This diagram is a flowchart of the dedicated digital space and data accumulation mechanism. The steps in the diagram are numbered as follows:

[0092] Lightweight Encoded Index (Unique Access Key) (S601) → Exclusive Digital Space Entry Verification (S602) → [Non-Self-Created / No-Transaction Content] Personal Exclusive Temporary Database (S603, Automatically Destroyed After 7-15 Days Without Transactions, Only for Review / Collection, Corresponding to the Third Processing Mode in Claims 4 and 6) / [Self-Created / Transaction Content] Permanent Exclusive Database (S604, Corresponding to Claims 4 and 8) → AI Database Retrieval (S605) → Intelligent Matching / Smart Allocation / Generation (S606) → [Secondary Creation Trigger] Secondary Creation Auxiliary Index Generation Module (S607) → Generate Auxiliary Index (S608, Bind User Ownership + Secondary Creation Information, Corresponding to Claims 3 and 7) → Secondary Creation Content (S609, Bind Original Index + Auxiliary Index) → [Secondary Creation Circulation] Generate Double-Index Encrypted Traceability Link (S610) / [Secondary Creation Transaction] Original Index + Auxiliary Index Legality Verification (S611) → Two-way Distribution of Secondary Transaction Profits (S612) → Permanent Binding of All Records to the Original Index + Auxiliary Index (S613) → Standardized Data Interface (S614, for subsequent "Subject with AI" system calls, call conditions: legal index + biometric verification) → 100% Data Sovereignty Belongs to the Corresponding Subject (S615) → Mirror Space Aggregation (S616): When an individual AI continuously runs, interacts, and accumulates in its exclusive digital space, and is aggregated by the space creator, the system automatically generates a unique "Mirror Space Index." This index, as a lightweight coded index applied in the set dimension, binds the space creator's ownership and establishes a permanent mapping relationship with all individual indexes, asset indexes, and behavior indexes within the space, forming a digital world that can be independently circulated, traceable, and sovereign (corresponding to claims 1, 6, and 12).

[0093] Note: Digital personality and individual AI are optional upgrade features for users. They are not a necessary step in this process and do not affect the operation of the system's core functions.

[0094] Appendix Figure 7 Note: The six-dimensional comparison diagram of the technical effects of this invention corresponds to the quantitative indicators in claims 1 and 6, and to Section 3.4.1 "Beneficial Effects at the Technical Level" and Section 5.3.1 "Technical System Characteristics" of the specification. This diagram illustrates the comparison between this invention and existing mainstream AIGC models in six dimensions: efficiency, ownership, computing power, memory, individual AI, and value distribution.

[0095] Appendix Figure 8Note: This chart compares the cost of this invention with existing mainstream AIGC generation modes, corresponding to the generation computing power consumption indicators in claims 1 and 6 and the "beneficial effects at the technical level" in Section 3.4.1 of the specification. The data in this chart clearly demonstrates the one ten-thousandth advantage of this invention in terms of computing power consumption, time cost, and energy cost, corresponding to the quantitative characteristics of "generation computing power consumption not exceeding 0.03 GFLOPS" and "generation latency not exceeding 100 ms" in claims 1 and 6. Detailed Implementation

[0096] 5.1 Unified Framework for Implementation Examples

[0097] (The method steps (steps 1-5) corresponding to claim 6)

[0098] The six industry-specific implementations of this invention (design, decoration, medical, consulting, fashion, and catering) all follow the following unified technical process, with only parameters such as encoding dimensions and matching thresholds slightly adjusted according to industry characteristics. (Partial engineering verification has been completed in the design and decoration industries: it can match hundreds of design and decoration cases in one second, and trace the data back to multiple practitioners. Currently, 120 visual asset certificates have been compliantly uploaded to the Guangzhou Data Center, with encoding response ≤100ms, matching accuracy ≥99.9%, and index storage ≤0.03% of the original image. The technical feasibility and stability have been fully implemented.)

[0099] Step 1, Data Upload and Privacy Pre-check (S1, corresponding to Figure 2 S202 / S203, Figure 3 (S302): Users upload visual data or initiate requests through the terminal, which automatically performs privacy anonymization processing. Consumer behavior data is synchronized and rights are confirmed, and browsing behavior is quantified into contribution coins. The rights confirmation process is completed seamlessly and automatically, requiring no additional action from the user.

[0100] Step 2, Encoding and Rights Confirmation (S2, corresponding to Figure 3 (S303-S305): The lightweight encoding server performs lightweight fusion encoding on visual data, generating a unique lightweight encoding index and synchronously binding the creator's ownership information. Encoding response time ≤ 100ms (single 2K image), encoding accuracy ≥ 99.9%, and post-encoding index storage size ≤ 0.03% of the original image. Encoding is equivalent to ownership confirmation and is completed automatically.

[0101] Step 3, matching driver (S3, corresponding to the "matching driver unit / step" in claims 1 and 6), and Figure 2 S204 Figure 3(S307): The system transforms user-input visual requirements (reference images, text descriptions, voice commands, etc.) into query features, performs similarity retrieval in a lightweight coded index, and outputs matching results. Matching response time is ≤1 second (million-level index), and the matching similarity threshold is ≥80% (adjustable by industry: 85% for the design industry, 90% for the medical industry, etc.). The matching result is set as the sole prerequisite for triggering all subsequent functions (corresponding to claims 1, 6, 9, and 10).

[0102] Step 4, Progressive processing (S4, corresponding to the "matching extension processing unit / step" in claims 1 and 6) and Figure 4 Based on the matching results, the following three processing modes are executed.

[0103] • First processing mode (intelligent matching) (S41, corresponding to the "first processing mode" in claims 1 and 6) and Figure 4 S403: Based on user instructions, visual features associated with the matching results are edited. All editing operations are bound to the original index as auxiliary metadata, and the system underlying layer forcibly prohibits the generation of new lightweight coded indexes. Processing latency ≤150ms.

[0104] • Second processing mode (generation) (S42, corresponding to the "second processing mode" in claims 1 and 6) and Figure 4 S404 (Chinese version): Based on the original indexes corresponding to multiple matching results, visual features are decoupled and recombined to generate fused visual content. The generation parameters are bound to the original index set used as auxiliary metadata. The generation result does not generate new independent indexes. The generation latency is ≤100ms, the computing power consumption is ≤0.03GFLOPS (less than one ten-thousandth of the existing AIGC mode), and 100 sets of solutions can be output within 1 second.

[0105] • Third processing mode (precipitation) (S43, corresponding to the "third processing mode" in claims 1 and 6) Figure 6 (S603 / S604): User operation data and preference data generated in the above processing modes are stored in a user-specific database with the original index as the unique access key, forming a user feature memory that can be used for subsequent matching and retrieval. Non-transaction data storage period is 7-15 days, and it will be automatically and irreversibly destroyed upon expiration, except in judicial evidence collection scenarios; during this period, only playback / favoritism is supported, and downloading / exporting is prohibited.

[0106] Step 5, Transaction Settlement (S5, corresponding to the "Transaction Settlement" step in claim 6) Figure 5 Digital RMB transactions are triggered based on matching results, and transaction records, payment information, tax payment certificates, and corresponding indexes are permanently bound. The initial transaction settlement delay is ≤3 seconds, and the subsequent transaction profit-sharing delay is ≤5 seconds. A transaction unlocking mechanism ensures that only users who complete compliant transactions can obtain the high-definition source files.

[0107] Step 6, secondary creation and dual index binding (S6, corresponding to claims 3, 7 and...) Figure 4 S406-S408 Figure 6 (S607-S613): When a user creates derivative works on purchased content, the system generates a secondary index based on the original index and binds it to the secondary creator's ownership information. The derivative content is simultaneously bound to both the original and secondary indexes; any subsequent use or transaction requires verification of both indexes. The secondary index cannot be transferred independently; if the original index becomes invalid, the secondary index is simultaneously invalidated.

[0108] Step 7, Mirror space aggregation (corresponding to claims 1, 6, 12 and...) Figure 6 (S616): When an individual AI continuously operates, interacts, and accumulates within its dedicated digital space, and is aggregated by the space creator (individual, organization, or multi-entity alliance), the system automatically generates a unique "mirror space index." This index is an application of lightweight coded indexes at the set dimension, binding the space creator's ownership information and establishing a permanent mapping relationship with all individual indexes, asset indexes, and behavior indexes within the space. The mirror space itself, as an independently transferable, traceable, and sovereign digital asset, requires dual verification through both the space index and the individual index for any access, interaction, or asset transfer.

[0109] All processing steps support LBS near-field matching with a positioning accuracy of ±10 meters, enabling localized precise docking within a range of 100 meters and 1 kilometer.

[0110] 5.2 Six Types of Industry-Specific Differentiated Parameters

[0111] In this industry-specific implementation, the encoding dimension is 256, and the matching threshold is set to ≥85%. Application scenarios include design-related visual data such as scheme renderings, construction drawings, and 3D models. These industry-differentiated parameters represent the application of the quantitative features (encoding accuracy ≥99.9%, matching similarity threshold ≥80%) from claims 1 and 6 in specific scenarios.

[0112] In the home renovation industry implementation example, the coding dimension is 384 dimensions, the matching threshold is ≥80%, and the application scenarios include renovation data such as floor plans, renderings, and material diagrams, supporting traceability for multiple stakeholders including designers, material suppliers, and construction companies. This industry-specific parameter represents the application of the quantified features in claims 1 and 6 in specific scenarios.

[0113] In the medical and aesthetic medicine industry implementation, the encoding dimension is 512 dimensions, the matching threshold is ≥90%, and the application scenarios include medical data such as CT images, MRI images, and aesthetic medicine plans. Deep anonymization hardware is used to ensure privacy compliance. This industry-differentiated parameter is the application of the quantitative features in claims 1 and 6 in specific scenarios, and reflects the synergy between LBS positioning and privacy anonymization in claim 5.

[0114] In the consulting industry implementation example, the coding dimension is 256 dimensions, the matching threshold is ≥85%, and the application scenarios include consulting data such as consulting cases and solution documents. The efficiency of consulting services is improved by more than 30% compared with the traditional model.

[0115] In the clothing industry implementation example, the encoding dimension is 192 dimensions, the matching threshold is ≥82%, and the application scenarios include clothing styling, style matching and other clothing data. It can be adapted to local low power consumption operation of smart wearable devices.

[0116] In the catering industry implementation example, the coding dimension is 128 dimensions, the matching threshold is ≥80%, and the application scenarios include catering data such as dishes, recipes, and banquet plans. It supports dual index binding between chefs and brand owners.

[0117] 5.3.1 Characteristics of the Technical System

[0118] The technical system of this invention has the following characteristics:

[0119] (1) Closed-loop technology system: The lightweight fusion coding and AIUGC large model are deeply coupled to form a closed-loop technology system of collection, coding, rights confirmation, processing, transaction and distribution.

[0120] (2) Uniqueness of core carrier: The generation of lightweight coding index relies on the synergistic effect of dedicated hardware, dedicated algorithm and dedicated model, and has the triple characteristics of identifiable algorithm features, immutable ownership metadata and verifiable hardware identification.

[0121] (3) Three-level matching system: The generation of new encoding indexes is prohibited throughout the process. Secondary creation only generates auxiliary indexes and is forcibly bound to the original index. All auxiliary metadata generated by the operation is permanently bound to the original index.

[0122] (4) Quantitative indicator verification: After engineering verification of 1.33 million visual data, the coding accuracy is ≥99.9%, the storage amount is ≤0.03% of the original image, the response time is ≤100ms, and the generation computing power is ≤0.03GFLOPS.

[0123] (5) Data ownership confirmation mechanism: Data ownership confirmation is completed automatically and synchronously with data generation, without the need for additional user operation.

[0124] (6) Value distribution between two entities: practitioners enjoy the revenue from creation, and consumers enjoy the revenue from behavior. The revenue is automatically distributed between the two through the smart contract of digital RMB.

[0125] 5.3.2 Engineering Verification and Institutional Response

[0126] The key technical indicators of the above embodiments, including encoding accuracy ≥99.9%, storage capacity ≤0.03%, and response time ≤100ms, have been verified through engineering of 1.33 million visual data and 120 data asset certificates have been listed on the Guangzhou Data Exchange.

[0127] Summarize:

[0128] This invention achieves a complete technical closed loop for the confirmation, circulation, and transaction of visual data through a lightweight coded index unique identification mechanism and a hard-coded architecture with "matching as the sole prerequisite." The system's underlying layer is forcibly bound to digital RMB settlement, prohibiting the generation of new indexes, and achieves value distribution between practitioners and consumers through three-level visual processing. After engineering verification with 1.33 million visual data samples, all core technical indicators have met the standards, providing an engineerable technical solution for the marketization of data elements.

Claims

1. A system for full visual data matching, rights confirmation, and traceability, characterized in that, include: The encoding unit is configured to perform lightweight fusion encoding on the collected visual data to generate a unique lightweight encoding index. The index is a permanent binding carrier between the visual data and the author's ownership information, and is the unique legal identifier for the visual data to enter the system for any subsequent processing. The lightweight fusion encoding has an encoding response time of no more than 100 milliseconds, an encoding accuracy of no less than 99.9%, and an index storage size of no more than 0.03% of the original image. The matching drive unit is configured to: receive externally input visual requirements and convert them into query features; perform similarity retrieval in the lightweight encoding index library based on the query features, output matching results, with a matching response time of no more than 1 second and a matching similarity threshold of no less than 80%; and set the matching results as the sole prerequisite for triggering all subsequent functions of the system. The matching extension processing unit is configured to perform one or more of the following operations based on the matching results output by the matching drive unit: First processing mode: edit the visual features associated with the matching results based on user instructions; all editing operations are bound to the original index as auxiliary metadata; the system's underlying layer forcibly prohibits the generation of new lightweight encoding indexes in this mode, with a processing delay of no more than 150 milliseconds. The second processing mode is to decouple and reorganize the visual features based on the original indexes corresponding to multiple matching results, generate the fused visual content, and bind the generated parameters as auxiliary metadata to the set of original indexes used. The generated result does not generate new independent indexes, the generation delay is no more than 100 milliseconds, and the computing power consumption is no more than 0.03 GFLOPS. The third processing mode: The user operation data generated in the above processing mode is stored in a user-exclusive database with the original index as the unique access key, forming a user feature memory that can be called for subsequent matching. The storage period for non-transactional data is 7-15 days, and it is automatically and irreversibly destroyed upon expiration, except in judicial evidence collection scenarios. The matching driving unit and the matching extension processing unit form a progressive execution relationship. Any visual data that has not been matched by the matching driving unit and output a matching result cannot enter the matching extension processing unit to perform any operation.

2. The system according to claim 1, characterized in that, The first processing mode includes: performing editing operations on the visual features associated with the matching results based on user instructions, and packaging the operation type, operation parameters, and operation timestamp into auxiliary metadata, and establishing a permanent mapping relationship with the original lightweight coded index; the system automatically verifies the index library after each editing operation to ensure that no new index records are added.

3. The system according to claim 1, characterized in that, The second processing mode includes: extracting features from the original indexes corresponding to multiple matching results and fusing them to generate the result; packaging the hash of the original index set used, the generation model identifier, the generation parameters, the generation timestamp, and the generation result hash into auxiliary metadata and establishing a permanent mapping relationship with all the original indexes used; the generated result itself does not have any flow attributes independent of the original index set, and any subsequent use of it requires simultaneous verification of all the original indexes it depends on.

4. The system according to claim 1, characterized in that, The third processing mode includes: storing users' historical matching records, operation preferences, and transaction data into a dedicated database with the original index as the unique identifier; the database can only be accessed by the entity holding the corresponding original index, and any third-party system that does not hold a legitimate index cannot call the data in the database for model training or inference; the database adopts hierarchical storage, with a permanent storage area storing self-created and traded content, and a temporary storage area storing non-traded content. The storage period for non-traded data is 7-15 days, and it is automatically and irreversibly destroyed upon expiration, except in judicial evidence collection scenarios.

5. The system according to claim 1, characterized in that, The system supports matching based on geographic location information, which is obtained through the terminal's positioning capabilities. The acquisition of geographic location information is controlled by user authorization.

6. A method for full visual data matching for rights confirmation and traceability, characterized in that, Includes the following steps: Step 1, Encoding and Rights Confirmation: Perform lightweight fusion encoding on the collected visual data to generate a unique lightweight encoding index; the index is a permanent binding carrier between the visual data and the author's ownership information; the encoding response time is no more than 100 milliseconds, the encoding accuracy is no less than 99.9%, and the index storage size is no more than 0.03% of the original image; Step 2, Matching-Driven: Receive externally input visual requirements and convert them into query features; based on the query features, perform similarity retrieval in the lightweight encoding index library, output matching results, the matching response time is no more than 1 second, and the matching similarity threshold is no less than 80%; set the matching result as the unique prerequisite for triggering all subsequent steps. Step 3, Matching Extension Processing: Based on the matching results, perform one or more of the following operations: First processing mode: Edit the visual features associated with the matching results based on user instructions. All editing operations are bound to the original index as auxiliary metadata. The generation of new lightweight coded indexes is prohibited throughout the process, and the processing delay is no more than 150 milliseconds; Second processing mode: Decouple and reorganize the visual features based on the original indexes corresponding to multiple matching results to generate fused visual content. The generation parameters are bound to the set of original indexes used as auxiliary metadata. The generation results do not generate new independent indexes. The generation delay is no more than 100 milliseconds, and the computing power consumption is no more than 0.03 GFLOPS; The third processing mode: The user operation data generated in the above processing mode is stored in a user-exclusive database with the original index as the unique access key, forming a user feature memory that can be used for subsequent matching and calling. The storage period for non-transactional data is 7-15 days, and it will be automatically and irreversibly destroyed upon expiration, except in judicial evidence collection scenarios; Step 4, transaction settlement: Digital RMB transactions are triggered based on the matching results. The settlement delay for the first transaction is no more than 3 seconds, and the profit sharing delay for the second transaction is no more than 5 seconds. Transaction records, payment information, tax payment certificates and corresponding indexes are permanently bound; Step 2 and Step 3 form a progressive execution relationship; Any visual data that has not been matched in Step 2 and output a matching result cannot enter Step 3 to perform any operation.

7. The method according to claim 6, characterized in that, In the second processing mode: the generated result itself does not have a flow attribute independent of the original index set, and any subsequent use of it requires simultaneous verification of all the original indexes it depends on.

8. The method according to claim 6, characterized in that, In the third processing mode: the user-exclusive database can only be accessed by the entity holding the corresponding original index. Any third-party system that does not hold a valid index cannot call the data in the database for model training or inference.

9. The system according to claim 1 or the method according to claim 6, characterized in that, In the matching driving unit, the matching retrieval includes, but is not limited to, any matching method such as similarity calculation, feature comparison, semantic understanding, and multimodal fusion. The judgment criteria for the matching result are dynamically set by the system according to the application scenario. However, regardless of the matching method or judgment criteria used, the output matching result is set as the only prerequisite for triggering all subsequent functions of this system.

10. The system according to claim 1 or the method according to claim 6, characterized in that, The aforementioned unique precondition is a hard-coded constraint at the system architecture level, which cannot be bypassed or removed by modifying configuration files, switching function switches, replacing modules, encapsulating interfaces, adapting middleware, or other means. Any visual data that has not been matched and has not output a matching result will be automatically identified and blocked by the unified permission verification module at the system's underlying layer, regardless of how it attempts to enter the subsequent processing stage (including but not limited to direct calls from internal system modules, imports from external interfaces, data format conversions, re-encoding of identifiers, indirect transmission through third-party systems, etc.). The blocking record is permanently bound to the index of the entity that initiated the operation.

11. The system according to claim 1 or the method according to claim 6, characterized in that, The system forcibly prohibits the generation of new lightweight coded indexes or any unique identifiers that are functionally equivalent to indexes; the "unique identifiers that are functionally equivalent to indexes" include, but are not limited to, derived identifiers, extended IDs, version numbers, packaging containers, identifiers imported after external recoding, and any other identification information that can independently identify the ownership or flow path of visual data.

12. The system according to claim 1, characterized in that, Also includes: The mirror space aggregation unit is configured to automatically generate a unique mirror space index when it detects that multiple lightweight coded indexes correspond to individual AI databases, visual digital asset libraries, and behavior records that form an aggregation relationship under user authorization. The mirror space index is an extension of the lightweight coded index in the set dimension, binds the ownership information of the space creator, and establishes a permanent mapping relationship with all individual indexes, asset indexes, and behavior indexes within the aggregation scope. As an independently transferable, traceable, and sovereign digital asset, any access, interaction, or asset transfer of the mirror space must be verified by both the mirror space index and the individual index.

13. The system according to claim 1 or the method according to claim 6, characterized in that, The storage period for non-transactional data is forcibly limited by the system. This period is set based on privacy protection regulations, storage cost optimization models, and the system's automatic cleanup cycle. Once the storage period is exceeded, the system will perform irreversible destruction. Except for legal scenarios for judicial evidence collection, no user operation (including but not limited to manual extension, paid retention, and re-import after export) can prevent or delay the destruction process.

Images