Agricultural chain supervision system and method based on trusted identity authentication and plot surveying and mapping

Abstract

The invention discloses an agricultural chain supervision system and method based on trusted identity authentication and plot surveying and mapping, belongs to the technical field of digital management, and aims to solve the technical problem of how to realize identity authentication, plot surveying and mapping and data trusted association of agricultural supervision. Comprising a hand-held terminal device, a digital authentication module and a planting registration module, the digital authentication module interacts with the hand-held terminal device and is used for receiving a fingerprint image, a face image and a certificate image, generating a farmer identity identifier UID based on a face feature vector, fingerprint feature data and structured certificate information of a farmer, and registering the planting registration module. Based on the UID, the face feature vector, the fingerprint feature data, the structured certificate information and the digital certificate of the peasant household, constructing the identity information of the peasant household; the planting registration module binds the identity information of the farmer, the space information of the planting plot and the planting information of the planting plot into ternary information based on the UID of the farmer.

Description

Technical Field

[0001] This invention relates to the field of digital management technology, specifically to an agricultural chain supervision system and method based on trusted identity authentication and land surveying. Background Technology

[0002] Currently, there are three core technical challenges in agricultural supply chain supervision:

[0003] 1. Identity authentication challenges: Farmers generally lack a unified and easily verifiable digital identity; traditional certificates are easily damaged and forged; poor network coverage makes online authentication difficult; biometric technology needs to adapt to local characteristics such as strong light and dark skin color, and existing technologies have insufficient accuracy in complex environments.

[0004] 2. Challenges in land surveying: unclear land boundaries and incomplete ownership registration; professional surveying equipment is expensive and complex to operate, making it difficult to popularize; consumer-grade GNSS equipment lacks accuracy in complex terrain (jungles, hills), failing to meet the needs of precise surveying; and there is a lack of efficient and intuitive technologies for collecting land boundaries and related planting information.

[0005] 3. Data credibility and correlation challenges: Manually entered data is prone to errors and difficult to verify; identity information, land information, and planting information are fragmented and cannot form a credible chain; there is a lack of effective technical means to strongly link and bind the identity of agricultural entities with their planting land, resulting in a lack of traceability and credibility of regulatory data, which cannot support accurate tax collection, subsidy distribution, and policy formulation.

[0006] How to achieve identity authentication, land surveying, and reliable data association in agricultural supervision are technical problems that need to be solved. Summary of the Invention

[0007] The technical objective of this invention is to address the above-mentioned shortcomings by providing an agricultural chain supervision system and method based on trusted identity authentication and land surveying, thereby solving the technical problem of how to achieve full-process supervision of the agricultural industry chain.

[0008] In a first aspect, the present invention provides an agricultural chain supervision system based on trusted identity authentication and land surveying, comprising a handheld terminal device, a digital authentication module, and a planting registration module;

[0009] The handheld terminal device is used to collect farmers' fingerprint images, facial images, and document images as authentication images. The authentication images are encrypted and sent to the digital authentication module. This is used to support each farmer in mapping their planting plots, collecting spatial information of the planting plots, and encrypting and sending the spatial information of the planting plots to the planting registration module.

[0010] The digital authentication module interacts with the handheld terminal device to receive fingerprint images, face images, and document images. It performs face recognition based on the face image to obtain a face feature vector and corresponding confidence level, performs fingerprint recognition based on the fingerprint image to obtain fingerprint feature data, and performs OCR recognition on the document image to obtain a document face feature vector, structured document information, and corresponding confidence level. It then authenticates farmers based on the face feature vector and document face feature vector. For authenticated farmers, it generates a farmer identity identifier (UID) based on the farmer's face feature vector, fingerprint feature data, and structured document information. Finally, it constructs the farmer's identity information based on the farmer's UID, face feature vector, fingerprint feature data, structured document information, and digital certificate.

[0011] The planting registration module interacts with the handheld terminal device to receive spatial information of each farmer's planting plot, collect metadata and planting information of the user's planting plot, bind the farmer's identity information, spatial information of the planting plot and planting information of the planting plot into three-element information based on the farmer's UID, calculate the hash value of the three-element information, store the hash value of the three-element information in the blockchain, and construct a Merkle tree based on the hash value of the three-element information. The hash value of the three-element information in the Merkle tree is the leaf value. The metadata includes timestamp, geographical location, device information and operator information. The timestamp is the time when the planting plot was surveyed by the handheld terminal device, the geographical location is the GPS coordinate information of the handheld terminal device, and the device information is the device ID and status of the handheld terminal device.

[0012] Preferably, the handheld terminal device is equipped with a capacitive sensor and a Gabor filter bank, which are used to collect fingerprint images of farmers through the capacitive sensor, and to enhance the collected fingerprint images through the Gabor filter bank to obtain enhanced fingerprint ridge images, and then send the enhanced fingerprint ridge images as fingerprint images to the digital authentication module.

[0013] Correspondingly, the digital authentication module is used to perform fingerprint recognition based on the enhanced fingerprint ridge image and extract fingerprint feature data, which includes the coordinates, direction, and type information of fingerprint minutiae.

[0014] Preferably, the digital authentication module is used to perform image preprocessing on the face image based on the Gamma correction algorithm. The image preprocessing includes adaptive brightness adjustment, illumination compensation, and nonlinear transformation to obtain a brightness-corrected face image. The brightness-corrected face image is then enhanced using the CLAHE algorithm to obtain a contrast-enhanced face image. The contrast-enhanced face image is then used as input to perform feature extraction and identity recognition through a face recognition model, outputting a face feature vector and confidence level.

[0015] Preferably, the face recognition model is a network model built on MobileNetV3-Small, which is used to extract features from the input image, enhance the extracted features based on the attention mechanism, and output the face feature vector and confidence score.

[0016] As a preferred option, user identity information is encrypted and stored in an identity information database.

[0017] Preferably, the digital authentication module is used to perform OCR recognition on document images as follows:

[0018] Image preprocessing is performed on the document images, including noise reduction, correction, enhancement, text region localization and segmentation, and image size and format are standardized.

[0019] For the preprocessed document images, features are extracted using the ResNet-34 model, converting the two-dimensional image into a one-dimensional sequence while preserving spatial location information, and outputting feature vectors.

[0020] Based on the output feature vector, the sequence features are decoded into text through LSTM layer for context understanding, and then the fields are classified and structured to output the ID card face feature vector, structured ID card information and corresponding confidence scores.

[0021] Preferably, the handheld terminal device is equipped with a GNSS chip, and when mapping the planting plot through the handheld terminal device, the following operations are included:

[0022] GNSS data is obtained by collecting coordinate points, current position, and traveled trajectory using a GNSS chip at a predetermined frequency.

[0023] The original GNSS data is filtered by an extended Kalman filter to extract the state vector. The extended Kalman filter is dynamically adjusted based on satellite parameters and the noise covariance matrix. Dynamic contours are generated based on the extracted state vectors, and closure detection is performed on the generated contours to obtain the WGS84 coordinate system boundary geometric data of the planting plot. The satellite parameters include the number of satellites, the horizontal accuracy attenuation factor HDOP, and the carrier-noise power density ratio C / N0.

[0024] Based on the WGS84 coordinate system boundary geometry data of the planting plot, the area of ​​the polygon is calculated using the Shoelace formula to obtain the area of ​​the planting plot. The confidence level of the area estimate is then calculated using a multi-parameter confidence model, which is an evaluation model based on the average HDOP value, the variance of the effective satellite number variation, the proportion of signal strength below the threshold, and the walking speed. This model is used to generate the area estimate and its confidence level interval.

[0025] As a preferred option, the planting registration module is used to collect planting information for each farmer's planting plot based on a pre-configured form. The planting information includes the plot name, main crop type, variety, expected yield, planting season, and sowing date.

[0026] Secondly, the present invention provides an agricultural chain supervision method based on trusted identity authentication and land parcel mapping, which realizes agricultural chain supervision through an agricultural chain supervision system based on trusted identity authentication and land parcel mapping as described in any one of the first aspects, comprising the following steps:

[0027] The fingerprint images, facial images, and document images of farmers are collected by handheld terminal devices as authentication images.

[0028] Face recognition is performed based on facial images to obtain facial feature vectors and corresponding confidence scores. Fingerprint recognition is performed based on fingerprint images to obtain fingerprint feature data. OCR recognition is performed on document images to obtain document facial feature vectors, structured document information, and corresponding confidence scores. Farmers are then authenticated based on the facial feature vectors and document facial feature vectors. For authenticated farmers, a farmer identity identifier (UID) is generated based on the farmer's facial feature vector, fingerprint feature data, and structured document information. The farmer's identity information is then constructed based on the farmer's UID, facial feature vector, fingerprint feature data, structured document information, and digital certificate.

[0029] For each certified farmer, the spatial information of their cultivated land is surveyed and collected using a handheld terminal device.

[0030] Metadata and planting information of user plots are collected. Based on the farmer's UID, the farmer's identity information, the spatial information of the planting plot, and the planting information of the planting plot are bound into three-element information. The hash value of the three-element information is calculated and stored in the blockchain. A Merkle tree is constructed based on the hash value of the three-element information. The hash value of the three-element information in the Merkle tree is the leaf value. The metadata includes timestamp, geographic location, device information, and operator information. The timestamp is the time when the planting plot was surveyed by the handheld terminal device. The geographic location is the GPS coordinate information of the handheld terminal device. The device information is the device ID and status of the handheld terminal device.

[0031] The agricultural chain supervision system and method based on trusted identity authentication and land parcel mapping of the present invention have the following advantages:

[0032] 1. By collecting farmers' facial images, fingerprint images, and document images through handheld terminal devices as authentication images, facial feature vectors, fingerprint feature data, and structured document information of farmers can be extracted based on the above authentication images. Farmer identity identifiers UID are generated based on the above three forms of information. Farmer identity information is constructed based on farmers' UID, facial feature vectors, fingerprint feature data, structured document information, and digital certificates, realizing multi-feature fusion farmer identity authentication.

[0033] 2. During face recognition, face images are collected, and the MobileNetV3-Small lightweight model (with only 2.1M parameters and an inference time of 45ms) and the TPM 2.0 hardware security chip are used to achieve an identity authentication accuracy of over 95% in a network-free environment. Compared with traditional online authentication methods, the applicability is improved by more than 80%, filling the technical gap in agricultural entity identity authentication in complex environments.

[0034] 3. Employing an improved extended Kalman filter algorithm (9-dimensional state vector, dynamic noise covariance matrix Q-adjustment) and real-time dynamic contour generation technology, it can adaptively adjust parameters based on environmental conditions such as real-time satellite signal strength (C / N0 threshold 35dB-Hz) and HDOP value (dynamic threshold 3.0). Even under harsh conditions such as forest obstruction (number of satellites <6) and hilly terrain, it can still maintain a mapping accuracy of 1.2-2.8 meters, which is more than 65% higher than the traditional consumer-grade GPS mapping method (accuracy 3-8 meters). The multi-parameter confidence assessment model provides a reliable area estimation confidence interval (typical value ±3-8%) for each plot, and the closure detection accuracy reaches 98.7%, providing scientific and technical support for agricultural land ownership confirmation and supervision.

[0035] 4. Through the collaborative work of the digital authentication module and the planting registration module, a 256-bit SHA-256 hash value is generated using a triplet strong binding algorithm (user ID, plot GUID, planting information). This achieves an immutable and strong association between the agricultural entity's identity and the planting plot. Data integrity is verified through a Merkle tree (the root hash value is stored on the blockchain), constructing a complete and trustworthy data chain of "who - what to plant - where to plant - how much to plant." Compared to the traditional manual reporting method, data accuracy has increased from 65% to 92% (an improvement of 41.5%), and data consistency has increased from 58% to 89% (an improvement of 53.4%). This provides a traceable and tamper-proof data foundation for accurate tax collection, subsidy distribution, and policy formulation.

[0036] 5. It provides an offline, easy-to-use technical solution. The handheld terminal device (IP68 protection rating) is designed to withstand harsh agricultural environments such as high temperature (-20℃ to +70℃), high humidity (95% RH), and strong light (1000 nits brightness of the display screen under sunlight). The MTBF (Mean Time Between Failures) reaches 8760 hours, and the device failure rate is reduced by more than 65% compared with ordinary terminals. The offline storage capacity supports 1000 user authentication records and 500 plots of data. The battery life is 12 hours on a single charge, which meets the needs of a day's work.

[0037] 6. Through digital identity authentication and precise land registration, the efficiency of agricultural supervision is greatly improved. The registration time for a single farmer is reduced from 45 minutes in the traditional way to 12 minutes (an efficiency improvement of 73%), reducing manual verification costs by more than 68%, saving a lot of manpower and resources. The system deployment cost is reduced by 42% compared with the traditional solution, the investment payback period is 18 months, and the modular design supports rapid deployment (3 months deployment cycle in a single province). It has great application potential in remote areas and has important commercial value and social significance in agricultural cooperation projects. Detailed Implementation

[0038] The present invention will be further described below with reference to specific embodiments, so that those skilled in the art can better understand and implement the present invention. However, the embodiments are not intended to limit the present invention. In the absence of conflict, the embodiments of the present invention and the technical features in the embodiments can be combined with each other.

[0039] This invention provides a digital agricultural chain supervision system to solve the technical problem of how to achieve full-process supervision of the agricultural industry chain.

[0040] Example 1:

[0041] This invention discloses an agricultural chain supervision system based on trusted identity authentication and land surveying, comprising a handheld terminal device, a digital authentication module, and a planting registration module.

[0042] The handheld terminal device is used to collect farmers' fingerprint images, facial images, and document images as authentication images. The authentication images are encrypted and sent to the digital authentication module, which supports each farmer in surveying and mapping their planting plots, collecting spatial information of the planting plots, and encrypting and sending the spatial information of the planting plots to the planting registration module.

[0043] In this embodiment, the handheld terminal device is equipped with a TPM chip. TPM 2.0 (Trusted Platform Module 2.0) is a dedicated security chip with a built-in hardware random number generator (HRNG) and encryption processing unit. Simultaneously, the handheld terminal device is equipped with a capacitive sensor and a Gabor filter bank. The capacitive sensor collects fingerprint images from farmers, and the Gabor filter bank enhances the collected fingerprint images to obtain enhanced fingerprint ridge images. These enhanced fingerprint ridge images are then sent as fingerprint images to the digital authentication module. Correspondingly, the digital authentication module performs fingerprint recognition based on the enhanced fingerprint ridge images, extracting fingerprint feature data, including the coordinates, direction, and type information of fingerprint minutiae.

[0044] The technical parameters of the Gabor filter bank are as follows:

[0045] Input: Raw fingerprint image data;

[0046] Filters in 8 directions: covering 0°, 22.5°, 45°, 67.5°, 90°, 112.5°, 135°, and 157.5°;

[0047] Frequency range: 0.1-0.4 (normalized frequency);

[0048] Standard deviation σ: 2.0; Output: Enhanced fingerprint ridge image;

[0049] Function: Enhances the contrast and clarity of fingerprint ridges, suppresses noise and blur, and highlights ridge features in different directions.

[0050] Meanwhile, the handheld terminal device is equipped with a GNSS chip. When surveying and mapping planting plots using the handheld terminal device, the following operations are included:

[0051] (1) GNSS data is obtained by collecting coordinate points, current position and traveled trajectory through a GNSS chip at a predetermined frequency;

[0052] (2) The original GNSS data is filtered by an extended Kalman filter to extract the state vector. Based on the satellite parameters, the extended Kalman filter is dynamically adjusted by the noise covariance matrix. Based on the extracted state vector, dynamic contour generation is performed, and the generated contour is closed to obtain the WGS84 coordinate system boundary geometric data of the planting plot. Among them, the satellite parameters include the number of satellites, the horizontal accuracy attenuation factor HDOP, and the carrier-noise power density ratio C / N0.

[0053] (3) Based on the WGS84 coordinate system boundary geometric data of the planting plot, the area of ​​the polygon is calculated using the Shoelace formula to obtain the area of ​​the planting plot, and the confidence level of the area estimate is calculated through a multi-parameter confidence model. The multi-parameter confidence model is an evaluation model based on the average HDOP value, the variance of the effective number of satellites, the proportion of signal strength below the threshold, and the walking speed, which is used to generate the area estimate and its confidence interval.

[0054] The digital authentication module interacts with handheld terminal devices to receive fingerprint images, facial images, and document images. It performs facial recognition based on the facial image, obtaining a facial feature vector and corresponding confidence level. It also performs fingerprint recognition based on the fingerprint image, obtaining fingerprint feature data. Furthermore, it performs OCR recognition on the document image, obtaining a document facial feature vector, structured document information, and corresponding confidence level. Based on the facial feature vector and document facial feature vector, it authenticates farmers. For authenticated farmers, it generates a farmer identity identifier (UID) based on the farmer's facial feature vector, fingerprint feature data, and structured document information. Finally, it constructs the farmer's identity information based on the farmer's UID, facial feature vector, fingerprint feature data, structured document information, and digital certificate. The user's identity information is encrypted and stored in an identity information database.

[0055] As a specific implementation of the digital authentication module, this module performs image preprocessing on face images based on the Gamma correction algorithm. Through image preprocessing, it performs adaptive brightness adjustment, illumination compensation, and nonlinear transformation to obtain a brightness-corrected face image. This image is then enhanced using the CLAHE algorithm to obtain a contrast-enhanced face image. Using this contrast-enhanced face image as input, a face recognition model performs feature extraction and identity recognition, outputting a face feature vector and confidence score. The face recognition model is a network model built on MobileNetV3-Small, used to extract features from the input image, enhance the extracted features using an attention mechanism, and output a face feature vector and confidence score.

[0056] In a specific implementation, the digital authentication module is used to perform OCR recognition on document images as follows:

[0057] (1) Perform image preprocessing on the document image, including noise reduction, correction, enhancement, text region localization and segmentation, and standardize the image size and format;

[0058] (2) For the preprocessed document images, feature extraction is performed using the ResNet-34 model to convert the two-dimensional image into a one-dimensional sequence, while retaining the spatial location information and outputting the feature vector;

[0059] (3) Based on the output feature vector, the sequence features are decoded into text through the LSTM layer for context understanding, and the fields are classified and structured to output the document face feature vector, structured document information and corresponding confidence.

[0060] The system in this embodiment can provide triple authentication: the system correlates and verifies the results of three authentication methods: facial recognition, fingerprint recognition, and document OCR, as follows:

[0061] (1) Consistency check: Verify the consistency between the photo on the certificate and the on-site facial recognition result;

[0062] (2) Identity binding: Permanently binding fingerprint features with document information;

[0063] (3) Cross-validation: The three authentication methods corroborate each other, improving the reliability of authentication.

[0064] And it provides feature-level fusion, namely:

[0065] (1) Integrate facial feature vectors, fingerprint feature templates, and document text information into a unified identity feature;

[0066] (2) Generate a unique farmer identification identifier (UserID);

[0067] (3) Establish a multimodal biometric database

[0068] And it provides decision-level fusion, namely:

[0069] (1) Confidence scores of the three identification methods;

[0070] (2) A weighted voting mechanism is used to determine the final authentication result;

[0071] (3) Set a threshold strategy to handle identification conflicts.

[0072] The planting registration module interacts with the handheld terminal device to receive spatial information of each farmer's planting plot, collect metadata and planting information of the user's planting plot, bind the farmer's identity information, spatial information of the planting plot and planting information of the planting plot into three-element information based on the farmer's UID, calculate the hash value of the three-element information, store the hash value of the three-element information in the blockchain, and construct a Merkle tree based on the hash value of the three-element information. In the Merkle tree, the hash value of the three-element information is the leaf value. The metadata includes timestamp, geographical location, device information and operator information. The timestamp is the time when the planting plot was surveyed by the handheld terminal device, the geographical location is the GPS coordinate information of the handheld terminal device, and the device information is the device ID and status of the handheld terminal device.

[0073] In practice, the planting registration module is used to collect planting information for each farmer's plot of land based on a pre-configured form. The planting information includes the plot name, main crop type, variety, expected yield, planting season, and sowing date.

[0074] As a specific implementation example of the system in this embodiment, the agricultural technician uses a ruggedized handheld terminal (IP68 protection rating) equipped with an ARM Cortex-A78 processor (2.4GHz), 8GB LPDDR5 memory, and 256GB UFS 3.1 storage, with a built-in TPM 2.0 security chip (supporting RSA-2048 / ECC P-384 key generation). In offline authentication mode, the MobileNetV3-Small face recognition model (model size 2.1MB, inference time 45ms) is launched, with an input image resolution of 640×480. Preprocessing is performed using Gamma correction (γ value adaptively adjusted between 0.6 and 1.8 according to ambient light) and the CLAHE algorithm (clipLimit = 2.0, tileGridSize = 8×8). Fingerprint recognition uses a capacitive sensor (508DPI resolution), with ridge enhancement through an 8-directional Gabor filter bank (frequency range 0.1-0.4, standard deviation σ = 2.0), tolerating a minutiae loss rate of up to 30%. The document OCR module is based on a CRNN architecture. The CNN feature extraction layer uses ResNet-34 (pre-trained weights with fine-tuning), and the LSTM layer has 256 hidden units. The average processing time for a single document is 1.2 seconds. Authentication data is encrypted using AES-256-GCM (256-bit key length) and stored in a secure area (512MB capacity). It supports offline storage of up to 1000 user authentication records.

[0075] After identity authentication, farmers use the terminal to walk along the plot boundary to conduct surveying. The terminal has a built-in multi-frequency GNSS receiver chip (model u-blox ZED-F9P) supporting GPS / GLONASS / BDS / Galileo constellations, supporting L1 / L2 / L5 frequency bands, and collecting coordinate points at a 10Hz sampling frequency. The system runs an improved extended Kalman filter algorithm in real time. The state vector X=[x,y,z,vx,vy,vz,ax,ay,az]T contains nine components: position, velocity, and acceleration. The state transition matrix Φ adopts a constant acceleration model. The noise covariance matrix Q is dynamically adjusted: when the number of satellites is <6, the diagonal element of Q is set to 0.1m2; when HDOP>3, it is set to 0.05m2; and when C / N0<35dB-Hz, it is set to 0.08m2. The observation noise covariance R is adaptively calculated based on signal quality: R = σ0² × (1 + 0.3 × HDOP + 0.02 × (45 - C / N0)), where σ0 = 0.5m. Real-time closure detection algorithm: Closure is determined when the distance between the current position and the starting point is <5m and the directional angles of three consecutive sampling points point to the starting point (angle tolerance ±30°), with an accuracy of 98.7%. The area calculation uses the Shoelace formula, and the confidence assessment model is C = 0.4 × exp(-0.5 × HDOP_avg) + 0.25 × exp(-σ_sat). 2 / 8)+0.2×(1-P_weak)+0.15×exp(-(v_avg-1.2)2 / 0.8). The structured form contains 17 fields, uses a triplet strong binding algorithm to generate 256-bit hash values, data integrity is verified by a Merkle tree, and the spatial index uses an R-tree structure.

[0076] During the planting registration process, the improved extended Kalman filter algorithm (9-dimensional state vector, dynamic noise covariance matrix Q-adjustment) and real-time dynamic contour generation technology in this embodiment can adaptively adjust parameters according to environmental conditions such as real-time satellite signal strength (C / N0 threshold 35dB-Hz) and HDOP value (dynamic threshold 3.0). Even under harsh conditions such as forest obstruction (number of satellites < 6) and hilly terrain, it can maintain a mapping accuracy of 1.2-2.8 meters, which is more than 65% higher than the accuracy of traditional consumer-grade GPS mapping methods (accuracy 3-8 meters). Multi-parameter confidence assessment model:

[0077] C=0.4×exp(-0.5×HDOP_avg)+0.25×exp(-σ_sat 2 / 8)+0.2×(1-P_weak)+0.15×exp(-(v_avg-1.2) 2 / 0.8),

[0078] It provides a reliable area estimation confidence interval (typical value ±3-8%) for each plot of land, with a closure detection accuracy of 98.7%, providing scientific and technical support for the confirmation and supervision of agricultural land rights.

[0079] The system in this embodiment, through the collaborative work of the digital authentication module and the planting registration module, uses a triplet strong binding algorithm (user ID, plot GUID, planting information) to generate a 256-bit SHA-256 hash value, achieving an immutable and strong association between the agricultural entity's identity and the planting plot. Data integrity is verified through a Merkle tree (the root hash value is stored on the blockchain), constructing a complete and trustworthy data chain of "who - what is planted - where is planted - how much is planted." Compared to the traditional manual reporting method, data accuracy has increased from 65% to 92% (an improvement of 41.5%), and data consistency has increased from 58% to 89% (an improvement of 53.4%), providing a traceable and tamper-proof data foundation for accurate tax collection, subsidy distribution, and policy formulation.

[0080] In this embodiment, the digital authentication module and the plant registration module provide an offline working mode. The system is built on an ARM Cortex-A78 processor and a TPM 2.0 security chip, deploying a MobileNetV3-Small lightweight model (2.1M parameters), achieving a 94% recognition accuracy rate among people with dark skin. Fingerprint recognition uses an 8-directional Gabor filter technology, maintaining a 91% recognition rate even with 30% minutiae loss. Document OCR uses a CRNN architecture, achieving a field recognition accuracy of 92.3%, with a single document processing time of only 1.2 seconds. All authentication data is encrypted and stored using AES-256-GCM, supporting offline authentication for 1000 users. The SHA-256 hash algorithm is used to strongly bind UserID, PlotGUID, and PlantInfo into triples, and data integrity is verified using a Merkle tree. The root hash is stored in the blockchain system (TPS≥3500). Data storage employs a layered architecture of Redis+MySQL+HDFS, protected by dual encryption of AES-256-GCM and RSA-2048. Spatial data is managed through R-tree indexes, supporting 17-field structured forms and 36 business rule validations, achieving a data accuracy of 92%, a 41.5% improvement over traditional manual data entry methods.

[0081] Example 2:

[0082] This invention discloses an agricultural chain supervision method based on trusted identity authentication and land parcel mapping. It implements agricultural chain supervision based on the system disclosed in Example 1, and includes the following steps:

[0083] Step S100: Collect the farmer's fingerprint image, facial image, and ID card image as authentication images using a handheld terminal device;

[0084] Step S200: Perform face recognition based on face image to obtain face feature vector and corresponding confidence level; perform fingerprint recognition based on fingerprint image to obtain fingerprint feature data; perform OCR recognition on document image to obtain document face feature vector, structured document information and corresponding confidence level; and authenticate farmers based on face feature vector and document face feature vector. For authenticated farmers, generate farmer identity identifier UID based on farmer's face feature vector, fingerprint feature data and structured document information; and construct farmer identity information based on farmer's UID, face feature vector, fingerprint feature data, structured document information and digital certificate.

[0085] Step S300: For each certified farmer, the spatial information of the plot of land they cultivate is collected by surveying and mapping the plot of land they cultivate through a handheld terminal device.

[0086] Step S400: Collect metadata and planting information of user planting plots. Based on the farmer's UID, bind the farmer's identity information, the spatial information of the planting plot, and the planting information of the planting plot into three-element information. Calculate the hash value of the three-element information, store the hash value of the three-element information in the blockchain, and construct a Merkle tree based on the hash value of the three-element information. The hash value of the three-element information in the Merkle tree is the leaf value. The metadata includes timestamp, geographical location, device information, and operator information. The timestamp is the time when the planting plot was surveyed by the handheld terminal device. The geographical location is the GPS coordinate information of the handheld terminal device. The device information is the device ID and status of the handheld terminal device.

[0087] The above provides a detailed description of the digital agricultural chain supervision system provided by the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. An agricultural chain supervision system based on trusted identity authentication and land mapping, characterized in that, The handheld terminal device, the digital authentication module, and a planting registration module are included. The handheld terminal device is configured to collect fingerprint images, face images, and certificate images of farmers as authentication images, and encrypt and send the authentication images to the digital authentication module to support each farmer to measure and map the farmland and collect spatial information of the farmland, and encrypt and send the spatial information of the farmland to the planting registration module. The digital authentication module interacts with the handheld terminal device to receive the fingerprint images, face images, and certificate images, perform face recognition based on the face images to obtain face feature vectors and corresponding confidence levels, perform fingerprint recognition based on the fingerprint images to obtain fingerprint feature data, and perform OCR recognition on the certificate images to obtain certificate face feature vectors, structured certificate information, and corresponding confidence levels, and authenticate the farmers based on the face feature vectors and the certificate face feature vectors, and for the farmers who pass the authentication, generate a farmer identity identifier UID based on the face feature vectors, the fingerprint feature data, and the structured certificate information of the farmers, and construct the identity information of the farmers based on the UID, the face feature vectors, the fingerprint feature data, and the structured certificate information of the farmers and a digital certificate. The planting registration module interacts with the handheld terminal device to receive the spatial information of the farmland of each farmer, collect metadata and planting information of the farmland of the user, bind the identity information of the farmers, the spatial information of the farmland, and the planting information of the farmland as three-element information based on the UID of the farmers, calculate the hash value of the three-element information, store the hash value of the three-element information to a blockchain, and construct a Merkle tree based on the hash value of the three-element information, wherein the hash value of the three-element information is a leaf value, and the metadata includes a timestamp, a geographic location, device information, and operator information, the timestamp is the time of measuring and mapping the farmland by the handheld terminal device, the geographic location is the GPS coordinate information of the handheld terminal device, and the device information is the device ID and state of the handheld terminal device.

2. The agricultural chain supervision system based on trusted identity authentication and land mapping according to claim 1, characterized in that, The handheld terminal device is configured with a capacitive sensor and a Gabor filter set to collect fingerprint images of farmers through the capacitive sensor, perform image enhancement on the collected fingerprint images through the Gabor filter set to obtain enhanced fingerprint ridge line images, and send the enhanced fingerprint ridge line images as the fingerprint images to the digital authentication module. Correspondingly, the digital authentication module is configured to perform fingerprint recognition based on the enhanced fingerprint ridge line images, and extract fingerprint feature data including fingerprint minutia point coordinates, directions, and type information. 3.The agricultural chain supervision system based on trusted identity authentication and land mapping according to claim 1, characterized in that, The digital authentication module is configured to perform image preprocessing on the face images based on a Gamma correction algorithm, perform brightness self-adaptive adjustment, illumination compensation, and nonlinear transformation through the image preprocessing to obtain brightness-corrected face images, perform image enhancement on the brightness-corrected face images through a CLAHE algorithm to obtain contrast-enhanced face images, and input the contrast-enhanced face images to a face recognition model to perform feature extraction and identity recognition, and output face feature vectors and confidence levels.

4. The agricultural chain supervision system based on trusted identity authentication and land mapping according to claim 3, characterized in that, The face recognition model is a network model constructed based on MobileNetV3-Small, which is used for feature extraction of an input image, enhancement of the extracted features based on an attention mechanism, and output of a face feature vector and a confidence.

5. The agricultural chain regulation system based on trusted identity authentication and land mapping according to claim 1, characterized in that, The identity information of the user is encrypted and stored in an identity information library.

6. The agricultural chain regulation system based on trusted identity authentication and plot mapping according to claim 1, characterized in that, The digital authentication module is configured to perform the following operations on the certificate image: The certificate image is subjected to image preprocessing, including denoising, correction, enhancement, and text region positioning and segmentation operations, and the image size and format are standardized; For the preprocessed certificate image, a ResNet-34 model is used for feature extraction, converting the two-dimensional image into a one-dimensional sequence while retaining the spatial position information, and outputting a feature vector; Based on the output feature vector, the context is understood through an LSTM layer, the sequence features are decoded into text, and field classification and structuring are performed, outputting a certificate face feature vector, structured certificate information, and the corresponding confidence.

7. The agricultural chain regulation system based on trusted identity authentication and plot mapping according to claim 1, characterized in that, The handheld terminal device is configured with a GNSS chip, and when the handheld terminal device is used to survey the planting plot, the following operations are included: The GNSS chip collects coordinate points, the current position, and the traveled trajectory at a predetermined frequency to obtain GNSS data; The original GNSS data is filtered by an extended Kalman filter to extract a state vector, and the extended Kalman filter is dynamically adjusted based on satellite parameters and a noise covariance matrix, and a dynamic contour is generated based on the extracted state vector, and the generated contour is closed detected to obtain the WGS84 coordinate system boundary geometry data of the planting plot, wherein the satellite parameters include the number of satellites, the horizontal dilution of precision factor HDOP, and the carrier-to-noise power density ratio C / N0; Based on the WGS84 coordinate system boundary geometry data of the planting plot, the Shoelace formula is used to calculate the area of the polygon to obtain the area of the planting plot, and a multi-parameter confidence model is used to calculate the confidence of the area estimation, which is an evaluation model based on the average HDOP value, the variance of the effective satellite number, the proportion of signal strength below the threshold, and the walking speed, used to generate the area estimation value and its confidence interval. 8.The agricultural chain supervision system based on trusted identity authentication and land mapping according to claim 1, characterized in that, The planting registration module is configured to collect the planting information of each farmer's planting plot based on a preconfigured form, including the plot name, the main crop type, the variety, the expected yield, the planting season, and the sowing date.

9. An agricultural chain supervision method based on trusted identity authentication and land mapping, characterized in that, An agricultural chain supervision system based on trusted identity authentication and plot surveying is used to implement agricultural chain supervision, including the following steps: The handheld terminal device collects the farmer's fingerprint image, face image, and certificate image as authentication images; Face recognition is performed based on the face image to obtain a face feature vector and a corresponding confidence level, fingerprint recognition is performed based on the fingerprint image to obtain fingerprint feature data, and OCR recognition is performed on the certificate image to obtain a certificate face feature vector, structured certificate information, and a corresponding confidence level. A farmer is authenticated based on the face feature vector and the certificate face feature vector. For a farmer who passes the authentication, a farmer identity identifier UID is generated based on the face feature vector, the fingerprint feature data, and the structured certificate information of the farmer. Identity information of the farmer is constructed based on the UID, the face feature vector, the fingerprint feature data, the structured certificate information, and a digital certificate of the farmer. For each authenticated farmer, a planting plot of the farmer is surveyed and mapped by a handheld terminal device to collect spatial information of the planting plot. Metadata and planting information of a planting plot of a user are collected. Identity information of the farmer, spatial information of the planting plot, and planting information of the planting plot are bound as triadic information based on the UID of the farmer. A hash value of the triadic information is calculated. The hash value of the triadic information is stored in a blockchain. A Merkle tree is constructed based on the hash value of the triadic information. The hash value of the triadic information is a leaf value in the Merkle tree. The metadata includes a timestamp, a geographic position, device information, and operator information. The timestamp is a time at which the planting plot is surveyed and mapped by the handheld terminal device. The geographic position is GPS coordinate information at which the handheld terminal device is located. The device information is a device ID and a state of the handheld terminal device.