An ocr-based digitized full-link sample monitoring method and system

Abstract

The present application relates to a kind of digital full-link sample monitoring method and system based on OCR, belong to sample digital management and control technical field.Wherein, the method includes: the sample characteristics of analysis collection and receiving process, constructs digital sample task scheduling library, and according to the sampling handover sheet, task is executed, and sample task label data is obtained;Sample task full-link monitoring is guided, and the monitoring state parameters of sample label and sample label are input into the full-link monitoring model to analyze sample order monitoring information, and output sample digital full-link monitoring instruction;Through sample label analysis algorithm, sample preparation record in the automatic monitoring process is verified, and sample preparation record is stored into distributed storage by combining index association upload method;In combination with time stamp tamper-proof storage method, the sample preparation record that has been uploaded is prevented from being tampered with and stored, and traceable digital sample preparation log is generated.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of sample digital management and control, and particularly relates to a digital full-link sample monitoring method and system based on OCR. BACKGROUND

[0002] In the field of experimental detection and sample management, the whole process of sample from collection, handover, transportation to sample preparation usually relies on manual recording and paper handover sheet for process control, information transmission is lagged and it is difficult to realize full-link visual tracking. The traditional sample management system only has basic functions of warehouse registration and circulation record, and lacks real-time monitoring of handover nodes, label identification accuracy and sample preparation process status. When the sample quantity is large or involves multiple circulation agencies, problems such as label mismatch, record loss and sample preparation process untraceable are likely to occur, resulting in incomplete sample state chain and monitoring blind area. In addition, the existing OCR recognition system is usually only used for information input, and is not deeply integrated with task scheduling model and sample order control logic, so it cannot establish consistent logical mapping of handover sheet content and subsequent scheduling execution instructions. At the same time, the video recording data in the sample preparation process are usually stored in the original way, without positioning and retrieval capability, and lack effective tamper-proofing mechanism, which is difficult to meet the requirements of compliance audit and digital evidence storage. Therefore, there is an urgent need for a digital full-link sample monitoring method based on OCR data analysis, task scheduling model and link monitoring engine to realize traceable, verifiable and evidence-based control of sample whole process state. SUMMARY

[0003] To solve the above problems in the prior art, the application provides a digital full-link sample monitoring method based on OCR,

[0004] The purpose of the application can be achieved by the following technical instructions:

[0005] S1: obtaining sample sampling data, analyzing sample characteristics in the collection and receiving process based on the sample sampling data, constructing a digital sample task scheduling library, recognizing the sampling handover sheet information of the digital sample task scheduling library through optical character recognition, and issuing execution tasks according to the sampling handover sheet to obtain sample task label data;

[0006] S2: guiding the full-link monitoring of sample tasks based on the sample task label data, and allocating transmission tasks for samples according to the digital sample task scheduling library, constructing a full-link monitoring model combined with the allocated transmission tasks, inputting the monitoring state parameters of the sample label and the sample label into the full-link monitoring model to analyze the sample order monitoring information, and outputting sample digital full-link monitoring instructions;

[0007] S3: Execute the sample digitization full-link monitoring instruction, verify the sample preparation record in the automatic monitoring process through a sample label analysis algorithm, quickly query the split sample opening process of the sample preparation record based on a motion fast-forward image group, and assign a sample number, while storing the sample preparation record into a distributed storage based on an index association upload method;

[0008] S4: According to the storage requirement of the sample preparation video, combine the timestamp tamper-proof storage method, store the uploaded sample preparation record, and based on the edge cache structured weak network environment, generate a traceable digital sample preparation log based on the sample video retention time.

[0009] Specifically, the sample sampling data acquisition method is: based on the full-link monitoring information, extracting the sample number and collection time period in the sampling single, labeling the label information in the sample collection and reception process through optical character recognition, and aligning the extracted sampling data with the timestamp of the sample label in the field, according to the image tilt correction algorithm, the noise suppression of the sampling transfer single information is generated. Structure sample sampling data.

[0010] Specifically, the construction method of the digital sample task scheduling library is:

[0011] Generate a task scheduling primary key by analyzing the handover node information in the sample sampling data, the task scheduling primary key generates a task index number through a hash mapping algorithm, establishes an index mapping table for the task index number, sample number, handover node identifier and transmission path parameters, and obtains a task index basic structure;

[0012] Based on the index mapping table, load the task execution priority, transmission node sequence and fault tolerance time window parameters for each sample scheduling primary key through the task priority calculation rule, and write the above parameters in the form of key-value pairs into the scheduling parameter memory pool, and construct the digital sample task scheduling library.

[0013] Specifically, the optical character recognition technology extracts the text area based on the character row block segmentation algorithm, removes low-confidence character blocks according to the character confidence matrix, and locates the row and column positions of the table structure in the transfer single.

[0014] Specifically, the sample task label data generation method is:

[0015] Based on the sampling handover record, generate a unique identification label for the sample task, assign a dynamic two-dimensional code for each sample using encryption coding technology, embed task metadata, and perform regular encoding preprocessing on the task metadata to form a label encoding field set;

[0016] Trigger a label encoding engine according to the label encoding field set, and respectively check the loading node sequence bits of the sample label and the sample label based on a preset label encoding rule, and obtain a label number sequence by hashing the loading node sequence bits;

[0017] Establish a bidirectional index mapping relationship between the label number sequence and the task scheduling primary key, and write the encoding structure of the sample label and the sample label into a label data cache area based on a key-value pair encapsulation mechanism to generate sample task label data.

[0018] Specifically, the construction method of the full-link monitoring model is:

[0019] Analyze the sample task label data and the assigned transmission task, generate a full-link state matrix according to the sample task label data, and associate and map the transmission node information, execution time window parameters, and sample monitoring state parameters in the sample task scheduling library;

[0020] Based on the link topology structure, the monitoring parameters of the sample task label are taken as inputs to construct a node-link association framework of the full-link monitoring model; the monitoring requirements of the sample task and the available state of the link node are mapped through a time-space task matrix to generate link monitoring data containing monitoring state and task allocation relationship;

[0021] Based on the link monitoring data, a task allocation monitoring logic is established, a link state update logic rule set is generated according to a link latency threshold, and a time window parameter and a sample monitoring state parameter are associated and mapped through an event-state mapping mechanism to establish a full-link monitoring model.

[0022] Specifically, the output method of the sample digital full-link monitoring instruction is:

[0023] According to the state matrix in the full-link monitoring model, an abnormal task node is identified, the link state matrix in the full-link monitoring model is scanned in real time, the current node state of the sample task label is compared with the target node sequence corresponding to the scheduling primary key, and if it is detected that the state flag bit is not completed within a preset time window, an instruction generation event is triggered and an abnormal node index is recorded;

[0024] The abnormal node index is combined with a scheduling priority parameter and a link resource occupation state, an instruction scheduling engine is called to generate an instruction candidate set of the sample verification instruction, and the candidate instructions are prioritized based on a task delay tolerance calculation model to form a monitoring instruction execution sequence;

[0025] The monitoring instruction execution sequence is pushed to the full-link monitoring model through a link scheduling message bus, and a node execution response is detected based on an instruction feedback mechanism; if it is detected that the node state still does not satisfy the expected jump condition of the link state matrix, the unfinished node is re-written into a scheduling waiting queue, and a sample digitized full-link monitoring instruction is output.

[0026] Specifically, the motion fast-forward image group extracts key frame images from the sample preparation video at fixed time intervals, generates a fast-forward index table based on the key frame time sequence, and quickly locates the sample splitting and sealing action time period through a frame skipping comparison algorithm.

[0027] Specifically, the sample label analysis algorithm verifies the sample preparation process method as follows:

[0028] The sample label scanning sequence in the sample preparation process is uniquely verified, a label reading module is called to collect the scanning sequence of the sample splitting label and the sample retaining label, and the scanned label code is aligned with the task scheduling primary key in the digital sample task scheduling library at the field level to generate a temporary label scanning data set;

[0029] The temporary label scanning data set is compared for consistency, and if the label number and the scheduling primary key corresponding relationship are detected to be mismatched or the scanning sequence has a multi-label conflict, a label exception detection event is triggered, and an exception type field and an exception node timestamp are marked;

[0030] The generated label verification exception record set is written into a sample preparation verification data storage area, and the abnormal label number and the corresponding task scheduling primary key are bidirectionally associated through an exception event index mapping table, the abnormal task is marked as a to-be-verified state, and sample preparation verification is triggered.

[0031] Specifically, the index association upload generates a corresponding digital index based on the unique number of each sample, maps the sample preparation record with the digital index, and when the edge cache network is available, triggers the index association upload, and uploads the mapped sample preparation record to the distributed storage library through the index.

[0032] Specifically, the extraction of the digital sample preparation log is based on the tamper-proof evidence of the sample preparation record, and the link operation record and the monitoring event data are analyzed:

[0033] The link operation record data identifies the label information of the sample on the link node through a two-dimensional code scanning technology, records the sequence of the link nodes passed by the sample in combination with a time synchronization mechanism, and generates link operation record data;

[0034] The monitoring event data is used to monitor the running state and task execution of the sample at each link node in real time through a full-link monitoring model, and the monitoring instruction is used to record the link node number and sample number, so that a traceable digital sample preparation log is generated.

[0035] Specifically, an OCR-based digital full-link sample monitoring system comprises the following steps:

[0036] A sample data acquisition module is used to acquire sample sampling data, analyze sample characteristics in the sampling and receiving process based on the sample sampling data, construct a digital sample task scheduling library, recognize sample handover sheet information of the digital sample task scheduling library through optical character recognition, and issue an execution task according to the sample handover sheet to obtain sample task label data.

[0037] A model-driven dispatching module is used to guide full-link monitoring of sample tasks based on the sample task label data, allocate a transmission task for the sample according to the digital sample task scheduling library, construct a full-link monitoring model in combination with the allocated transmission task, input monitoring state parameters of sample splitting labels and sample reservation labels into the full-link monitoring model to analyze sample dispatching monitoring information, and output a sample digital full-link monitoring instruction.

[0038] A sample label verification module is used to execute the sample digital full-link monitoring instruction, verify sample preparation records in an automatic monitoring process through a sample label analysis algorithm, query a split and sealed sample process of the sample preparation records based on a motion fast-forward image group and allocate a sample number, and store the sample preparation records into a distributed storage library in combination with an index association upload method.

[0039] A log cache generation module is used to store evidence of the uploaded sample preparation records according to the needs of sample preparation video storage in combination with a timestamp tamper-proof storage method, and generate a traceable digital sample preparation log based on an edge cache structured weak network environment and a sample preparation video retention time.

[0040] The present application has the following advantages:

[0041] The application realizes automatic structured analysis of sample handover sheet information and generation of scheduling master key by introducing an OCR recognition technology and a digital task scheduling library linkage mechanism, avoids data delay and field missing problems caused by traditional manual input, and significantly improves the accuracy and real-time performance of sample task distribution. By constructing a full-link monitoring model and introducing a sample separation label and a sample reservation label state matrix, the application can realize dynamic state tracking at each node of sample collection, handover, transmission and sample preparation, and solve the technical bottleneck that sample flow links are invisible and cannot be intervened in existing systems. At the same time, the application applies a motion fast-forward image group index method, greatly improves the video retrieval efficiency of the sample preparation process, supports the backtracking of splitting and sealing sample action nodes within seconds, and improves the process audit capability. In the data storage link, the application combines a timestamp writing mechanism to realize tamper-proof storage of sample preparation records, and ensures the compliance and traceability of sample full-process data. Overall, the full-link digital monitoring mechanism constructed by the application significantly improves the automation, security and regulatory transparency of experimental sample management, and has wide application value. BRIEF DESCRIPTION OF DRAWINGS

[0042] For the convenience of those skilled in the art, the present application will be further described below with reference to the accompanying drawings.

[0043] Figure 1 A structure diagram of the present application, a digital full-link sample monitoring method and system based on OCR.

[0044] Figure 2 An execution diagram of the full-link monitoring model in the present application, a digital full-link sample monitoring method and system based on OCR.

[0045] Figure 3 A flowchart of sample preparation video in the present application, a digital full-link sample monitoring method and system based on OCR. DETAILED DESCRIPTION

[0046] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined application purpose, the specific embodiments, structures, features and effects of the present application will be described in detail below with reference to the accompanying drawings and preferred embodiments.

[0047] Please refer to Figure 1 A digital full-link sample monitoring method based on OCR:

[0048] S1: Obtain sample sampling data, analyze sample characteristics in the collection and reception process based on the sample sampling data, construct a digital sample task scheduling library, recognize sample handover sheet information of the digital sample task scheduling library through optical character recognition, and according to the sample handover sheet, execute tasks to obtain sample task label data;

[0049] S2: guiding the full-link monitoring of the sample task based on the sample task label data, and assigning a sample to a transmission task according to the digital sample task scheduling library, constructing a full-link monitoring model in combination with the assigned transmission task, inputting the monitoring state parameters of the sample labeling label and the sample labeling label into the full-link monitoring model to analyze the sample dispatching monitoring information, and outputting a sample digital full-link monitoring instruction;

[0050] S3: executing the sample digital full-link monitoring instruction, verifying the sample preparation record in the automatic monitoring process through a sample label analysis algorithm, rapidly querying the split and sample sealing process of the sample preparation record based on a motion fast-forward image group and assigning a sample number, and storing the sample preparation record into a distributed storage library in combination with an index association upload method;

[0051] S4: according to the storage requirement of the sample preparation video, in combination with a timestamp tamper-proof storage method, storing the uploaded sample preparation record, and based on an edge cache structured weak network environment, retaining the sample preparation video for a certain period of time, and generating a traceable digital sample preparation log.

[0052] Specifically, the sample sampling data acquisition method is: based on the full-link monitoring information, extracting the sample number and the sampling time period in the sampling single, labeling the label information in the sample collection and reception process through optical character recognition, and aligning the extracted sampling data with the timestamp of the sample label in the field, according to the image tilt correction algorithm, the noise suppression of the sampling transfer single information is generated. Structured sample sampling data.

[0053] Specifically, the construction method of the digital sample task scheduling library is:

[0054] The task scheduling primary key is generated by analyzing the transfer node information in the sample sampling data, the task scheduling primary key is generated by a hash mapping algorithm, the task index number is established with the sample number, the transfer node identification and the transmission path parameters to form an index mapping table, and the task index basic structure is obtained.

[0055] Based on the index mapping table, the task execution priority, the transmission node sequence and the fault tolerance time window parameters are loaded for each sample scheduling primary key through the task priority calculation rule, and the above parameters are written into the scheduling parameter memory pool in the form of key-value pairs to construct the digital sample task scheduling library.

[0056] Specifically, the optical character recognition technology extracts the text area based on the character row block segmentation algorithm, removes low confidence character blocks according to the character confidence matrix, and locates the row and column positions of the table structure in the transfer single.

[0057] The digitized full-link sample monitoring system in this embodiment is deployed in a scheduling control platform of a cloud-edge collaborative architecture, as shown in Figure 2 The specific technical stack is as follows:

[0058] OCR recognition service: Tesseract / PaddleOCR / image preprocessing module;

[0059] Message queue: Kafka / Pulsar;

[0060] Real-time scheduling calculation: Flink / Spark Streaming;

[0061] Rule reasoning engine: Drools / OpenL Tablets;

[0062] Distributed object storage: MinIO / Ceph;

[0063] State cache: Redis / Hazelcast;

[0064] Database: PostgreSQL / TimescaleDB;

[0065] Timestamp trusted service: NTP distributed synchronization service;

[0066] The specific implementation process is as follows:

[0067] The elasticity expansion of the scheduling and control components is realized through the Kubernetes / K3s cluster. The data collection and task distribution channel realizes the asynchronous flow of the OCR recognition task and the link monitoring instruction based on the Kafka / Pulsar message queue, so as to ensure the stability of data transmission in a high-concurrency scenario. As shown in Figure 1 The OCR recognition engine is deployed as an independent recognition microservice, which uses the Tesseract OCR or PaddleOCR engine framework and combines image preprocessing algorithms (tilt correction, Gaussian filtering, and character region segmentation) to realize the structured analysis of the sample delivery order.

[0068] The full-link state monitoring and task scheduling logic runs in the Flink Streaming / Spark StructuredStreaming real-time calculation framework, which is used for dynamic calculation and update of the label state matrix and the link node time window, as shown in Figure 3The rule reasoning module shown adopts the Drools rule engine / OpenL Tablets to determine task priority, trigger monitoring instructions, and match abnormal scheduling fallback strategies. The distributed storage part in the system is composed of a MinIO / Ceph object storage cluster, which is used to carry video streams and fast-forward image index data in the sample preparation process. At the same time, structured logs and sample preparation record data are stored through PostgreSQL + TimescaleDB, where TimescaleDB is used for high-frequency link state data time series writing query optimization.

[0069] Redis / Hazelcast memory computing services are deployed on edge nodes to cache sample task label mapping tables and node state snapshot data. In terms of tamper-proof evidence, the system uses SHA-256 / SM3 hash algorithms to calculate the digest of sample log blocks, and uses NTP time synchronization services to write timestamps to ensure log credibility.

[0070] Specifically, the method for generating sample task label data is:

[0071] Based on the sample handover record, a unique identification label for the sample task is generated, and an encrypted coding technology is used to assign a dynamic two-dimensional code to each sample, embed task metadata, and perform regular coding preprocessing on the task metadata to form a label coding field set;

[0072] According to the label coding field set, a label coding engine is triggered, and based on the pre-set label coding rules, the split sample label and the reserved sample label loading node sequence bits are respectively verified, and the loading node sequence bits are hashed to obtain a label number sequence;

[0073] The label number sequence and the task scheduling primary key establish a bidirectional index mapping relationship, and based on the key-value pair encapsulation mechanism, the coding structure of the split sample label and the reserved sample label is written into the label data cache area to generate sample task label data.

[0074] Specifically, the method for constructing the full-link monitoring model is:

[0075] The sample task label data and the assigned transmission task are analyzed, a full-link state matrix is generated according to the sample task label data, and the transmission node information, execution time window parameters, and sample monitoring state parameters in the sample task scheduling library are associated and mapped;

[0076] Based on the link topology structure, the monitoring parameters of the sample task label are taken as input to construct a node-link association framework of the full-link monitoring model; the monitoring requirements of the sample task and the available state of the link node are mapped through a time-space task matrix to generate link monitoring data containing monitoring state and task allocation relationship;

[0077] A task allocation monitoring logic is established based on the link monitoring data, a link state update logic rule set is generated according to a link latency threshold, and a time window parameter and a sample monitoring state parameter are associated and mapped through an event-state mapping mechanism to establish a full-link monitoring model.

[0078] In this embodiment, the system introduces a time-state offset detection model for the sample in the collection, handover, transmission and sample preparation node flow process, which is used to judge whether the sample completes node transition within the preset time window. Assuming that the sample task in a certain link node Node i should execute the time window as follows:

[0079] ,

[0080] The offset of the actual scanning record time of the sample from the time window is defined as:

[0081] ,

[0082] If Δt i ≦ε, where ε is a link fault tolerance threshold (calculated dynamically according to the task priority level), the current node state is judged to be normal jump; otherwise, the instruction re-distribution mechanism is triggered, the task is marked as Delay-Pending state and written into the delay queue.

[0083] The link fault tolerance threshold ε is generated by the following formula:

[0084] ,

[0085] Where, T hist is the average passing time of the sample of this type in the historical data of the corresponding node, σ T is the standard deviation of the passing time, a and b are coefficients adaptively allocated according to the sample risk level, and a lower tolerance is set for high-risk samples.

[0086] The scheduling re-distribution trigger condition formula is:

[0087] ,

[0088] The task entering the secondary scheduling will be written into the scheduling priority queue Q again and reordered according to the following priority calculation function:

[0089] ,

[0090] wherein W task is a sample task level weight, λ is a delay penalty coefficient for dynamically lifting the scheduling priority of overdue tasks, Q is a scheduling queue, P riority is a scheduling priority value.

[0091] Specifically, the output method of the sample digital full-link monitoring instruction is:

[0092] According to the state matrix in the full-link monitoring model, an abnormal task node is identified, the link state matrix in the full-link monitoring model is scanned in real time, the current node state of the sample task label is compared with the target node sequence corresponding to the scheduling primary key, if it is detected that the state flag bit is not completed within the preset time window, an instruction generation event is triggered and the abnormal node index is recorded;

[0093] The abnormal node index is combined with the scheduling priority parameter and the link resource occupation state, an instruction scheduling engine is called to generate an instruction candidate set of sample preparation verification instructions, and the candidate instructions are prioritized based on a task delay tolerance calculation model to form a monitoring instruction execution sequence;

[0094] The monitoring instruction execution sequence is pushed to the full-link monitoring model through a link scheduling message bus, and a node execution response is detected based on an instruction feedback mechanism, if it is detected that the node state still does not satisfy the expected jump condition of the link state matrix, the unfinished node is re-written into a scheduling waiting queue, and the sample digital full-link monitoring instruction is output.

[0095] Specifically, the motion fast-forward image group extracts key frame images from the sample preparation video at fixed time intervals, and generates a fast-forward index table based on the key frame time sequence, and quickly locates and splits the sample preparation and sample sealing action time period through a frame skipping comparison algorithm.

[0096] Specifically, the method for verifying the sample preparation process by the sample label analysis algorithm is:

[0097] The uniqueness of the sample label scanning sequence in the sample preparation process is verified, a label reading module is called to collect the scanning sequence of the sample label and the sample label, and the label code obtained by scanning is aligned with the task scheduling primary key in the digital sample task scheduling library at the field level, to generate a temporary label scanning data set;

[0098] The temporary label scanning data set is compared for consistency, if it is detected that the label number and the scheduling primary key corresponding relationship do not match or the scanning sequence has a multi-label conflict, a label abnormality detection event is triggered, and the abnormal type field and the abnormal node timestamp are marked;

[0099] The generated label check exception record set is written into the sample preparation check data storage area, the exception label number is bidirectionally associated with the corresponding task scheduling primary key through the exception event index mapping table, the exception task is marked as a to-be-verified state, and sample preparation checking is triggered.

[0100] Specifically, the index association upload generates a corresponding digital index based on the unique number of each sample, maps the sample preparation record to the digital index, triggers index association upload when the edge cache network is available, and uploads the mapped sample preparation record to the distributed storage through the index.

[0101] Specifically, the extraction of the digitized sample preparation log is based on the tamper-proof evidence sample preparation record, and the link operation record and monitoring event data are analyzed:

[0102] The link operation record data identify the label information of the sample on the link node through the two-dimensional code scanning technology, record the sequence of the link nodes passed by the sample in combination with the time synchronization mechanism, and generate the link operation record data;

[0103] The monitoring event data real-time monitor the running state and task execution of the sample at each link node through the full-link monitoring model, and record the link node number and sample number in combination with the monitoring instruction execution log, to generate a traceable digitized sample preparation log.

[0104] Specifically, an OCR-based digitized full-link sample monitoring system, characterized by comprising:

[0105] A sample data acquisition module: acquires sample sampling data, analyzes the sample characteristics in the acquisition and reception process based on the sample sampling data, constructs a digitized sample task scheduling library, recognizes the sampling handover sheet information of the digitized sample task scheduling library through optical character recognition, and according to the sampling handover sheet, executes the task to obtain sample task label data;

[0106] A model-driven dispatching module: guides the full-link monitoring of the sample task based on the sample task label data, allocates a transmission task for the sample according to the digitized sample task scheduling library, constructs a full-link monitoring model in combination with the allocated transmission task, inputs the monitoring state parameters of the split sample label and the reserved sample label into the full-link monitoring model to analyze the sample dispatching monitoring information, and outputs a sample digitized full-link monitoring instruction;

[0107] A sample preparation label check module: executes the sample digitized full-link monitoring instruction, checks the sample preparation record in the automatic monitoring process through a sample label analysis algorithm, rapidly queries the split and sealed sample process of the sample preparation record based on a motion fast-forward image group and allocates a sample number, and simultaneously stores the sample preparation record into a distributed storage in combination with an index association upload method;

[0108] The log cache generation module: according to the demand of sample preparation video storage, combined with the time stamp tamper-proof evidence method, the sample preparation record uploaded is stored, and based on the edge cache structured weak network environment, the sample preparation video retention time is generated. The digital sample preparation log can be traced.

[0109] The above is only the preferred embodiment of the present application, not any form of limitation on the present application, although the present application has been disclosed as above, however, not to limit the present application, any person skilled in the art, within the scope of the present application, can make some changes or modifications of the above disclosed technology content as equivalent embodiments, but as long as it does not deviate from the technical content of the present application, any simple modification, equivalent change and modification of the above embodiment according to the technical essence of the present application, still belongs to the scope of the present application.

Claims

1. A digital end-to-end sample monitoring method based on OCR, characterized in that, include: S1: Obtain sample sampling data, analyze the sample characteristics of the collection and reception process based on the sample sampling data, construct a digital sample task scheduling library, identify the sampling handover information of the digital sample task scheduling library through optical character recognition technology, and issue execution tasks according to the sampling handover, thereby obtaining sample task tag data. The method for constructing the digital sample task scheduling library is as follows: A task scheduling primary key is generated by analyzing the handover node information in the sample sampling data. The task scheduling primary key is used to generate a task index number through a hash mapping algorithm. An index mapping table is established by the task index number, sample number, handover node identifier and transmission path parameters to obtain the basic structure of the task index. Based on the index mapping table, the task execution priority, transmission node sequence and fault tolerance time window parameters are loaded for each sample scheduling primary key according to the task priority calculation rules, and the above parameters are written into the scheduling parameter memory pool in the form of key-value pairs to construct a digital sample task scheduling library. The method for generating the sample task label data is as follows: A unique identifier tag for the sample task is generated based on the sampling handover record. A dynamic QR code is assigned to each sample using encryption encoding technology, task metadata is embedded, and the task metadata is preprocessed with regular encoding to form a tag encoding field set. The tag encoding engine is triggered based on the tag encoding field set, and the loading node sequence bits of the sampled tags and the retained tags are verified based on the preset tag encoding rules. The tag number sequence is obtained by hashing the loading node sequence bits. A bidirectional index mapping relationship is established between the label number sequence and the task scheduling primary key, and the encoding structure of the sampling label and the retention label is written into the label data cache based on the key-value pair encapsulation mechanism to generate sample task label data. S2: Guide the end-to-end monitoring of sample tasks based on the sample task tag data, allocate transmission tasks to samples according to the digital sample task scheduling library, construct an end-to-end monitoring model in combination with the allocated transmission tasks, input the monitoring status parameters of the sample distribution tag and the retention tag into the end-to-end monitoring model to analyze the sample dispatch monitoring information, and output the sample digital end-to-end monitoring instruction. The method for constructing the end-to-end monitoring model is as follows: The sample task tag data and the assigned transmission tasks are analyzed, and a full-link state matrix is ​​generated based on the sample task tag data. The transmission node information, execution time window parameters and sample monitoring status parameters in the sample task scheduling library are associated and mapped. Based on the link topology, the monitoring parameters of the sample task labels are used as input to construct the node-link association framework of the full-link monitoring model; the monitoring requirements of the sample tasks are mapped to the available status of the link nodes through the time-space task matrix, and link monitoring data containing monitoring status and task allocation relationship is generated. Based on the link monitoring data, a task allocation monitoring logic is established, a link status update logic rule set is generated according to the link waiting time threshold, and the time window parameters and sample monitoring status parameters are associated and mapped through the event-state mapping mechanism to establish a full-link monitoring model. S3: Execute the sample digitization full-link monitoring instruction, verify the sample preparation record during the automated monitoring process through the sample label analysis algorithm, quickly query the sample preparation record splitting and sealing process based on the motion fast forward image group and assign sample number, and at the same time store the sample preparation record into the distributed repository by combining the index association upload method. The method for verifying the sample preparation process using the sample label analysis algorithm is as follows: The uniqueness of the sample label scanning sequence during the sample preparation process is verified. The label reading module is called to collect the scanning sequence of the sample splitting label and the sample retention label. The obtained label code is aligned with the task scheduling primary key in the digital sample task scheduling library at the field level to generate a temporary label scanning dataset. A consistency comparison is performed on the temporary label scanning dataset. If a mismatch is detected between the label number and the scheduling primary key or multiple label conflicts are detected in the scanning sequence, a label anomaly detection event is triggered, and the anomaly type field and the anomaly node timestamp are marked. Write the generated label verification anomaly record set into the sample preparation verification data storage area, and use the anomaly event index mapping table to bidirectionally associate the anomaly label number with the corresponding task scheduling primary key, mark the anomaly task as pending verification status, and trigger sample preparation verification. S4: Based on the requirements for storing sample preparation videos, and combined with the timestamp anti-tampering evidence preservation method, the uploaded sample preparation records are preserved, and a traceable digital sample preparation log is generated based on the retention time of sample preparation videos in the edge-cached structured weak network environment.

2. The method according to claim 1, characterized in that, The method for obtaining the sample sampling data is as follows: extract the sample number and collection time period from the sampling handover form based on the end-to-end monitoring information, mark the label information in the sample collection and reception process through optical character recognition, align the extracted sampling data with the timestamp of the sample label, suppress noise in the sampling handover form information according to the image tilt correction algorithm, and generate structured sample sampling data.

3. The method according to claim 1, characterized in that, The optical character recognition technology extracts text regions based on a character row block segmentation algorithm, eliminates low-confidence character blocks according to the character confidence matrix, and locates the row and column positions of the table structure in the handover form.

4. The method according to claim 1, characterized in that, The method for outputting the sample digitization end-to-end monitoring command is as follows: Abnormal task nodes are identified based on the state matrix in the full-link monitoring model. The link state matrix in the full-link monitoring model is scanned in real time. The current node status of the sample task label is compared with the target node sequence corresponding to the scheduling primary key. If the status flag bit is detected to have not completed the transition within the preset time window, an instruction generation event is triggered and the abnormal node index is recorded. The abnormal node index, combined with the scheduling priority parameter and the link resource occupancy status, calls the instruction scheduling engine to generate a candidate set of sample verification instructions, and sorts the candidate instructions by priority based on the task delay tolerance calculation model to form a monitoring instruction execution sequence. The monitoring instruction execution sequence is pushed to the full-link monitoring model through the link scheduling message bus, and the node execution response is detected based on the instruction acknowledgment mechanism. If the node status is detected as not meeting the expected jump conditions of the link status matrix, the incomplete node is rewritten into the scheduling waiting queue, and the sample digitization full-link monitoring instruction is output.

5. The method according to claim 2, characterized in that, The motion fast-forward image group extracts key frame images from the sample preparation video at fixed time intervals, and generates a fast-forward index table based on the key frame time series. The time period of the splitting and sealing action is quickly located by the frame skipping comparison algorithm.

6. The method according to claim 1, characterized in that, The index-associated upload generates a corresponding digital index based on the unique number of each sample, maps the sample preparation record to the digital index, and triggers the index-associated upload when the edge caching network is available, uploading the mapped sample preparation record to the distributed repository through the index.

7. The method according to claim 1, characterized in that, The extraction of the digital sample preparation log is based on tamper-proof and evidence-based sample preparation records, and analyzes the link operation records and monitoring event data: The link operation record data uses QR code scanning technology to identify the tag information of the sample on the link node, and combines a time synchronization mechanism to record the sequence of link nodes traversed by the sample to generate link operation record data; The monitoring event data is monitored in real time by the full-link monitoring model to monitor the running status and task execution of the sample at each link node. Combined with the monitoring instruction execution log, the link node number and sample number are recorded to generate a traceable digital sample preparation log.

8. A digital end-to-end sample monitoring system based on OCR, characterized in that, include: Sample data acquisition module: acquires sample sampling data, analyzes sample characteristics during the acquisition and reception process based on the sample sampling data, constructs a digital sample task scheduling library, identifies the sampling handover information of the digital sample task scheduling library through optical character recognition technology, issues execution tasks according to the sampling handover, and obtains sample task tag data. The method for constructing the digital sample task scheduling library is as follows: A task scheduling primary key is generated by analyzing the handover node information in the sample sampling data. The task scheduling primary key is used to generate a task index number through a hash mapping algorithm. An index mapping table is established by the task index number, sample number, handover node identifier and transmission path parameters to obtain the basic structure of the task index. Based on the index mapping table, the task execution priority, transmission node sequence and fault tolerance time window parameters are loaded for each sample scheduling primary key according to the task priority calculation rules, and the above parameters are written into the scheduling parameter memory pool in the form of key-value pairs to construct a digital sample task scheduling library. The method for generating the sample task label data is as follows: A unique identifier tag for the sample task is generated based on the sampling handover record. A dynamic QR code is assigned to each sample using encryption encoding technology, task metadata is embedded, and the task metadata is preprocessed with regular encoding to form a tag encoding field set. The tag encoding engine is triggered based on the tag encoding field set, and the loading node sequence bits of the sampled tags and the retained tags are verified based on the preset tag encoding rules. The tag number sequence is obtained by hashing the loading node sequence bits. A bidirectional index mapping relationship is established between the label number sequence and the task scheduling primary key, and the encoding structure of the sampling label and the retention label is written into the label data cache based on the key-value pair encapsulation mechanism to generate sample task label data. Model-driven dispatch module: Based on the sample task tag data, it guides the end-to-end monitoring of sample tasks, allocates transmission tasks to samples according to the digital sample task scheduling library, constructs an end-to-end monitoring model in combination with the allocated transmission tasks, inputs the monitoring status parameters of the sample distribution tag and the retention tag into the end-to-end monitoring model to analyze the sample dispatch monitoring information, and outputs digital end-to-end monitoring instructions for samples. The method for constructing the end-to-end monitoring model is as follows: The sample task tag data and the assigned transmission tasks are analyzed, and a full-link state matrix is ​​generated based on the sample task tag data. The transmission node information, execution time window parameters and sample monitoring status parameters in the sample task scheduling library are associated and mapped. Based on the link topology, the monitoring parameters of the sample task labels are used as input to construct the node-link association framework of the full-link monitoring model; the monitoring requirements of the sample tasks are mapped to the available status of the link nodes through the time-space task matrix, and link monitoring data containing monitoring status and task allocation relationship is generated. Based on the link monitoring data, a task allocation monitoring logic is established, a link status update logic rule set is generated according to the link waiting time threshold, and the time window parameters and sample monitoring status parameters are associated and mapped through the event-state mapping mechanism to establish a full-link monitoring model. Sample preparation label verification module: executes the sample digitization full-link monitoring instructions, verifies the sample preparation records during the automated monitoring process through sample label analysis algorithm, quickly queries the sample preparation record splitting and sealing process based on motion fast forward image group and assigns sample number, and stores the sample preparation record into a distributed repository using index association upload method. The method for verifying the sample preparation process using the sample label analysis algorithm is as follows: The uniqueness of the sample label scanning sequence during the sample preparation process is verified. The label reading module is called to collect the scanning sequence of the sample splitting label and the sample retention label. The obtained label code is aligned with the task scheduling primary key in the digital sample task scheduling library at the field level to generate a temporary label scanning dataset. A consistency comparison is performed on the temporary label scanning dataset. If a mismatch is detected between the label number and the scheduling primary key or multiple label conflicts are detected in the scanning sequence, a label anomaly detection event is triggered, and the anomaly type field and the anomaly node timestamp are marked. Write the generated label verification anomaly record set into the sample preparation verification data storage area, and use the anomaly event index mapping table to bidirectionally associate the anomaly label number with the corresponding task scheduling primary key, mark the anomaly task as pending verification status, and trigger sample preparation verification. Log caching generation module: Based on the requirements for storing sample preparation videos, and combined with the timestamp anti-tampering evidence storage method, the module stores the uploaded sample preparation records and generates a traceable digital sample preparation log based on the retention time of sample preparation videos in the edge caching structured weak network environment.

Images