Method and system for digital tracking of the entire cycle of production of LCD display screens

Abstract

The present application relates to the technical field of manufacturing digitization, in particular to a LCD display production full-cycle digital traceability method and system, comprising the following steps: obtaining operation state and heat source time, screening stress reversal and state continuous section, extracting action number to construct jump section associated node set, identifying image abnormal position and structure mutation, establishing image and operation path correspondence, extending label associated action paragraph, and obtaining label path traceability structure set. In the present application, through the time sequence matching of stress change and heat source switching, the state jump section in the process is extracted, the behavior chain is constructed combined with the continuous connection relationship of action number, the structure fracture form and bright spot distribution direction in the image are associated, the corresponding path between the image content and the operation process is established, the operation paragraph and the interleaved segment covered by the defect label are expanded, the dynamic linkage of image performance, process behavior and time sequence is formed, and the abnormal characteristics are traced and reproduced in the process level.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of manufacturing digitalization, in particular to a LCD display screen production full-cycle digital traceability method and system. BACKGROUND

[0002] The technical field of manufacturing digitalization involves integrated management of information flow and logistics in the production process, and its core matters include data collection, process tracking, resource scheduling, quality management and decision support in each stage of the product life cycle. The technical path usually relies on industrial internet, internet of things, edge computing, digital twin and enterprise resource planning to build a full-process digital system covering design, processing, assembly, testing, logistics and after-sales, so as to realize real-time monitoring and intelligent management of the whole production process. Among them, the traditional LCD display screen production full-cycle digital traceability method refers to the establishment of a product identification system based on bar code or radio frequency identification in the process of manufacturing liquid crystal display screens, in order to realize the whole-process identification and association of information such as key components, processes, equipment and operators. The production process is recorded by cooperating with the manufacturing execution system, and the centralized storage and management of data at each link are completed by combining the database system, so as to realize the step-by-step tracking and structured storage of core data such as raw material source, process parameter, quality inspection record and inventory status. Related methods generally use two-dimensional code laser engraving, radio frequency identification label pasting, sensor data acquisition and manufacturing execution system linkage recording to complete the process.

[0003] In the prior art, data association is performed relying on bar code identification and system recording. In the scene of multi-process parallel or behavior short cycle overlap, there is a lack of discrimination mechanism for continuous change of operation behavior state, and the time sequence breakpoints in the key process are easily missed, resulting in jumps and gaps in the data chain, and a lack of dynamic correspondence between images and behavior data, which cannot map the actual process position and time section of image abnormalities. For example, when the stress changes dramatically and the heat source switches quickly, the related operation number cannot be associated with the image node in time, resulting in problems such as unclear defect cause identification and incomplete label traceability link, which limits the in-depth analysis of the source of image defects and the accurate restoration of the behavior path. SUMMARY

[0004] In order to solve the technical problems existing in the prior art, the embodiments of the present application provide a LCD display screen production full-cycle digital traceability method.

[0005] In order to achieve the above-mentioned purpose, the present application adopts the following technical scheme, the LCD display screen production full-cycle digital traceability method comprises the following steps:

[0006] S1: Obtain the initial operation state of the panel, collect the heat source switching time and the heated starting point, locate the continuously reversed position in the stress curve, select the processing section with time overlap and continuous state according to the front and rear conveying and heating behavior, and obtain the process jump section list;

[0007] S2: Based on the process jump section list, extract the corresponding action number from the scheduling record, identify the action sequence, eliminate the interrupt and jump items, extract the adjacent number section before and after, and obtain the jump section association node set;

[0008] S3: Based on the jump section association node set, extract the structure change position in the image, analyze the fracture line direction, bright area distribution and gray scale fluctuation characteristics, correspond to the panel number, identify the position matching relationship, extract the structure mutation segment of the image sequence, and obtain the defect label structure list;

[0009] S4: Based on the defect label structure list, extract the image acquisition time action number, select the time continuous sequence item, connect the image content and operation number, and obtain the path mapping table corresponding to the label;

[0010] S5: Based on the path mapping table corresponding to the label, extract the image number and operation sequence, extend the label associated action paragraph, extract the connection path in the covering relationship, and obtain the label path tracing structure set.

[0011] As a further scheme of the present application, the process jump section list includes the heat source switching time point, the initial heating time point, the stress curve direction reversal position, the operation state transformation period, the heating and conveying behavior track, the jump section association node set includes the action number sequence, the time interval mapping, the continuous connection node, the behavior chain structure, the defect label structure list includes the structure change position, the boundary fracture line direction, the bright spot distribution form, the gray scale jump area, the bright area distribution range, the image number position matching, the path mapping table corresponding to the label includes the image acquisition time, the action number interval, the image and action sequence mapping, the image performance and time sequence comparison, and the label path tracing structure set includes the operation number sequence, the image time sequence, the label number sorting, the action covering paragraph, and the interleaved section.

[0012] As a further scheme of the present application, the continuously reversed position refers to the section in which the stress direction between adjacent measuring points continuously reverses in the stress change curve, indicating the dynamic change characteristics of the material under the action of heat and stress;

[0013] The jump item refers to the number section with interruption, non-triggering and time span anomaly in the action number and time sequence.

[0014] As a further scheme of the present application, the fracture line direction refers to the direction of line fracture and morphological mutation occurring at the bright-dark boundary edge in the quality inspection image, and the structural anomaly and crack propagation trend are identified.

[0015] The gray scale fluctuation feature refers to the region in which the gray value in the image continuously jumps and the gradient changes in space.

[0016] As a further scheme of the present application, the specific steps of S1 are:

[0017] S101: Obtain the working state of the panel after entering the initial stage, collect the time corresponding to the heat source switching signal, extract the heat receiving starting point of the panel, compare the time interval between the heat source signal and the panel heat receiving, extract the time sequence and construct a continuous sequence to obtain the heat receiving time starting sequence;

[0018] S102: Based on the heat receiving time starting sequence, extract the stress change curve under the synchronous time axis, search for the segment with continuous occurrence direction reversal, extract the position segment with alternating slope between adjacent points, and filter the change sequence according to time continuity to obtain the stress direction reversal segment sequence;

[0019] S103: Based on the stress direction reversal segment sequence, extract the temperature control and conveying action numbers on both sides of the corresponding segment, compare the overlapping range in the action duration time, filter the segment with continuous relationship in the action overlapping part, extract the action corresponding number, and obtain the process jump segment list.

[0020] As a further scheme of the present application, the specific steps of S2 are:

[0021] S201: Based on the jump period in the process jump segment list, extract the action number in the corresponding time from the dispatching table, extract the start and end time of the number, analyze the trigger sequence between adjacent numbers, filter the number segment adjacent in time, extract the number set connected in sequence, and obtain the action time connection sequence set;

[0022] S202: Based on the action time connection sequence set, search for the time interval value between the number sequence, identify the data segment without triggering and with interruption, remove the segment with time span and disconnected before and after the number, extract the number set with continuous time and no interruption, and obtain the continuous execution number sequence;

[0023] S203: Based on the continuous execution number sequence, extract the trigger time difference value between the numbers, analyze the continuation relationship order between actions, select the number chain adjacent in time, extract the start and end mapping nodes of each number, and obtain the jump segment association node set.

[0024] As a further scheme of the present application, the specific steps of S3 are:

[0025] S301: based on the hopping section association node set, the quality inspection image of the corresponding panel is extracted, the structural change area in the monitoring image is extracted, the edge line and the fracture trend of the bright-dark boundary are extracted, the image block with morphological mutation is identified, the boundary contour offset trend in the continuous frame is compared, the pixel set with change characteristics is extracted, and a structural change characteristic matrix is obtained;

[0026] S302: based on the structural change characteristic matrix, the spatial gradient value of the brightness distribution is extracted, the trend and diffusion range of the bright spot area are analyzed, the concentrated area of the gray scale continuous jump is identified, the bright area distribution and the gray scale gradient information are compared, the edge mutation area block is extracted, and a brightness and gray scale associated segment set is obtained.

[0027] S303: based on the brightness and gray scale associated segment set, the panel number corresponding to each image is extracted, the corresponding relationship of the number and the image position in the time sequence is analyzed, the bright area distribution and the number mapping section are retrieved, the morphological characteristics in the image sequence are compared with the number, and a defect label structure list is obtained.

[0028] As a further scheme of the application, the specific steps of S4 are:

[0029] S401: based on the defect label structure list, the time period corresponding to image acquisition in the panel execution log is collected, the number interval where the time period is located in the log is extracted, the start and end time of the number in the number interval is retrieved, the number corresponding to the continuous acquisition time is screened, and an image time corresponding number sequence is obtained.

[0030] S402: based on the image time corresponding number sequence, the numbers are arranged according to the time sequence, the adjacent action content in the continuous number section is extracted, the continuity logic between the actions before and after the same number is identified, the image acquisition action and the adjacent operation behavior are concatenated, and an image action continuity segment set is obtained.

[0031] S403: based on the image action continuity segment set, the corresponding information of the image number and the action sequence number in the log is extracted, the image content and the behavior segment are aligned according to the time direction sequence, the action path range corresponding to each image is delimited, and a label corresponding path mapping table is obtained.

[0032] As a further scheme of the application, the specific steps of S5 are:

[0033] S501: based on the label corresponding path mapping table, the operation number set associated with the image sequence number is extracted, the image label numbers in the same batch are expanded, the corresponding numbers of the differential images in the number set are sequentially connected in time sequence, and a label number extension chain group is obtained.

[0034] S502: Based on the label number extension chain group, retrieve the position of the number in the operation chain, select the image sequence corresponding to the adjacent number in the time sequence, concatenate the logical connection relationship between the images in each group of label number chain, and obtain the image action position sequence set;

[0035] S503: Based on the image action position sequence set, position the front and back distribution range of the image number in the operation behavior, aggregate the operation paragraph and the staggered position section covered by the label number, and obtain the label path tracing structure set.

[0036] An LCD display screen production full-cycle digital tracing system comprises:

[0037] The initial state recognition module acquires the panel initial stage operation state, collects the heat source switching time and the heated starting point, positions the continuous position of the direction reversal in the stress curve, selects the processing section with time overlap and continuous state according to the front and back conveying and heating behavior, and obtains the process jump section list.

[0038] The scheduling behavior extraction module extracts the action number in the corresponding time in the scheduling table based on the jump period in the process jump section list, identifies the action sequence, eliminates the interruption and jump items, extracts the adjacent number section in sequence, and obtains the jump section association node set.

[0039] The defect feature extraction module extracts the structure change position in the image based on the jump section association node set, analyzes the boundary fracture line direction, bright area distribution and gray scale fluctuation characteristics, corresponds to the panel number, identifies the position matching relationship, extracts the image sequence structure mutation segment, and obtains the defect label structure list.

[0040] The action image association module extracts the image acquisition time action number based on the defect label structure list, selects the time continuous sequence item, connects the image content and the operation number, and obtains the path mapping table corresponding to the label.

[0041] The path chain tracing module extracts the image number and operation sequence based on the path mapping table corresponding to the label, extends the label associated action paragraph, extracts the connection path in the covering relationship, and obtains the label path tracing structure set.

[0042] Compared with the prior art, the advantages and positive effects of the present application are:

[0043] In the present application, by matching the timing of stress change and heat source switching, the state jump section in the process is extracted, the action number continuous connection relationship is combined to construct the behavior chain, the structure fracture morphology and bright spot distribution direction in the image are associated, the corresponding path between the image content and the operation process is established, the operation paragraph and the interleaved segment covered by the defect label are expanded, the dynamic linkage of image expression, process behavior and time sequence is formed, and the abnormal characteristics are promoted to trace and reproduce the path at the process level. BRIEF DESCRIPTION OF DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0045] Figure 1 The step flowchart of the present application is shown in the figure.

[0046] Figure 2 The S1 refinement schematic diagram of the present application is shown in the figure.

[0047] Figure 3 The S2 refinement schematic diagram of the present application is shown in the figure.

[0048] Figure 4 The S3 refinement schematic diagram of the present application is shown in the figure.

[0049] Figure 5 The S4 refinement schematic diagram of the present application is shown in the figure.

[0050] Figure 6 The S5 refinement schematic diagram of the present application is shown in the figure.

[0051] Figure 7 The system module diagram of the present application is shown in the figure. DETAILED DESCRIPTION

[0052] The technical solutions in the present application will be described below in combination with the drawings.

[0053] In the embodiments of the present application, the words such as "example", "for example" are used to represent as an example, illustration or description. Any embodiment or design scheme described as "example" in the present application should not be interpreted as more preferred or more advantageous than other embodiments or design schemes. Rather, the word "example" is intended to present the concept in a specific way. In addition, in the embodiments of the present application, the meaning expressed by "and / or" can be both, or can be one of the two.

[0054] In the embodiments of the present application, "image" and "picture" can be used interchangeably, and it should be pointed out that the meanings expressed are consistent when the distinction is not emphasized.

[0055] In the embodiments of the present application, sometimes the subscript such as W1 can be written in the form of non-subscript such as W1, and the meanings expressed are consistent when the distinction is not emphasized.

[0056] In order to make the technical problems, technical solutions and advantages to be solved by the present application more clear, the following will be described in detail in conjunction with the drawings and specific embodiments.

[0057] Please refer to Figure 1 The embodiments of the present application provide an LCD display screen production full-cycle digital traceability method, comprising the following steps:

[0058] S1: Obtain the operation state of the panel after entering the initial stage, collect the switching time of the heat source action, extract the time point when each panel starts to be heated according to the sequence, compare the positions where the direction in the stress curve reverses continuously, track the heating and conveying behaviors before and after the positions, extract the time period when the operation time overlaps and the state continuously changes, and obtain a process jump section list;

[0059] S2: Based on the jump time period in the process jump section list, extract the action number in the corresponding time in the scheduling table, develop the behavior chain according to the chronological relationship of the action, eliminate the contents that are not executed in the middle and have an interval, track the numbers that are adjacent and continuously connected between the actions, and obtain a jump section associated node set;

[0060] S3: Based on the jump section associated node set, identify the positions where the structure changes from the quality inspection image, extract the direction form of the line direction of the boundary fracture and the distribution of the bright spots, analyze the image area in the gray scale jump set, separate the distribution range of the bright area, and correspond each image to the position according to the panel number, and obtain a defect label structure list;

[0061] S4: Based on the defect label structure list, extract the action number interval in which the image acquisition time is located from the panel execution log, compare the time period corresponding to the image with the time when the operation occurs, extract the time-continuous sequence content in the number where the image acquisition point is located, and according to the time direction, connect the image performance and the corresponding action to obtain a path mapping table corresponding to the label;

[0062] S5: based on the path mapping table corresponding to the label, extracting the operation number corresponding to the image serial number, unfolding the numbers included in the label in the same batch according to the time extension order, pushing the position sequence between images according to the arrangement relationship of the action in the execution chain, tracking the coverage paragraph and the interleaving section of the label on the operation behavior, and obtaining the label path tracing structure set.

[0063] The process jump section list includes the heat source switching time point, the initial heating time point, the stress curve direction reversal position, the operation state transformation period, the heating and conveying behavior track, the jump section associated node set includes the action number sequence, the time interval mapping, the continuous connection node, the behavior chain structure, the defect label structure list includes the structure change position, the boundary fracture line direction, the bright spot distribution form, the gray scale jump area, the bright area distribution range, the image number position matching, the path mapping table corresponding to the label includes the image acquisition time, the action number interval, the image and action sequence mapping, the image performance and time sequence comparison, and the label path tracing structure set includes the operation number sequence, the image time sequence, the label number sorting, the action coverage paragraph, and the interleaving section.

[0064] Please refer to Figure 2 , the specific steps of S1 are:

[0065] S101: obtaining the operation state of the panel after entering the initial stage, collecting the time corresponding to the heat source switching signal, extracting the panel heating starting point, comparing the time interval between the heat source signal and the panel heat feeling, extracting the time sequence and constructing the continuous sequence, and obtaining the heating time starting sequence;

[0066] The working state of the panel after entering the initial stage is acquired, the heat source signal end of the production line where the panel is located is monitored, the voltage signal amplitude corresponding to the current change when the heating device is powered on is collected, and the triggering moment of the heating action is recorded through the signal collection device. At the same time, a thermal sensitive probe is arranged at the temperature sensing point of the panel to sense the first rising position of the temperature change of the panel surface, and the critical point when the temperature sensing signal change rate first exceeds the upper limit threshold of the static fluctuation determined through statistical analysis is recorded as the heating starting point. An independent number is set for each panel during the collection process, and the number and the temperature sensing signal are stored in the collection data table. By comparing the triggering time of the heat source signal of each panel and the detection value of the sudden increase of the change rate of the temperature sensing point, the response sequence between them is determined, and the effective heat sensing response and the background noise signal are distinguished according to the signal change amplitude. When the temperature change amplitude continuously exceeds the statistical threshold, it is determined as an effective heating signal segment, and the difference between the starting point of the signal segment and the heating signal time is compared. The signal offset between the adjacent two points is converted into time interval value, and the delay between the heat source action and the panel response is determined according to the time interval value. In the sequence of multiple panels, the heat sensing response sequence is arranged according to the relative size of the time interval, the heat source signal number and the panel number are corresponded through the number mapping method, a signal sequence table is formed, the logical judgment of the number sequence is carried out, and when the heat source signal number increases and the temperature sensing number correspondingly decreases, it is determined that there is a time overlapping section in the sequence, which is included in the statistical range of continuous heating interval. Subsequently, the continuous response is regarded as the same heating starting point group according to the fact that the number of continuous temperature sensing signals exceeds the determination threshold value determined through experiments, and finally the heating starting point groups of all panels are aggregated according to the number, the number items of the earliest response in each group are extracted, and the continuous heat sensing response chain is arranged according to the number items, and the heating time starting sequence is obtained.

[0067] S102: Based on the heating time starting sequence, a stress change curve under a synchronous time axis is extracted, a segment with continuous occurrence direction reversal is searched, a position segment with alternating change of slopes between adjacent points is extracted, and a change sequence is filtered according to time continuity, to obtain a stress direction reversal segment sequence.

[0068] First, the stress measurement points after the heat starting position of each panel are extracted one by one, the numerical difference between any two adjacent stress points is calculated, and the change direction of the continuous difference value is judged. If the current difference is positive and the previous difference is negative, or the current is negative and the previous is positive, the position is recorded as a direction reversal point. In this way, the whole data is traversed, the positioning operation of the reversal point is completed, and then the interval time between each group of adjacent reversal points is taken as the basis for selection. The points whose time interval exceeds the average distance of the adjacent panel thermal response section by a preset multiple M (for example, 3 times) are screened out, only the parts with adjacent time and continuous amplitude change are retained, and a continuous slope change sequence is constructed. In the sequence, the slope change direction of the adjacent three points is analyzed in turn. If the directions of two slopes are both reversed and not equal, the three points are determined as the slope alternating section. A plurality of segments are aggregated and processed, the frequency of repeated reversal of the slope direction is calculated, and the segments whose reversal times are less than the preset minimum reversal times K (for example, 5 times) are screened out. In the remaining paragraphs, the reversal points are sequentially numbered using the time axis, and the time interval value of each numbered point and the previous numbered point is calculated. When all intervals are within the value range of the panel thermal response interval, the change sequence has time continuity. The sequence meeting the characteristics is included in the section result set. Taking the numbered sequence as an index, the stress curve changes of different panels are compared and classified. The parts with time overlap or direction change occurring at the same time are grouped into the same sequence. Finally, the stress direction reversal section sequence is obtained.

[0069] S103: Based on the stress direction reversal section sequence, the temperature control and conveying action numbers on both sides of the corresponding section are extracted, the overlapping range in the action duration is compared, the segments with continuous relationship in the action overlapping part are screened out, the action corresponding numbers are extracted, and the process jump section list is obtained;

[0070] First, the time number corresponding to the start and end positions of each section is extracted, and the action records occurring on both sides are retrieved in the temperature control log and the conveying log with the time number as the index. The start and end numbers of each action record are extracted, the time interval is constructed, and the time interval is compared with the stress section time range. The overlapping interval extraction operation is performed on the time interval with intersection, and the overlapping start and end numbers are extracted to form the action intersection interval list. Then, the number connection relationship of adjacent overlapping intervals in the list is compared pair by pair, and it is judged whether the current overlapping end number and the next overlapping start number are consecutive. If multiple overlapping sections satisfy the condition, the continuous action section is aggregated and identified, and the corresponding action number set is recorded. The single action section that does not satisfy the number continuous feature is removed. For each group of aggregated continuous sections, the action numbers involved in the internal are extracted for number aggregation, and the number aggregation result is de-duplicated to generate a unique set of action numbers. Then, all action numbers with continuous relationship are arranged in sequence, and the corresponding temperature control or conveying action source is marked in each group. The information is mapped to the start stress section to finally obtain the process jump section list.

[0071] Please refer to Figure 3 , the specific steps of S2 are:

[0072] S201: Based on the jump time period in the process jump section list, the action numbers in the corresponding time are extracted from the dispatching table, the number start and end time is extracted, the trigger sequence between adjacent numbers is analyzed, the number section with adjacent time before and after is filtered, the number set connected in sequence is extracted, and the action time connection sequence set is obtained;

[0073] First, the starting number and the ending number corresponding to each jump record are extracted respectively, the number section is taken as the time index parameter, the action number set is searched according to the number range in the scheduling data table, the trigger time corresponding to the action number is obtained, the number position of the starting point and the ending point of the trigger time is recorded, and the corresponding mapping relationship between the starting number and the ending number is established, then each group of action numbers is arranged in ascending order according to the number, to form an action set with a clear sequence relationship, and then the trigger time interval between adjacent action numbers is compared based on the set, for the adjacent number trigger time difference within the preset number difference threshold (for example, the difference is less than or equal to 3), it is judged that it has a continuous execution relationship, and the number of this kind is divided into a continuous number set in the same time period, and for the action number pair exceeding the three-bit number interval, it is placed in different sets respectively to avoid false association. In the operation process, for example, if the jump section number is L037 to L042, the action number sequence in the section in the scheduling table is A41, A42, A43 and A47, the number difference between A41 and A43 is 1, so it is a continuous section and can be classified into a group, and the number difference between A43 and A47 is 4, which is not within the continuous range, so A47 needs to be classified into another group. The trigger time of each number in each group is extracted and sorted in order, and finally the action time connection sequence set is obtained.

[0074] S202: Based on the action time connection sequence set, the time interval value between the number sequence is searched, the data section without trigger and with interruption is identified, the piece with disconnected number before and after the time span is removed, the number set with continuous time and no interruption is extracted, and the continuous execution number sequence is obtained;

[0075] First, each set of numbered sets in the sequence is processed by group by group, extracting the trigger time corresponding to each number, constructing the trigger time sequence, subtracting the trigger time values of adjacent two numbers to obtain the time interval value, and sequentially processing all number pairs to form a time interval sequence. For each value in the time interval sequence, an abnormal interval is identified using a threshold determination method. The determination reference value is set as the median interval value of the current numbered set. If a time interval is greater than twice the median value, it is determined that there is an untriggered segment or an interrupted segment, and the abnormal section number range is marked in the numbered set. Further check whether these sections are in the boundary position in the numbered set. If it is at the beginning or end of the set, the number segment is directly removed. If it is located inside the set, the abnormal section is separated from the numbered set to form a separate segment. In the operation process, for example, the numbered set is B41, B42, B43, B50, B51, and the time interval between B43 and B50 is four times the median interval value. It can be judged that this is an interrupted segment, and the numbers between B43 and B50 are removed, while B41 to B43 and B50 to B51 are retained as two effective paragraphs. Repeat the operation to process all numbered sets and remove numbered sections with interruptions or trigger failures. Finally, only numbered combinations with sequential trigger time and no time discontinuity are retained to obtain a continuous execution number sequence.

[0076] S203: Based on the continuous execution number sequence, the trigger time difference between the numbers is extracted, the action continuation relationship order is analyzed, the number chain adjacent in time is selected, the start and end mapping nodes of each number are extracted, and the jump segment association node set is obtained.

[0077] First, the action trigger start and end time corresponding to each number is extracted, a time table is constructed, the trigger start time between two adjacent numbers in the number sequence and the end time of the previous number are operated by difference, and a sequence of action trigger time differences is obtained. If the end time of number B12 is 96 seconds and the start time of number B13 is 97 seconds, the time difference between the two is 1 second, which is lower than the operation gap reference threshold value determined by statistical analysis or experimental verification, and it is judged to be a continuous number relationship; if the interval between two numbers exceeds the threshold value, it is considered as non-continuous operation, the number pair is removed, all number pairs with adjacent time difference below the reference threshold value are retained, and the number chain is connected in order to obtain a time continuous number chain table. Next, for each number in each number chain table, the input and output device node information involved in the action execution is extracted in turn, and a mapping index table is constructed. In specific operation, for example, the number B13 action is a conveying operation, the start node is the conveyor A section, and the end node is the switching table B section, so the corresponding start and end mapping nodes of the number are A section to B section. Repeat the extraction process for all numbers in the number chain, and complete the mapping of numbers to nodes in order. Finally, in the node mapping set of the number chain, the start and end node information of all numbers is extracted, the transition of the number chain to the node chain is completed, and the jump section associated node set is obtained.

[0078] Please refer to Figure 4 , the specific steps of S3 are:

[0079] S301: Based on the jump section associated node set, the quality inspection image corresponding to the panel is extracted, the structural change area in the monitoring image is monitored, the edge line and fracture trend of the bright-dark boundary are extracted, the image block with morphological mutation is identified, the boundary contour shift trend in the continuous frame is compared, the pixel set with change characteristics is extracted, and the structural change feature matrix is obtained;

[0080] First, the image number and node number are established one-to-one correspondence relationship from the quality inspection image storage area extraction of the batch of all panel images, and then the local area detection operation is performed on each image. First, the brightness distribution interval is extracted in the gray level, and the pixel area with a gray value difference of more than 30 degrees is identified as the light and dark boundary area. Then, the continuous change of pixel intensity on the edge of the boundary area is tracked, the direction of the edge line is extracted, and if the direction of the continuous pixel gradient changes more than 45 degrees, it is marked as a broken point, and the extension direction and length range of the broken line are recorded. By comparing the line direction difference between the two images before and after the same area, if the direction offset angle is in the interval of 10 degrees to 60 degrees and the line overlap rate is less than 50%, it is judged that the area has structural changes. The coordinate range and pixel block size of the area are extracted to form a preliminary change area set. Then, the brightness distribution center of each area in the set is detected, and the part with a brightness change amplitude of more than 20% of the average brightness of the original image area is classified as a bright spot change point, and its boundary contour line is extracted. Then, according to the coordinate interval of these change points, the adjacent frames are compared, and the boundary offset trend in the continuous frames is analyzed. If the offset of the adjacent frame boundary is more than three times in the same direction, it is determined that the area is a morphological mutation image block, and the corresponding pixel area information is extracted. Then, the gray gradient, direction vector and boundary offset displacement value of all morphological mutation image blocks are merged to form a pixel change feature set. Finally, all pixel change information in the set is arranged in matrix form, each row represents the feature parameter sequence of the change area in a single frame, and each column corresponds to the change trend value between consecutive frames to obtain a structural change feature matrix.

[0081] S302: Based on the structural change feature matrix, the spatial gradient value of the brightness distribution is extracted, the direction and diffusion range of the bright spot area are analyzed, the concentrated area of gray scale continuous jump is identified, the bright area distribution and gray scale gradient information are compared, the edge mutation area block is extracted, and the brightness and gray scale associated segment set is obtained.

[0082] First, the brightness change data of each pixel point in space is extracted from the matrix, the gray difference of adjacent pixels in horizontal and vertical directions is calculated, each pixel is compared with the gray difference in four directions around it, and the pixel points with a difference value exceeding 25 are determined to have a significant brightness gradient. Then, the pixel points are clustered in the matrix, divided into multiple brightness distribution regions, and the spatial gradient value of each region is calculated. The direction vector of the gradient value is used to identify the direction of the bright spot region. In an actual sample, the gradient direction of the bright spot region at the right lower corner of the panel is concentrated between 60 degrees and 90 degrees, indicating that the brightness of the region spreads along the vertical direction. Then, the distribution of gray continuous change is measured in each bright spot region, the pixel points with a gray change amplitude between 10 degrees and 40 degrees are defined as gray step points, the concentration ratio of these step points in the region is calculated, and if the concentration ratio exceeds 30%, the region is marked as a gray step concentration area. On this basis, the distribution vector of the bright area is compared with the gray gradient direction point by point, and if the included angle between the two directions is less than 20 degrees, it is determined that the brightness change and the gray change tend to be consistent, and these regions are regarded as potential edge mutation areas. Further detect the brightness gradient difference of the edge, if the gradient difference value exceeds 50, it is confirmed that the region has edge mutation, then extract the edge contour line and region block position index of the region, and record in the change data set. By comparing the brightness gradient distribution and gray change relationship of multiple regions, all region information meeting the mutation characteristics is summarized, and finally the brightness and gray correlation segment set is obtained.

[0083] S303: Based on the brightness and gray correlation segment set, the panel number corresponding to each image is extracted, the corresponding relationship of the number and image position in the time sequence is analyzed, the bright area distribution and number mapping section are retrieved, the morphological features in the image sequence are compared with the number, and the defect label structure list is obtained.

[0084] First, the number index of each image in the acquisition record is extracted, the panel tracking table is retrieved through the number index, the panel number information corresponding to the image number is obtained, and the panel number is matched with the image file. In a batch, if the image serial number is I126, the corresponding panel number is P07, the mapping relationship of I126-P07 is established, then all image numbers are sorted according to the order of panel numbers, and the position information in the production circulation is extracted. The corresponding relationship between the panel number and the image position in the time sequence is unfolded into one-dimensional arrangement, the distance change between the continuous numbers is analyzed, whether the adjacent panel images maintain continuity in time is judged, if the distance between the two adjacent images in the sequence exceeds twice the average interval, the group number is rejected, and the image sequence with uniform time interval is retained. Further retrieve the distribution area of the bright area in the sequence, detect the area with brightness value exceeding 30 of the average value of gray scale in each image, and extract the geometric position parameters of the bright area. Match these parameters with the time sequence section corresponding to the panel number to determine the number mapping section corresponding to each bright area. Then compare the shape features of the image sequence in the mapping section, analyze the shape contour, bright area area and edge direction change of the continuous frames under the same number, if the edge direction of the adjacent frames is consistent and the area change ratio exceeds 25%, it is judged that the corresponding area of the number exists shape change, the number and the change area are bound, and the brightness distribution characteristics and gray scale information are recorded. Finally, all the number and the matching data of the corresponding image shape features are integrated to obtain the defect label structure list.

[0085] Please refer to Figure 5 , the specific steps of S4 are:

[0086] S401: Based on the defect label structure list, the time period corresponding to the image acquisition in the panel execution log is collected, the number interval where the time period is located in the log is extracted, the start and end time of the number in the number interval is retrieved, the numbers corresponding to the continuous acquisition time are screened, and the image time corresponding number sequence is obtained.

[0087] First, the image acquisition record corresponding to the label is retrieved in the panel execution log, the time period information of each image acquisition is extracted, and it is matched with the work interval corresponding to the panel number, for example, when a panel number is P12, the corresponding image acquisition time period is located in the seventh interval of the continuous production stage, the time node sequence of the starting point and the ending point is recorded, then the action number range in these intervals is retrieved in the log data in time sequence, the numbers in the number interval are extracted, and a start and end time mark is added to each number, by comparing the start and end time of the number with the image acquisition time, it is judged which numbers cover the image acquisition process, in one retrieval, when the start time of action number A2334 is earlier than the start point of image acquisition, and the end time of number A2336 is later than the end point of acquisition, it is determined that the number interval A2334 to A2336 is the effective corresponding number interval, then the connection relationship between the continuous numbers in the interval is analyzed, if the interval between the end time of the adjacent number and the start time of the next number does not exceed 2 time units (for example, 2ms), it is determined as time continuous number, if the interval exceeds the value, the number is rejected, the continuous sequence is reserved, the numbers meeting the time connection rule are arranged in order, and the corresponding acquisition interval range is recorded, on this basis, the mapping relationship between the image acquisition time period and the number is established, each acquisition time period is matched to the corresponding number chain, and the results corresponding to different panels are organized into a set in order of the number, to obtain the image time corresponding number sequence.

[0088] S402: based on the image time corresponding number sequence, arranging the numbers in time sequence, extracting the adjacent action content in the continuous number segment, identifying the connection logic between the actions before and after the same number, concatenating the image acquisition action and the adjacent operation behavior, to obtain the image action connection segment set;

[0089] First, arrange in chronological order, index the time relationship between each group of numbers, extract the start and end time corresponding to the number, and judge whether there is a time overlap or a time-adjacent logical relationship by looking up the end and start nodes of adjacent numbers. In the sample of the first numbering sequence B101-B105, if the end time of B102 is immediately followed by the start time of B103, it is determined that the sequential connection relationship is established. Further retrieve the action types corresponding to each number in the continuous numbering segment, for example, the image acquisition action number B103 is followed by the conveying action B102 and the positioning action B104. Extract B102, B103, and B104, and confirm whether there is a device state switching instruction or a difference in executing equipment before and after them. If B102 and B104 operate the same equipment code, and B103 is image acquisition, it is determined that the image acquisition action and the adjacent conveying and positioning actions have a continuous behavior connection process. Then extract the image acquisition action number and find the action number before and after it in time. Search for whether it is a logically allowed action type in terms of action content, for example, image acquisition cannot appear between no positioning or conveying instructions. If the type connection logic is met, the three are grouped into a segment. Cross-compare the action numbers in each segment to exclude the case where there are interruptions or gaps before and after the image acquisition number. If there are more than 2 time units (e.g., 2 ms) of number jumps, they are not counted. Finally, extract each number before and after the image acquisition number in all continuous numbering intervals, record their number, start and end time, operation content, and device number, and organize them into an image acquisition behavior corresponding operation chain segment set in chronological order to obtain an image action connection segment set.

[0090] S403: Based on the image action connection segment set, extract the corresponding information of the image number and the action sequence number in the log, align the image content and the behavior segment according to the time direction order, and determine the action path range corresponding to each image to obtain a path mapping table corresponding to the label;

[0091] First, the number of each image and the action number it is in are extracted. By checking the records in the log, the image number is recorded one by one corresponding to the number of the action it belongs to. Then, according to the order of the records in the log, the execution time of each image corresponding to the action is arranged into a continuous data list. Further, the image collection corresponding action number and its start and end time are sequentially arranged on the time axis. It is judged whether the image appearance time is located in the time period of the action number. If the image recording time is within the action execution time period, it is confirmed that the corresponding relationship is established. For example, if the image number I105 recording time is between the start and end points of the number A210 execution, then I105 corresponds to A210. By traversing the operation list, the image number is aligned to the action time period. Then, the adjacent action numbers and contents corresponding to the number are extracted from the log. If the consecutive numbers A209, A210 and A211 are conveying, photographing and marking actions respectively, then according to the time axis order, the three are included in the behavior path of image I105. Then, the above process is performed for all image numbers one by one. The action numbers with continuous relationship in time are screened out, and the accurate insertion position of the image collection point in the action list is found out. According to the time sequence, the image numbers are grouped. The action number segment covered in each group is extracted, and the number, start and end time, action name and image number are combined to form a group of comparison items, which are finally collected into the complete behavior segment path corresponding to each image, and the path mapping table corresponding to the label is obtained.

[0092] Please refer to Figure 6 The specific steps of S5 are as follows:

[0093] S501: Based on the path mapping table corresponding to the label, the operation number set associated with the image serial number is extracted. The image label numbers in the same batch are expanded. The corresponding numbers of the differential images in the number set are sequentially connected in time sequence to obtain a label number extension chain group.

[0094] First, the image serial number corresponding to the operation number set is extracted from the mapping table, and each image number associated operation item is separated and rearranged to extract the image label number belonging to the same batch, for example, when the batch code is B07, the image number sequence is I701 to I709, and the operation number set is O101 to O115. Then, for each image number, read its corresponding action number sequence in the mapping table and count the start and end order relationship between the numbers. If the execution time period of the numbers exists partially overlaps or continuous connection, it is determined that the relationship can be extended, and the numbering chain is reordered according to the time starting point order to form a time increasing numbering chain. On this basis, the difference between the numbering sets corresponding to different image labels is searched, and the numbering sets of each group of adjacent images are cross-compared. If there is the same number, it is classified into the same label segment, otherwise it is classified into the differentiated segment. For example, image I703 corresponds to numbers O107 to O110, and image I704 corresponds to numbers O111 to O114. The difference between the numbering intervals is O111 to O114. This segment is recorded as a new segment of the label extension sequence. Continue to recursively propagate according to the time sequence, and sequentially connect all image numbers until the extension matching of all label numbers in the batch is completed. In the whole process, the connection starting point, end point and corresponding label identification number of each number are recorded to ensure that the numbering chain in the same batch is arranged continuously in time and the logical order is consistent. Finally, all image number associated operation numbers are aggregated in time sequence to obtain the label number extension chain group.

[0095] S502: Based on the label number extension chain group, the position of the number in the operation chain is searched, the image sequence corresponding to the adjacent numbers in the time sequence is selected, the logical connection relationship between the images in each group of label number chains is extracted, and the image action position sequence set is obtained.

[0096] First, read all the operation numbers contained in each set of label numbers, locate each number corresponding operation on the time axis, and record its start and end time period and the order before and after in the operation chain. Then, for the two adjacent numbers in the same chain group, find the order relationship of the two in the log, for example, the record position of number A in the log is the 15th, and number B is the 16th. It is confirmed that the two are directly adjacent. Then extract the images associated with the pair of numbers to form the logical connection chain of image A and image B in the image sequence. By traversing the entire extended chain group, the mapping order of the operation numbers corresponding to the continuous images is constructed. Further, the images corresponding to adjacent numbers are paired, each pair of images corresponds to the transfer logic between operation numbers. If image X corresponds to number P, image Y corresponds to number Q, and P and Q are continuous in time, it is considered that there is a logical connection between image X and image Y. The images are connected in sequence according to the number order. Each group of images forms an image sequence fragment under the action chain. Then, the number of each image in the fragment and the number of the image connected before form a mapping relationship, for example, image I502 and I503 correspond to numbers O208 and O209 respectively. It is determined that the end time of O208 is earlier than the start time of O209, and it is confirmed that the two images are directly connected. Then it is extended to the subsequent images to form a complete image number order table. Finally, the front and rear relationship between each image in the image sequence and the adjacent image in the operation number order is matched to obtain the connection path index between the images. The front and rear connection path between the images in each group of label number chain is output to obtain the image action position sequence set.

[0097] S503: Based on the image action position sequence set, locate the front and rear distribution range of the image number on the operation behavior, aggregate the operation paragraphs and staggered position sections covered by the label numbers, and obtain the label path tracking structure set;

[0098] First, the number information of each image sequence is read, the relative position of each image number in the operation behavior chain is determined, the number is compared with the operation time period associated with it, the start and end positions of the image number on the operation behavior are extracted, for example, image I605 corresponds to number O312, and its execution time is t1 to t2, then the number corresponds to all the recording contents in this range on the operation behavior, then the time coverage relationship between consecutive numbers is counted, whether there is an overlapping part in the time section between adjacent numbers is judged, if the end time of number O312 overlaps with the start time of number O313, it is judged that there is a time connection between the two numbers in the behavior, and the section is defined as an overlapping segment, then the whole sequence is traversed, the operation interval covered by the image corresponding to each group of numbers is checked one by one, adjacent segments are concatenated in chronological order to form a set of distribution sections of the label number on the operation behavior, then according to the range of each label number on the time axis, the operation paragraphs corresponding to the label number are divided, and the multiple operation intervals covered by the same label are aggregated to obtain the complete behavior range penetrated by the label, for example, label T705 corresponds to numbers O310 to O315 on the time axis, covering continuous action paragraphs such as conveying, detecting and packaging, then these paragraphs are regarded as the behavior interval of the same label path, the same operation is performed in all labels, the number coverage interval of each label and the overlapping position are uniformly arranged to form an overlapping relationship matrix between labels, and finally the operation coverage paragraphs and overlapping sections corresponding to all label numbers are collected in the time direction to obtain a set of label path tracing structures.

[0099] Please refer to Figure 7 An LCD display screen production full-cycle digital traceability system comprises:

[0100] The initial state recognition module obtains the panel initial stage operation state, collects the heat source switching time and the heated starting point, locates the continuous position of direction reversal in the stress curve, selects the processing section with time overlap and continuous state according to the front and rear conveying and heating behavior, and obtains the process jump section list;

[0101] The scheduling behavior extraction module extracts the action number in the corresponding time in the scheduling table based on the jump period in the process jump section list, identifies the action sequence, eliminates the interruption and jump items, extracts the adjacent number section in sequence, and obtains the jump section association node set;

[0102] The defect feature extraction module extracts the structure change position in the image based on the jump section association node set, analyzes the boundary fracture line direction, bright area distribution and gray scale fluctuation characteristics, identifies the position matching relationship corresponding to the panel number, extracts the structure mutation segment of the image sequence, and obtains the defect label structure list;

[0103] The action image association module extracts the image acquisition time action number based on the defect label structure list, filters the time continuous sequence items, connects the image content and the operation number, and obtains a path mapping table corresponding to the label;

[0104] The path chain tracing module extracts the image number and the operation sequence based on the path mapping table corresponding to the label, extends the label associated action paragraph, extracts the connection path in the covering relationship, and obtains a label path tracing structure set.

[0105] The above is only a specific embodiment of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed by the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for digital tracking of the whole cycle of LCD display production, characterized in that, Comprising the following steps: S1: Obtain the initial operation state of the panel, collect the heat source switching time and the initial heating starting point, locate the continuously reversed direction position in the stress curve, select the processing section with time overlap and continuous state according to the front and rear conveying and heating behavior, and obtain the process jump section list; S2: Based on the process jump section list, extract the corresponding action number from the scheduling record, identify the action sequence, eliminate the interrupt and jump items, extract the adjacent number section before and after, and obtain the jump section association node set; S3: Based on the jump section association node set, extract the structure change position in the image, analyze the fracture line direction, bright spot distribution and gray scale fluctuation characteristics, correspond to the panel number, identify the position matching relationship, extract the image sequence structure mutation segment, and obtain the defect label structure list; S4: Based on the defect label structure list, extract the image acquisition time action number, select the time continuous sequence item, connect the image content and operation number, and obtain the path mapping table corresponding to the label; S5: Based on the path mapping table corresponding to the label, extract the image number and operation sequence, extend the label associated action paragraph, extract the connection path in the covering relationship, and obtain the label path tracking structure set; The specific steps of S5 are: S501: Based on the path mapping table corresponding to the label, extract the operation number set associated with the image serial number, expand the image label number within the same batch, sequentially connect the corresponding numbers of the differential images in the number set in time sequence, and obtain the label number extension chain group; S502: Based on the label number extension chain group, search the position of the number in the operation chain, select the image sequence corresponding to the adjacent numbers in the time sequence, concatenate the number before and after relationship, extract the logical connection relationship between images in each group of label number chain, and obtain the image action position sequence set; S503: Based on the image action position sequence set, locate the before and after distribution range of the image number in the operation behavior, aggregate the operation paragraph and staggered position section covered by the label number, and obtain the label path tracking structure set; The process jump section list includes the heat source switching time point, the initial heating time point, the stress curve direction reversal position, the operation state transformation period, the heating and conveying behavior track, the jump section association node set includes the action number sequence, the time interval mapping, the continuous connection node and the behavior chain structure, the defect label structure list includes the structure change position, the boundary fracture line direction, the bright spot distribution form, the gray scale jump area, the bright area distribution range and the image number position matching, the path mapping table corresponding to the label includes the image acquisition time, the action number interval, the image and action sequence mapping, the image performance and time sequence comparison, and the label path tracking structure set includes the operation number sequence, the image time sequence, the label number sorting, the action covering paragraph and the staggered section.

2. The method of claim 1, wherein, The continuously reversed position refers to the section in which the stress direction between adjacent measuring points continuously reverses in the stress change curve, indicating the dynamic change characteristics of the material under the action of heat and stress; The jump item refers to the number section with interruption, non-triggering and time span anomaly in the action number and time sequence.

3. The method of claim 1, wherein the method comprises: The fracture line direction refers to the direction of line fracture and morphological mutation at the bright-dark boundary edge in the quality inspection image, and identifies structural abnormalities and crack propagation trends; The gray scale fluctuation feature refers to the region in which the gray value in the image continuously jumps and gradient changes in space.

4. The method of claim 1, wherein the method is characterized by, The specific steps of S1 are: S101: Obtain the working state of the panel after entering the initial stage, collect the time corresponding to the heat source switching signal, extract the panel heating starting point, compare the time interval between the heat source signal and the panel heat sensing, extract the time sequence and construct a continuous sequence to obtain the heating time starting sequence; S102: Based on the heating time starting sequence, extract the stress change curve under the synchronous time axis, search for the segment where the continuous occurrence direction reverses, extract the position segment where the slope between adjacent points alternately changes, and filter the change sequence according to the time continuity to obtain the stress direction reversal segment sequence; S103: Based on the stress direction reversal segment sequence, extract the temperature control and conveying action numbers on both sides of the corresponding segment, compare the overlapping range in the action duration, filter the segment where the action overlap part has a continuous relationship, extract the action corresponding number, and obtain the process jump segment list.

5. The method of claim 1, wherein the method is characterized by, The specific steps of S2 are: S201: Based on the jump period in the process jump segment list, extract the action number in the corresponding time from the scheduling table, extract the number start and end time, analyze the trigger sequence between adjacent numbers, filter the number segments adjacent in time, extract the number set connected in sequence, and obtain the action time connection sequence set; S202: Based on the action time connection sequence set, search for the time interval value between the number sequence, identify the data segment that is not triggered and has interruptions, remove the segment where the time span and the front and rear number are disconnected, extract the number set that is continuous in time without interruption, and obtain the continuous execution number sequence; S203: Based on the continuous execution number sequence, extract the trigger time difference value between the numbers, analyze the continuation relationship order between actions, select the number chain adjacent in time, extract the start and end mapping nodes of each number, and obtain the jump segment association node set.

6. The method of claim 1, wherein, The specific steps of S3 are: S301: Based on the jump segment association node set, extract the quality inspection image of the corresponding panel, monitor the structural change area in the image, extract the edge line and fracture direction of the bright-dark boundary, identify the image block with morphological mutation, compare the boundary contour offset trend in the continuous frames, extract the pixel set with change characteristics, and obtain the structural change feature matrix; S302: Based on the structural change feature matrix, extract the spatial gradient value of the brightness distribution, analyze the direction and diffusion range of the bright spot area, identify the concentrated area of gray scale continuous jump, correspondingly compare the bright area distribution and gray scale gradient information, extract the region block with edge mutation, and obtain the brightness and gray scale associated segment set; S303: Based on the brightness and gray scale associated segment set, extract the panel number corresponding to each image, analyze the corresponding relationship between the number and the image position in the time sequence, search for the bright area distribution and number mapping segment, compare the morphological features in the image sequence with the number, and obtain the defect label structure list.

7. The method of claim 1, wherein the method is used for the entire life cycle of the LCD display panel. The specific steps of S4 are: ​ S401: Based on the defect label structure list, the image acquisition corresponding time period in the panel execution log is collected, the number interval where the time period is located in the log is extracted, the start and end time of the number interval is retrieved, the number corresponding to the continuous acquisition time is screened, and the image time corresponding number sequence is obtained; S402: Based on the image time corresponding number sequence, the numbers are arranged in chronological order, the adjacent action content in the continuous number segment is extracted, the continuity logic between the actions before and after the same number is identified, the image acquisition action and the adjacent operation behavior are concatenated, and the image action continuity segment set is obtained; S403: Based on the image action continuity segment set, the corresponding information of image number and action sequence number in the log is extracted, the image content and behavior segment are aligned according to the time direction sequence, the action path range corresponding to each image is delimited, and the label corresponding path mapping table is obtained.

8. A LCD display production whole cycle digitalized traceability system, characterized in that, The system is used to realize the LCD display screen production full cycle digital traceability method of any one of claims 1-7, and the system comprises: The initial state recognition module acquires the panel initial stage operation state, collects the heat source switching time and the heated starting point, locates the continuous position of direction reversal in the stress curve, screens the processing section with time overlap and continuous state according to the front and rear conveying and heating behavior, and obtains the process jump section list; The scheduling behavior extraction module extracts the action number in the corresponding time in the scheduling table based on the jump time period in the process jump section list, identifies the action sequence, eliminates the interruption and jump items, extracts the adjacent number segment before and after the connection, and obtains the jump section association node set; The defect feature extraction module extracts the structure change position in the image based on the jump section association node set, analyzes the boundary fracture line direction, bright area distribution and gray scale fluctuation characteristics, corresponds to the panel number, identifies the position matching relationship, extracts the image sequence structure mutation segment, and obtains the defect label structure list; The action image association module extracts the image acquisition time action number based on the defect label structure list, screens the time continuous sequence item, connects the image content and operation number, and obtains the label corresponding path mapping table; The path chain traceability module extracts the image number and operation sequence based on the label corresponding path mapping table, extends the label associated action paragraph, extracts the connection path in the covering relationship, and obtains the label path traceability structure set.

Images