A digital display instrument digital gauge generation system

Abstract

The application discloses a digital display instrument digital gauge generation system and relates to the technical field of image metrology, which comprises an image enhancement module, an algorithm evaluation module and the like.

Description

Technical Field

[0001] This invention relates to the field of image measurement technology, and in particular to a digital display instrument digital measuring tool generation system. Background Technology

[0002] As the application of machine vision, deep learning, and generative artificial intelligence in industrial inspection, automated reading, and instrument inspection continues to deepen, image recognition algorithms for digital display instruments are gradually shifting from manual sample collection and training to a technical approach that combines real-world acquisition, simulation modeling, and generative enhancement. Related solutions typically first acquire instrument images or scene images, and then use generative models such as stable diffusion models to enhance two-dimensional imaging features such as texture, color, lighting, and background to expand the sample coverage and alleviate the problems of high on-site acquisition costs and insufficient sample morphology. References also record that machine vision algorithms are deeply applied in industrial manufacturing inspection and measurement, while generative image enhancement can construct new image samples based on scene description, semantic distribution, and depth features.

[0003] In the evaluation of digital display instrument recognition algorithms, existing generative augmentation typically focuses on visual realism and sample diversity. The generation process emphasizes simulating background, lighting, texture, or viewpoint disturbances, rather than binding standard display readings, frame-level reference values, and maximum permissible indication errors as unified constraints. Therefore, in scenarios where instrument readings change dynamically, decimal point positions are fixed, unit symbols are involved in recognition, and character boundaries are affected by pen segment degradation, if the augmentation process alters the consistency of digital pen segments, symbol pixels, or frame sequence readings, the algorithm's recognition deviation is difficult to reliably correspond to a standard uncertainty with metrological significance. This is because the generation of augmented samples is mainly based on image distribution similarity, rather than on the coordinated control of frame-by-frame reference values, reference value protection areas, and inverse uncertainty constraints. References also point out that images generated by traditional data augmentation do not have corresponding measurement uncertainties and cannot be used as a metrological standard for evaluating the accuracy of digital display instrument recognition algorithms. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a digital display instrument digital gauge generation system to solve the problem of coupling between reading reference value protection and uncertainty control in the generation of digital instrument gauges.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention provides a digital display instrument / digital gauge generation system, comprising: a reference indication generation module, which acquires the instrument type and maximum permissible indication error input by the user, calls the corresponding instrument template to form a task configuration data package, configures standard display values ​​for each frame in chronological order based on the task configuration data package, generates frame-by-frame reference indications, and determines imaging perturbation parameters in reverse with the maximum permissible indication error as a constraint; and a segment mask construction module, which maps digital characters to segment display structures according to the frame-by-frame reference indications, assigns segment degradation features to each segment according to the maximum permissible indication error, generates segment degradation state data, generates display layers based on the segment degradation state data and imaging perturbation parameters, and locks the digital character small... The pixel area corresponding to the point unit symbol is set as the reference value locking area, forming a reference value protection mask. The image enhancement module defines the non-protected area according to the reference value protection mask, superimposes scene imaging feature parameters in the non-protected area and performs generative enhancement processing, while keeping the displayed content in the protected area consistent with the frame-by-frame reference value, and outputs candidate image digital gauges. The algorithm evaluation module binds the candidate image digital gauges and the frame-by-frame reference values ​​according to the frame number and synthesizes them into video digital gauges. It then inputs each image recognition algorithm to be verified, obtains the frame-by-frame recognition results of each algorithm, compares them with the frame-by-frame reference values, calculates the corresponding algorithm recognition accuracy and recognition standard uncertainty, and generates a digital gauge algorithm evaluation report.

[0007] As a preferred embodiment of the digital display instrument and digital gauge generation system of the present invention, the specific steps of obtaining the instrument type and maximum permissible indication error input by the user, and calling the corresponding instrument template to form a task configuration data package are as follows: The instrument type entered by the user is compared with the instrument name field in the instrument template library. If the comparison matches, the instrument type code corresponding to the instrument name field is read. The display resolution in the corresponding instrument template is obtained according to the instrument type code, and the maximum permissible indication error is normalized according to the display resolution, and the corrected maximum permissible indication error is output. Retrieve an instrument template that matches the instrument type code from the instrument template library, obtain an instrument template copy, and configure the uncertainty constraints and disturbance parameter boundaries in the instrument template copy according to the corrected maximum permissible indication error to form a task configuration data package.

[0008] As a preferred embodiment of the digital display instrument digital gauge generation system of the present invention, the frame-by-frame reference value is generated by generating a frame number according to the frame rate and duration in the task configuration data packet, and assigning a standard display value to each frame number according to the display resolution and reading step size in the instrument template copy, and binding the frame number with the corresponding standard display value.

[0009] In a preferred embodiment of the digital display instrument digital gauge generation system of the present invention, the specific steps for determining the imaging disturbance parameters in reverse with the maximum permissible indication error as a constraint are as follows: Candidate disturbance intensities are set using the corrected maximum permissible indication error as a reverse constraint condition, and the candidate disturbance intensities are mapped to the disturbance parameter boundaries in the instrument template copy to determine the candidate imaging disturbance parameters. The instrument display content is pre-rendered based on frame-by-frame reference readings and candidate imaging disturbance parameters to generate candidate display images. The candidate display images are then used for reading recognition and comparison with reference readings to obtain recognition deviation values. The standard uncertainty is calculated based on the recognition deviation values. When the standard uncertainty does not meet the maximum permissible indication error after correction, the deviation between the standard uncertainty and the maximum permissible indication error after correction is used as the basis for reverse adjustment to update the candidate disturbance intensity, and the candidate imaging disturbance parameters are re-determined until the standard uncertainty meets the maximum permissible indication error after correction. The candidate imaging disturbance parameters that meet the conditions are then determined as the imaging disturbance parameters.

[0010] As a preferred embodiment of the digital display instrument digital gauge generation system of the present invention, the specific steps of mapping digital characters to segment display structures based on frame-by-frame reference values, and assigning segment degradation features to each segment according to the maximum permissible indication error to generate segment degradation state data are as follows: Extract the frame display string from the frame-by-frame reference readings, split the frame display string into individual numeric characters, and write each numeric character into the corresponding character display unit in the instrument template copy to form frame-by-frame character positioning data; Each numeric character is matched with the seven-segment stroke encoding table to determine the corresponding illuminated stroke segment number, and the illuminated stroke segment number is written into the stroke segment field of the corresponding character display unit to form the stroke segment display structure. The maximum permissible indication error is matched with the segment degradation level table to determine the corresponding segment degradation level. The brightness degradation state and edge degradation state are configured for the lit segments in the segment display structure according to the segment degradation level to form segment degradation level data. The base brightness of the illuminated segment is replaced according to the brightness degradation state, and morphological erosion is performed on the outline boundary of the illuminated segment according to the edge degradation state. The brightness replacement result and the edge erosion result are written into the status field of the corresponding illuminated segment to form segment degradation state data.

[0011] As a preferred embodiment of the digital display instrument digital gauge generation system of the present invention, the specific steps of generating a display layer based on pen segment degradation state data and imaging disturbance parameters, and locking the pixel area corresponding to the decimal point unit symbol of the digital character as a reference indication locking area to form a reference indication protection mask are as follows: Draw the basic outline of the pen segment according to the coordinates of the lit pen segment in the pen segment degradation state data, fill the pen segment brightness according to the pen segment degradation state data and remove the pen segment gap area to form the initial display layer, and perform sub-pixel anti-aliasing rasterization processing on the initial display layer according to the imaging perturbation parameters to form grayscale transition pixels at the edge of the pen segment to form the display layer. Extract the valid pixel areas corresponding to the numeric characters, decimal points, and unit symbols from the display layer, merge the valid pixel areas into the initial locked area, perform morphological dilation and closing operations on the initial locked area in sequence, mark the processed area as the reference value locked area, and write the reference value locked area into the binary mask layer to form the reference value protection mask.

[0012] As a preferred embodiment of the digital display instrument and digital gauge generation system of the present invention, the specific steps of defining the unprotected area based on the reference indication protective mask, superimposing scene imaging feature parameters in the unprotected area and performing generative enhancement processing are as follows: Traverse the pixel markers in the reference value protection mask, keep the pixels corresponding to the protection values ​​locked, remark the pixels corresponding to the non-protection values ​​as upgradable, and merge all the upgradable pixels to form the unprotected area; According to the unprotected area, the image content corresponding to the enhanced state is cropped pixel by pixel in the display layer, and the cropped image content is written to the same coordinate position in the blank layer to form the image area to be enhanced. Lighting gradient, local shadow and background texture noise are sequentially written into the image area to be enhanced, and bilateral filtering is performed on the background texture noise after writing to keep the background texture noise continuous at the edge position, forming a scene feature superposition area. The processing range is limited by the non-protected area. Boundary feathering is performed on the boundary pixels of the scene feature superposition area and the protected area, while keeping the pixels corresponding to the protected value from participating in the processing, thus forming an enhanced candidate image.

[0013] As a preferred embodiment of the digital display instrument digital gauge generation system of the present invention, the specific steps of simultaneously maintaining the display content within the protected area consistent with the frame-by-frame reference values ​​and outputting candidate image digital gauges are as follows: Copy the protected area pixels corresponding to the reference indication protection mask from the display layer, and overwrite the protected area pixels with the same coordinates in the enhanced candidate image to form a reference indication consistent image; The protected area in the reference value consistent image is checked against the frame-by-frame reference value. When the check passes, the reference value consistent image is determined as a candidate image digital gauge.

[0014] As a preferred embodiment of the digital display instrument digital gauge generation system of the present invention, the specific steps of binding candidate image digital gauges and frame-by-frame reference values ​​according to frame number to synthesize video digital gauges are as follows: binding candidate image digital gauges to frame-by-frame reference values ​​one by one according to frame number, calculating image verification values ​​for each frame after binding, generating a frame-level verification index table, sorting and encoding candidate image digital gauges according to the frame number in the frame-level verification index table, and writing corresponding timestamps and reference value metadata to generate video digital gauges.

[0015] In a preferred embodiment of the digital display instrument digital gauge generation system of the present invention, the specific steps for generating the digital gauge algorithm evaluation report are as follows: The video digital gauge is decoded and verified frame by frame, and when the image verification value of the decoded image frame is consistent with the frame-level verification index table, the video digital gauge is marked as an evaluable state. The video digital gauges in the evaluable state are input frame by frame into the image recognition algorithm to be verified, and the returned recognition strings are aligned according to the decimal position of the corresponding frame number to generate a frame-by-frame recognition value table of the algorithm. The algorithm connects the frame-by-frame recognition value table and the frame-level verification index table according to the frame number. The frame-by-frame recognition values ​​under the same frame number are compared with the standard display values ​​corresponding to the frame-by-frame reference values. Frames with consistent recognition and frames with discrepancies are marked. The algorithm recognition accuracy is calculated based on the proportion of consistent frames in the evaluation frames. An effective deviation sequence is formed based on the difference between the frame-by-frame recognition values ​​and the standard display values. The recognition standard uncertainty is calculated based on the effective deviation sequence. The algorithm recognition accuracy and recognition standard uncertainty are written into the evaluation field according to the algorithm identifier to generate a digital gauge algorithm evaluation report.

[0016] The beneficial effects of this invention are as follows: By defining the pixel areas corresponding to digital characters, decimal points, and unit symbols through a reference value protection mask, and only superimposing scene imaging feature parameters and performing generative enhancement processing in the non-protected areas, regional isolation between display content protection and scene disturbance enhancement is achieved. This ensures that candidate image digital gauges maintain consistent frame-by-frame reference values ​​even under varying illumination, local shadows, and background texture changes. Consequently, it provides a stable benchmark for subsequent video digital gauge synthesis, image verification, frame-by-frame numerical comparison of algorithm recognition, and calculation of recognition standard uncertainty. This enhances the reference value traceability, processing control capabilities, and evaluation conclusion verification capabilities during the evaluation of digital display instrument recognition algorithms. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of a digital display instrument and digital measuring tool generation system.

[0019] Figure 2 A flowchart for task configuration and reference value generation.

[0020] Figure 3 This is a flowchart of the segment mask construction and reference indication protection mask formation.

[0021] Figure 4 This is a flowchart for image enhancement processing and candidate image digital gauge output. Detailed Implementation

[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0025] Reference Figures 1-4 As one embodiment of the present invention, this embodiment provides a digital display instrument digital gauge generation system, comprising the following steps: The reference indication generation module obtains the instrument type and maximum permissible indication error input by the user, calls the corresponding instrument template to form a task configuration data package, configures standard display values ​​for each frame in chronological order based on the task configuration data package, generates frame-by-frame reference indications, and determines the imaging disturbance parameters in reverse with the maximum permissible indication error as a constraint.

[0026] It should be noted that the instrument name, instrument model, measurement range, resolution, and accuracy class entered by the user are obtained through the instrument information input field. The corresponding instrument template is called according to the instrument name and instrument model. The error representation method and error parameters that match the measurement range, resolution, and accuracy class are extracted from the instrument template. When the error is expressed as fixed error, reading proportional error, range proportional error, or segmented error, the upper limit and lower limit of the allowable indication error of the current indication range are determined according to the corresponding conversion relationship. The upper limit, lower limit, error unit, and applicable indication range are associated to form the maximum allowable indication error.

[0027] The instrument type entered by the user is compared with the instrument name field in the instrument template library. If the comparison matches, the instrument type code corresponding to the instrument name field is read.

[0028] Furthermore, spaces are removed from the user-inputted instrument type, and full-width and half-width characters are standardized. The instrument name content is retained, resulting in a user-inputted instrument type with complete character consistency processing. Each instrument name field is read in the order of records in the instrument template library, and the user-inputted instrument type with complete character consistency processing is compared character by character with the currently read instrument name field. When the user-inputted instrument type with complete character consistency processing matches the character content and character order of the currently read instrument name field, the comparison stops, the instrument type code in the record containing the currently read instrument name field is read, and the instrument type code is output as the retrieval basis for the corresponding instrument template.

[0029] It should also be noted that the instrument template library is formed by obtaining the instrument description files and instrument display area calibration files corresponding to each instrument type, extracting the instrument name field, instrument type code, display resolution and reading step size from the instrument description file, extracting the character display unit and display area coordinates from the instrument display area calibration file, and writing the extraction results into the same storage file according to the instrument type code.

[0030] The display resolution in the corresponding instrument template is obtained according to the instrument type code, and the maximum permissible indication error is normalized according to the display resolution, and the corrected maximum permissible indication error is output.

[0031] Furthermore, the instrument type code field is searched one by one in the instrument template library according to the instrument type code. When an instrument template with the same instrument type code is found, the display resolution and maximum permissible indication error are extracted from the corresponding instrument template. The maximum permissible indication error is converted to a unit of measurement consistent with the display resolution, and the upper limit and lower limit of the permissible indication error are retained respectively. The number of resolution intervals corresponding to the maximum permissible indication error is calculated based on the display resolution, and the error resolution correspondence result is generated. The permissible indication change range of the imaging disturbance is determined according to the error resolution correspondence result, forming the imaging disturbance constraint value. The disturbance parameter boundary is configured according to the imaging disturbance constraint value, while the maximum permissible indication error is retained as the permissible error boundary for subsequent identification results and frame-by-frame reference indications to determine the deviation.

[0032] It should also be noted that an instrument template with the same instrument type code is retrieved from the instrument template library to obtain an instrument template copy. The uncertainty constraints and disturbance parameter boundaries in the instrument template copy are configured according to the corrected maximum permissible indication error to form a task configuration data package.

[0033] Furthermore, the instrument type code field is searched one by one in the instrument template library according to the instrument type code. When an instrument template with the same instrument type code is found, it is copied to obtain an instrument template copy. The corrected maximum permissible indication error is configured as the uncertainty constraint in the instrument template copy, so that the uncertainty constraint in the instrument template copy has the target limiting conditions required for subsequent reading identification and uncertainty recalculation. Based on the corrected maximum permissible indication error, the preset disturbance parameter boundary level in the instrument template copy is matched to determine the disturbance parameter boundary corresponding to the corrected maximum permissible indication error. The instrument type code, instrument template copy, corrected maximum permissible indication error, uncertainty constraint, and disturbance parameter boundary are associated and combined to form a task configuration data package.

[0034] Frame numbers are generated based on the frame rate and duration in the task configuration data packet. Standard display values ​​are assigned to each frame number according to the display resolution and reading step size in the instrument template copy. The frame numbers are then bound to the corresponding standard display values ​​to form frame-by-frame reference values.

[0035] Furthermore, the frame rate and duration are read from the task configuration data packet. The number of image frames to be generated is calculated based on the frame rate and duration. Frame numbers are generated continuously from the first frame according to the number of image frames until the frame number corresponding to the last frame is generated. The display resolution and reading step size are read from the instrument template copy. The display resolution is used as the minimum number of digits allowed to be displayed in the standard display value, and the reading step size is used as the interval rule for the change of the standard display value between different frame numbers. The standard display value corresponding to each frame number is determined sequentially according to the arrangement of the frame numbers. The display digits of each standard display value are corrected according to the display resolution so that each standard display value can be displayed according to the display resolution in the instrument template copy. After the standard display values ​​of all frame numbers are determined, each frame number and its corresponding standard display value are recorded as a set of frame-by-frame readings, forming a frame-by-frame reference value according to the arrangement of the frame numbers.

[0036] Candidate disturbance intensities are set using the corrected maximum permissible indication error as a reverse constraint condition, and the candidate disturbance intensities are mapped to the disturbance parameter boundaries in the instrument template copy to determine the candidate imaging disturbance parameters.

[0037] Furthermore, the corrected maximum permissible indication error and the boundary of the disturbance parameters in the instrument template copy are read, and the corrected maximum permissible indication error is used as the inverse constraint condition for the candidate disturbance intensity. Based on the corrected maximum permissible indication error, the corresponding uncertainty constraint range is retrieved in the instrument template copy, and a candidate disturbance intensity matching the uncertainty constraint range is selected so that the candidate disturbance intensity can represent the degree of degradation that the currently generated image needs to achieve. When performing disturbance parameter boundary mapping on the candidate disturbance intensity, the boundary range of each disturbance parameter boundary in the instrument template copy is read first, and then the candidate value of each disturbance parameter is determined according to the candidate disturbance intensity within the boundary range, so that each candidate value does not exceed the disturbance parameter boundary in the instrument template copy. After the candidate values ​​of each disturbance parameter are determined, the candidate imaging disturbance parameters are obtained. The candidate imaging disturbance parameters are used for subsequent pre-rendering of candidate display images based on frame-by-frame reference indications.

[0038] The instrument display content is pre-rendered based on frame-by-frame reference values ​​and candidate imaging perturbation parameters to generate candidate display images. The candidate display images are then used for reading recognition and comparison with reference values ​​to obtain recognition deviation values. The standard uncertainty is calculated based on the recognition deviation values.

[0039] Furthermore, the frame number and standard display value in the frame-by-frame reference values ​​are read, and the standard display values ​​are used to generate the corresponding instrument display content according to the order of the frame numbers. The pixel brightness of the instrument display content is changed according to the brightness change parameter in the candidate imaging disturbance parameters. The character edges of the instrument display content are blurred according to the edge blurring parameter in the candidate imaging disturbance parameters. Noise pixels are added to the display area of ​​the instrument display content according to the noise interference parameter in the candidate imaging disturbance parameters. The pixel position of the instrument display content within the display area is moved according to the position offset parameter in the candidate imaging disturbance parameters, resulting in a candidate display image corresponding to each frame number. The instrument display area in the candidate display image is located, and the numerical characters, decimal point, and unit symbol are extracted from the instrument display area to obtain the recognition reading corresponding to the candidate display image. The recognition reading is compared with the standard display value in the frame-by-frame reference values ​​according to the frame number to obtain the recognition deviation value corresponding to each frame number. The recognition deviation values ​​corresponding to all frame numbers are statistically calculated to obtain the dispersion of the recognition deviation value within the frame-by-frame range, and the dispersion is used as the standard uncertainty, expressed as: ; in: It is the average deviation value of the recognition deviation values ​​corresponding to all frame numbers; It is the first The recognition deviation value corresponding to each frame number; It is the statistical sequence number corresponding to the frame sequence number; It is the total number of frame sequence numbers involved in the statistical calculation; ; in: It identifies the degree of dispersion of the deviation value within each frame and uses it as the standard uncertainty.

[0040] When the standard uncertainty does not meet the maximum permissible indication error after correction, the deviation between the standard uncertainty and the maximum permissible indication error after correction is used as the basis for reverse adjustment to update the candidate disturbance intensity, and the candidate imaging disturbance parameters are re-determined until the standard uncertainty meets the maximum permissible indication error after correction. The candidate imaging disturbance parameters that meet the conditions are then determined as the imaging disturbance parameters.

[0041] Furthermore, the standard uncertainty and the allowable range corresponding to the corrected maximum permissible indication error are read, and the standard uncertainty is compared with the allowable range corresponding to the corrected maximum permissible indication error. When the standard uncertainty is less than the lower limit of the allowable range corresponding to the corrected maximum permissible indication error, it is confirmed that the imaging disturbance of the candidate display image is insufficient. The candidate disturbance intensity is increased according to the preset disturbance intensity update step size, and the brightness change parameter, edge blur parameter, noise interference parameter, and position offset parameter are re-determined within the disturbance parameter boundary in the instrument template copy based on the increased candidate disturbance intensity. When the standard uncertainty is greater than the upper limit of the allowable range corresponding to the corrected maximum permissible indication error, it is confirmed that the imaging disturbance of the candidate display image is too strong. The candidate disturbance intensity is decreased according to the preset disturbance intensity update step size. The candidate disturbance intensity is determined, and the brightness change parameter, edge blur parameter, noise interference parameter, and position offset parameter are redefined within the disturbance parameter boundary in the instrument template copy based on the reduced candidate disturbance intensity. A candidate display image is then generated again based on the redefined candidate imaging disturbance parameters. The regenerated candidate display image is then used for reading identification and reference value comparison, and the standard uncertainty is recalculated. This process of updating the candidate disturbance intensity, redefining the candidate imaging disturbance parameters, regenerating the candidate display image, and recalculating the standard uncertainty is repeated until the standard uncertainty falls within the allowable range corresponding to the corrected maximum allowable indication error. The candidate imaging disturbance parameter corresponding to the point where the standard uncertainty falls within the allowable range corresponding to the corrected maximum allowable indication error is determined as the imaging disturbance parameter.

[0042] It should also be noted that the disturbance intensity update step size is determined by the disturbance parameter boundary level in the instrument template copy. The disturbance intensity change interval corresponding to the disturbance parameter boundary level is read, and the disturbance intensity change interval is used as the candidate disturbance intensity update step size.

[0043] The segment mask construction module maps digital characters into segment display structures based on frame-by-frame reference values, assigns segment degradation features to each segment according to the maximum allowable value error, generates segment degradation state data, generates display layers based on segment degradation state data and imaging perturbation parameters, and locks the pixel area corresponding to the decimal point symbol of the digital character as the reference value locking area to form a reference value protection mask.

[0044] The frame display string is extracted from the frame-by-frame reference readings, the frame display string is split into individual numeric characters, and each numeric character is written into the corresponding character display unit in the instrument template copy to form frame-by-frame character positioning data.

[0045] Furthermore, based on the standard display value corresponding to each frame number in the frame-by-frame reference values, the number of decimal places of the standard display value is determined according to the display resolution in the instrument template copy, and the unit content attached to the standard display value is determined according to the unit symbol in the instrument template copy, generating a frame display string corresponding to each frame number; the numeric characters are split character by character according to the character arrangement order in the frame display string, retaining the character position of each numeric character in the frame display string; the corresponding character display unit is determined from the instrument template copy according to the character position, so that the coordinate range of the numeric character, character position, and character display unit is corresponding; after completing the correspondence between all numeric characters and character display units under the same frame number, the frame-by-frame character positioning data is obtained.

[0046] It should also be noted that the character display unit in the instrument template copy is generated by reading the character position and character position coordinates in the instrument display area calibration file, arranging the character position coordinates from left to right according to the character position, and generating each character position and its corresponding character position coordinates.

[0047] Each numeric character is matched with a seven-segment stroke encoding table to determine the corresponding illuminated stroke segment number, and the illuminated stroke segment number is written into the stroke segment field of the corresponding character display unit to form a stroke segment display structure.

[0048] Furthermore, based on the numerical characters corresponding to each frame number in the frame-by-frame character positioning data, the same numerical characters are searched one by one in the seven-segment segment coding table; when the same numerical characters are found in the seven-segment segment coding table, the illuminated segment number corresponding to the same numerical characters is obtained; based on the character position corresponding to the numerical characters in the frame-by-frame character positioning data, the corresponding character display unit in the instrument template copy is determined; the illuminated segment number is used as the segment field content of the corresponding character display unit to generate a segment display structure containing frame number, numerical characters, character position, character display unit, and illuminated segment number.

[0049] It should also be noted that the seven-segment code table is obtained by reading the correspondence between the seven-segment numeric characters and the illuminated segment numbers stored in the instrument template library.

[0050] The maximum permissible indication error is matched with the segment degradation level table to determine the corresponding segment degradation level. Then, according to the segment degradation level, the brightness degradation state and edge degradation state are configured for the lit segments in the segment display structure to form segment degradation level data.

[0051] Furthermore, using the maximum permissible indication error as a screening constraint for the degree of segment degradation, candidate segment degradation levels and their corresponding brightness degradation and edge degradation states are extracted item by item according to the level order in the segment degradation level table. For the same frame number, the brightness degradation and edge degradation states corresponding to the candidate segment degradation levels are applied to the lit segments in the segment display structure to form candidate segment images. Indication verification is then performed on the candidate segment images to obtain candidate recognition values. The candidate recognition values ​​are compared with the standard display values ​​in the frame-by-frame reference values ​​to determine whether the candidate recognition values ​​are within the permissible values ​​limited by the maximum permissible indication error. Within the boundary, the current candidate segment degradation level is retained when the candidate recognition value is within the allowable value boundary, and the current candidate segment degradation level is removed when the candidate recognition value exceeds the allowable value boundary. After completing the value verification of each candidate segment degradation level, the candidate segment degradation level with the highest degradation degree is selected from the retained results, and the corresponding brightness degradation state and edge degradation state are associated with each lit segment under the same frame number. When none of the candidate segment degradation levels pass the value verification, the basic segment state is determined as the segment degradation state corresponding to the current frame number. After the segment degradation state of each frame number is determined, segment degradation level data is formed.

[0052] It should also be noted that the segment degradation level table is formed by the segment base brightness, brightness degradation state and edge degradation state in the instrument template. Each segment degradation level is arranged in order of segment display clarity and is associated with the corresponding brightness degradation state and edge degradation state.

[0053] It should also be noted that the standard display value is derived from the display value bound to the current frame number in the frame-by-frame reference value.

[0054] The base brightness of the illuminated segment is replaced according to the brightness degradation state, and morphological erosion is performed on the outline boundary of the illuminated segment according to the edge degradation state. The brightness replacement result and the edge erosion result are written into the status field of the corresponding illuminated segment to form segment degradation state data.

[0055] Furthermore, based on the brightness degradation state in the segment degradation level data, the degradation value corresponding to the brightness degradation state is obtained, and the original basic brightness value of the illuminated segment is replaced with the degradation value to obtain the brightness replacement result. The brightness replacement result refers to the pixel brightness state after the brightness of the illuminated segment is reduced. Based on the edge degradation state in the segment degradation level data, the morphological erosion kernel size corresponding to the edge degradation state is obtained. The center point of the morphological erosion kernel is placed sequentially at each boundary pixel position of the illuminated segment outline boundary, and it is checked whether the pixels within the coverage area of ​​the morphological erosion kernel all belong to the illuminated segment area. When there are pixels within the coverage area of ​​the morphological erosion kernel that do not belong to the illuminated segment region, the boundary pixels corresponding to the center point of the morphological erosion kernel are removed from the illuminated segment region. When all pixels within the coverage area of ​​the morphological erosion kernel belong to the illuminated segment region, the pixels corresponding to the center point of the morphological erosion kernel are retained. After completing the erosion operation on all boundary pixels of the illuminated segment outline, the edge erosion result is obtained. The edge erosion result refers to the outline shape state of the illuminated segment after the edge shrinkage is completed. According to the frame sequence number and the illuminated segment number, the brightness replacement result and the edge erosion result are mapped to the illuminated segment to form segment degradation state data.

[0056] The basic outline of the pen segment is drawn according to the coordinates of the lit pen segment in the pen segment degradation state data. The pen segment brightness is filled according to the pen segment degradation state data and the pen segment gap area is removed to form the initial display layer. Subpixel anti-aliasing rasterization processing is performed on the initial display layer according to the imaging perturbation parameters to form grayscale transition pixels at the edge of the pen segment, thus forming the display layer.

[0057] Furthermore, based on the coordinates of the illuminated segments in the segment degradation state data, a basic outline of the illuminated segments is generated in the blank display layer. This basic outline is formed by a closed pixel region defined by the coordinates of the illuminated segments. Based on the brightness replacement results in the segment degradation state data, the pixels inside the basic outline are assigned corresponding degraded brightness values. Based on the edge erosion results in the segment degradation state data, the shape of the illuminated segment outline after edge shrinkage is preserved. Based on the segment gap areas in the segment degradation state data, pixels corresponding to the segment gap areas are removed from the illuminated segment outline shape after brightness filling, forming the initial display layer. The sub-pixel offset parameter in the imaging perturbation parameters moves the segment outline boundary in the initial display layer to a non-integer pixel position and calculates the segment area coverage ratio for each pixel passed through by the segment outline boundary. When the segment area coverage ratio is full coverage, the degraded brightness value of the corresponding pixel is retained. When the segment area coverage ratio is partial coverage, the grayscale value of the corresponding pixel is calculated between the degraded brightness value and the background brightness value according to the segment area coverage ratio, so that a grayscale transition pixel is formed at the segment outline boundary, transitioning from the degraded brightness value to the background brightness value. After completing the sub-pixel anti-aliasing rasterization processing of all segment outline boundaries, the display layer is formed.

[0058] It should also be noted that subpixel anti-aliasing rasterization refers to the grayscale smoothing of the edges of the pen segment that fall at non-integer pixel positions when converting the pen segment outline boundary from geometric coordinates to pixel display results; specifically, it involves calculating the coverage ratio of the pen segment outline boundary to adjacent pixels, and allocating the pen segment brightness and background brightness according to the coverage ratio, so that the pen segment edges do not present abrupt stepped jagged edges, but present continuous grayscale transition pixels.

[0059] The sub-pixel offset parameter in the imaging perturbation parameters is determined by the candidate perturbation intensity within the perturbation parameter boundary in the instrument template copy.

[0060] Extract the valid pixel areas corresponding to the numeric characters, decimal points, and unit symbols from the display layer, merge the valid pixel areas into the initial locked area, perform morphological dilation and closing operations on the initial locked area in sequence, mark the processed area as the reference value locked area, and write the reference value locked area into the binary mask layer to form the reference value protection mask.

[0061] Furthermore, in the display layer, effective pixel areas belonging to digit characters, decimal points, and unit symbols are extracted according to their pixel brightness ranges. All effective pixel areas under the same frame number are grouped into the same locking range according to their pixel coordinates, forming an initial locking region. When performing morphological dilation on the initial locking region, the center of the dilation kernel is placed sequentially at the boundary pixel positions of the initial locking region, and adjacent pixels within the coverage area of ​​the dilation kernel are included in the initial locking region, creating an expanded protection range around the digit characters, decimal points, and unit symbols. When performing closing operations on the initial locking region after morphological dilation, morphological dilation is first performed to connect adjacent broken areas, followed by morphological erosion to restore the region boundaries, filling the small holes and boundary gaps inside the digit characters, decimal points, and unit symbols. The region after morphological dilation and closing operations is marked as a reference value locking region, and in the binary mask layer, the pixels corresponding to the reference value locking region are marked as protected values, while pixels outside the reference value locking region are marked as unprotected values, forming a reference value protection mask.

[0062] It should also be noted that the binary mask layer is a pixel marker layer used to distinguish between the reference value locked area and the non-protected area. Each pixel in the binary mask layer has only a protected value or a non-protected value. The protected value corresponds to the pixel position where the numeric characters, decimal point and unit symbol need to be consistent with the reference value, and the non-protected value corresponds to the pixel position where subsequent enhancement processing is allowed.

[0063] The binary mask layer is obtained by first generating a blank pixel marker layer with the same size as the display layer, then marking the pixel positions corresponding to the reference value locking area as protected values, and marking the pixel positions outside the reference value locking area as unprotected values.

[0064] The dilation kernel center is the center position of the structuring element used to align the position of the pixel to be processed in morphological dilation processing.

[0065] The center of the expansion nucleus is determined according to the size of the expansion nucleus, and is usually taken as the position corresponding to the geometric center of the expansion nucleus.

[0066] The image enhancement module defines the unprotected area based on the reference indication protection mask, overlays scene imaging feature parameters in the unprotected area and performs generative enhancement processing, while keeping the displayed content in the protected area consistent with the frame-by-frame reference indication, and outputs candidate image digital gauges.

[0067] Traverse the pixel markers in the reference value protection mask, keep the pixels corresponding to the protection values ​​locked, remark the pixels corresponding to the non-protection values ​​as upgradable, and merge all the upgradable pixels to form the unprotected area.

[0068] Furthermore, the pixel markers are checked one by one according to the pixel coordinates in the reference value protection mask. When the current pixel is marked as a protected value, the display layer position corresponding to the current pixel is kept locked, so that the current pixel does not participate in the subsequent generative enhancement process. When the current pixel is marked as a non-protected value, the display layer position corresponding to the current pixel is marked as an enhanceable state, so that the current pixel can participate in the subsequent generative enhancement process. After checking all pixel markers in the reference value protection mask, all pixels in the enhanceable state are grouped into the same processing area according to the pixel coordinate range to form an unprotected area, which is used to limit the scope of the subsequent generative enhancement process.

[0069] The image content corresponding to the enhanced state is extracted pixel by pixel from the unprotected area in the display layer, and the extracted image content is written to the same coordinate position in the blank layer to form the image area to be enhanced.

[0070] Furthermore, following the pixel coordinate order in the unprotected area, the same pixel coordinate positions in the display layer are located one by one, and the pixel values ​​at the same pixel coordinate positions are obtained. The pixel values ​​include grayscale values ​​or color channel values. A blank layer with the same size as the display layer is generated, and each pixel coordinate position in the blank layer is set to an unfilled state. According to the pixel coordinates in the unprotected area, the pixel values ​​at the same pixel coordinate positions in the display layer are placed at the same pixel coordinate positions in the blank layer. The pixel coordinate positions corresponding to the protected values ​​remain unfilled in the blank layer and do not participate in the placement of image content. After the pixel values ​​corresponding to all pixel coordinates in the unprotected area are placed, the image area to be enhanced is formed.

[0071] Lighting gradients, local shadows, and background texture noise are sequentially written into the image area to be enhanced. Bilateral filtering is then applied to the background texture noise after it has been written, so that the background texture noise remains continuous at the edge position, forming a scene feature superposition area.

[0072] Furthermore, in the image area to be enhanced, a lighting gradient is first generated according to the pixel coordinate arrangement direction, so that the pixel brightness in the image area to be enhanced forms a continuous change of brightness along the preset direction; then, the brightness of the corresponding pixel is reduced according to the position range of the local shadow, so that the image area to be enhanced forms a local occlusion or local dark area effect; background texture noise is generated in the image area to be enhanced, so that the unprotected area has subtle granular changes; when performing bilateral filtering on the background texture noise, neighboring pixels are selected according to spatial proximity and pixel value similarity, and the filtering result of the current pixel is calculated using the selected neighboring pixels, so that the background texture noise is smoothed in flat areas, while maintaining a continuous transition at the edge position, forming a scene feature superposition area.

[0073] The processing range is limited by the non-protected area. Boundary feathering is performed on the boundary pixels of the scene feature superposition area and the protected area, while keeping the pixels corresponding to the protected value from participating in the processing, thus forming an enhanced candidate image.

[0074] Furthermore, using the unprotected area as the scope of the boundary feathering process, the adjacency relationship between unprotected and protected pixels is checked one by one in the reference protection mask, and unprotected pixels adjacent to protected pixels are identified as boundary pixels. Starting from the boundary pixels, the boundary feathering width range is determined inwards from the unprotected area. Within the boundary feathering width range, the feathering weight is determined according to the pixel distance from the unprotected pixel to the protected pixel. The closer the pixel distance is to the protected pixel, the closer the feathering weight is to the original pixel content of the display layer; the farther the pixel distance is from the protected pixel, the closer the feathering weight is to the pixel content of the scene feature overlay area. The original pixel content of the display layer and the pixel content of the scene feature overlay area are blended according to the feathering weight to obtain the boundary feathered pixel content. Pixels corresponding to protected values ​​do not participate in the boundary feathering process and retain their original pixel content in the display layer. After completing the pixel blending of the entire boundary feathering width range within the unprotected area, an enhanced candidate image is formed.

[0075] Copy the protected area pixels corresponding to the reference indication protection mask from the display layer, and overwrite the protected area pixels with the same coordinates in the enhanced candidate image to form a reference indication consistent image.

[0076] Furthermore, based on the protection value pixel coordinates in the reference value protection mask, the same pixel coordinate positions in the display layer are located one by one, and the protected area pixels at the same pixel coordinate positions in the display layer are extracted; based on the protection value pixel coordinates, the same pixel coordinate positions in the enhancement candidate image are located one by one, and the original pixels at the same pixel coordinate positions in the enhancement candidate image are replaced with the protected area pixels extracted from the display layer; after completing the replacement of the protected area pixels corresponding to all protection value pixel coordinates, a reference value consistent image is formed.

[0077] The protected area in the reference value consistent image is checked against the frame-by-frame reference value. When the check passes, the reference value consistent image is determined as a candidate image digital gauge.

[0078] Furthermore, based on the pixel coordinates of the protected values ​​in the protective mask, the protected area is extracted from the reference value consistent image; according to the standard display value of the corresponding frame number in the frame-by-frame reference values, a standard display string for the corresponding frame number is generated; the numeric characters, decimal point, and unit symbol in the protected area are compared digit by digit with the standard display string to confirm whether the order of the numeric characters, the position of the decimal point, and the content of the unit symbol in the protected area are consistent with the standard display string; when the order of the numeric characters, the position of the decimal point, and the content of the unit symbol in the protected area are all consistent with the standard display string, the consistency check is passed, and the reference value consistent image is determined as a candidate image digital gauge.

[0079] The algorithm evaluation module combines candidate image digital gauges with frame-by-frame reference values ​​according to frame number to form a video digital gauge. It then inputs each image recognition algorithm to be verified, obtains the frame-by-frame recognition results of each algorithm, compares them with the frame-by-frame reference values, calculates the corresponding algorithm recognition accuracy and recognition standard uncertainty, and generates a digital gauge algorithm evaluation report.

[0080] Candidate image digital gauges are bound one by one to frame-by-frame reference values ​​according to their frame numbers. Image verification values ​​are calculated for each bound frame to generate a frame-level verification index table. Candidate image digital gauges are sorted and encoded according to the frame numbers in the frame-level verification index table, and corresponding timestamps and reference value metadata are written to generate video digital gauges.

[0081] Furthermore, the candidate image digital gauges are read frame by frame according to the frame number, and each candidate image digital gauge in each frame is bound to the standard display value corresponding to the same frame number in the frame-by-frame reference value. Then, the image verification value is calculated for each candidate image digital gauge after binding. The image verification value is obtained by hash verification of the pixel content of the candidate image digital gauge. The frame number, standard display value and image verification value are written into the same frame record to generate a frame-level verification index table. After the frame-level verification index table is generated, the candidate image digital gauges are read in the order of the frame sequence number in the frame-level verification index table, and the candidate image digital gauges are sorted and encoded so that the arrangement order of the candidate image digital gauges in the encoding file is consistent with the frame sequence number order of the frame-by-frame reference values. At the same time, a corresponding timestamp is written for each frame after encoding according to the frame sequence number, and the frame-by-frame reference values ​​in the frame-level verification index table are written as reference value metadata into the encoding file to generate video digital gauges.

[0082] The video digital gauge is decoded and verified frame by frame, and when the image verification value of the decoded image frame is consistent with the frame-level verification index table, the video digital gauge is marked as an evaluable state.

[0083] Furthermore, the video digital gauge is decoded frame by frame. Decoded image frames are extracted sequentially according to the timestamp and frame number written in the video digital gauge, and an image check value is calculated for each decoded image frame. The image check value is obtained by hashing the pixel content of the decoded image frame. Then, the corresponding image check value is read from the frame-level check index table according to the frame number, and the image check value of the decoded image frame is compared with the image check value in the frame-level check index table frame by frame. The frame-by-frame comparison is used to confirm whether the pixel content of the candidate image digital gauge is consistent during the sorting encoding and decoding process. When the image check value of all decoded image frames is consistent with the image check value of the corresponding frame number in the frame-level check index table, the video digital gauge is marked as evaluable.

[0084] The video digital gauges in the evaluable state are input frame by frame into the image recognition algorithm to be verified, and the returned recognition strings are aligned according to the decimal position of the corresponding frame number to generate a frame-by-frame recognition value table of the algorithm.

[0085] Furthermore, after the video digital gauge is marked as evaluable, the decoded image frames in the video digital gauge are read frame by frame according to the frame sequence number in the frame-level verification index table. The decoded image frames corresponding to the same frame sequence number are then sequentially fed into the Tesseract OCR recognition algorithm, the CRNN text recognition algorithm, and the PaddleOCR text recognition algorithm. The Tesseract OCR recognition algorithm performs character recognition on the digital character regions in the decoded image frames and outputs the recognition string returned by the Tesseract OCR recognition algorithm. The CRNN text recognition algorithm performs sequence recognition on the arrangement of digital characters in the decoded image frames and outputs the recognition string returned by the CRNN text recognition algorithm. The PaddleOCR text recognition algorithm performs text detection and text recognition on the digital character regions in the decoded image frames and outputs the recognition string returned by the PaddleOCR text recognition algorithm.

[0086] After obtaining the recognition strings returned by the TesseractOCR, CRNN, and PaddleOCR algorithms, the corresponding decimal point positions are read from the frame-level verification index table according to the frame sequence number. Format alignment processing is then performed on the recognition strings returned by the TesseractOCR, CRNN, and PaddleOCR algorithms respectively. This format alignment process includes retaining numeric characters and decimal point characters in the recognition strings, deleting space characters, unit symbols, and non-reading characters, and then correcting the decimal point position in the recognition strings according to the decimal point position corresponding to the same frame sequence number in the frame-level verification index table, ensuring that the number of decimal places in the recognition strings is consistent with the standard displayed value in the frame-by-frame reference indication.

[0087] After format alignment is completed, the recognition string returned by the TesseractOCR recognition algorithm is written into the corresponding frame-by-frame recognition value of the TesseractOCR recognition algorithm, the recognition string returned by the CRNN text recognition algorithm is written into the corresponding frame-by-frame recognition value of the CRNN text recognition algorithm, and the recognition string returned by the PaddleOCR text recognition algorithm is written into the corresponding frame-by-frame recognition value of the PaddleOCR text recognition algorithm. The frame number, the corresponding frame-by-frame recognition value of the TesseractOCR recognition algorithm, the corresponding frame-by-frame recognition value of the CRNN text recognition algorithm, and the corresponding frame-by-frame recognition value of the PaddleOCR text recognition algorithm are written into the same frame record to generate the algorithm frame-by-frame recognition value table.

[0088] It should be noted that the Tesseract OCR recognition algorithm is an open-source optical character recognition algorithm that can directly recognize text characters in images and output the recognized string. In the scenario of digital display instrument reading recognition, the Tesseract OCR recognition algorithm is suitable for reading recognition of decoded image frames with relatively clear character boundaries and relatively regular number arrangement, and can be used as a representative algorithm of traditional optical character recognition methods.

[0089] CRNN is a text recognition algorithm for image sequence recognition. It is called Convolutional Recurrent Neural Network for text recognition. It typically completes the recognition of text sequences in an image through convolutional feature extraction, sequence modeling, and transcription output. In the scenario of recognizing readings of digital display instruments, CRNN is suitable for recognizing the entire sequence of continuously arranged digit characters and can be regarded as a representative algorithm of deep learning sequence recognition methods.

[0090] PaddleOCR is an open-source text recognition tool that includes text detection and recognition capabilities. It can locate text regions in images and output recognized strings. In digital display instrument reading recognition scenarios, PaddleOCR is suitable for reading recognition of decoded image frames with slight tilt, complex backgrounds, or changes in the position of the display area. It can be used as a representative algorithm for engineered text recognition methods.

[0091] The algorithm connects the frame-by-frame recognition value table and the frame-level verification index table according to the frame number. The frame-by-frame recognition values ​​under the same frame number are compared with the standard display values ​​corresponding to the frame-by-frame reference values. Frames with consistent recognition and frames with discrepancies are marked. The algorithm recognition accuracy is calculated based on the proportion of consistent frames in the evaluation frames. An effective deviation sequence is formed based on the difference between the frame-by-frame recognition values ​​and the standard display values. The recognition standard uncertainty is calculated based on the effective deviation sequence. The algorithm recognition accuracy and recognition standard uncertainty are written into the evaluation field according to the algorithm identifier to generate a digital gauge algorithm evaluation report.

[0092] Furthermore, the frame-by-frame recognition value table and the frame-level verification index table are connected according to the frame number, so that the frame-by-frame recognition value under the same frame number corresponds to the standard display value in the same record. Then, using the standard display value as the comparison benchmark, the recognition deviation value is calculated for the frame-by-frame recognition values ​​corresponding to the Tesseract OCR recognition algorithm, the CRNN text recognition algorithm, and the PaddleOCR text recognition algorithm, respectively. The recognition deviation values ​​are then summarized according to the algorithm identifier to form the effective deviation sequence for the Tesseract OCR recognition algorithm, the CRNN text recognition algorithm, and the PaddleOCR text recognition algorithm. Then, the uncertainty of the evaluation value of the corresponding image recognition algorithm to be verified is calculated based on each effective deviation sequence. The algorithm identifier, frame-by-frame recognition value, effective deviation sequence, and algorithm evaluation uncertainty are written into the evaluation result field to generate a digital gauge algorithm evaluation report containing the evaluation content of each image recognition algorithm to be verified.

[0093] In summary, this invention achieves regional isolation between display content protection and scene disturbance enhancement by using a reference value protection mask to define the pixel areas corresponding to digital characters, decimal points, and unit symbols, and by superimposing scene imaging feature parameters and performing generative enhancement processing only in the non-protected areas. This ensures that candidate image digital gauges maintain consistent frame-by-frame reference values ​​even under varying illumination, local shadows, and background texture changes. Consequently, it provides a stable benchmark for subsequent video digital gauge synthesis, image verification, frame-by-frame numerical comparison of algorithm recognition, and calculation of recognition standard uncertainty. This enhances the reference value traceability, processing control capabilities, and evaluation conclusion verification capabilities during the evaluation of digital display instrument recognition algorithms.

[0094] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A digital display instrument digital gauge generation system, characterized in that, include: The reference indication generation module obtains the instrument type and maximum permissible indication error input by the user, calls the corresponding instrument template to form a task configuration data package, configures standard display values ​​for each frame in chronological order based on the task configuration data package, generates frame-by-frame reference indications, and determines the imaging disturbance parameters in reverse with the maximum permissible indication error as a constraint. The segment mask construction module maps digital characters into segment display structures based on frame-by-frame reference values, assigns segment degradation features to each segment according to the maximum allowable value error, generates segment degradation state data, generates display layers based on segment degradation state data and imaging perturbation parameters, and locks the pixel area corresponding to the decimal point unit symbol of the digital character as the reference value locking area to form a reference value protection mask. The image enhancement module defines the unprotected area based on the reference value protection mask, overlays scene imaging feature parameters in the unprotected area and performs generative enhancement processing, while keeping the displayed content in the protected area consistent with the frame-by-frame reference value, and outputs candidate image digital gauges. The algorithm evaluation module combines candidate image digital gauges with frame-by-frame reference values ​​according to frame number to form a video digital gauge. It then inputs each image recognition algorithm to be verified, obtains the frame-by-frame recognition results of each algorithm, compares them with the frame-by-frame reference values, calculates the corresponding algorithm recognition accuracy and standard uncertainty, and generates a digital gauge algorithm evaluation report.

2. The digital display instrument digital gauge generation system as described in claim 1, characterized in that, The steps for obtaining the instrument type and maximum permissible indication error input by the user, and then calling the corresponding instrument template to form a task configuration data package are as follows: The instrument type entered by the user is compared with the instrument name field in the instrument template library. If the comparison matches, the instrument type code corresponding to the instrument name field is read. The display resolution in the corresponding instrument template is obtained according to the instrument type code, and the maximum permissible indication error is normalized according to the display resolution, and the corrected maximum permissible indication error is output. Retrieve an instrument template that matches the instrument type code from the instrument template library, obtain an instrument template copy, and configure the uncertainty constraints and disturbance parameter boundaries in the instrument template copy according to the corrected maximum permissible indication error to form a task configuration data package.

3. The digital display instrument digital gauge generation system as described in claim 2, characterized in that, The frame-by-frame reference value is generated by generating a frame number based on the frame rate and duration in the task configuration data packet, assigning a standard display value to each frame number according to the display resolution and reading step size in the instrument template copy, and binding the frame number with the corresponding standard display value.

4. The digital display instrument digital gauge generation system as described in claim 2, characterized in that, The specific steps for determining the imaging perturbation parameters in reverse, constrained by the maximum permissible indication error, are as follows: Candidate disturbance intensities are set using the corrected maximum permissible indication error as a reverse constraint condition, and the candidate disturbance intensities are mapped to the disturbance parameter boundaries in the instrument template copy to determine the candidate imaging disturbance parameters. The instrument display content is pre-rendered based on frame-by-frame reference readings and candidate imaging disturbance parameters to generate candidate display images. The candidate display images are then used for reading recognition and comparison with reference readings to obtain recognition deviation values. The standard uncertainty is calculated based on the recognition deviation values. When the standard uncertainty does not meet the maximum permissible indication error after correction, the deviation between the standard uncertainty and the maximum permissible indication error after correction is used as the basis for reverse adjustment to update the candidate disturbance intensity, and the candidate imaging disturbance parameters are re-determined until the standard uncertainty meets the maximum permissible indication error after correction. The candidate imaging disturbance parameters that meet the conditions are then determined as the imaging disturbance parameters.

5. The digital display instrument digital gauge generation system as described in claim 3, characterized in that, The specific steps for mapping digital characters to segment display structures based on frame-by-frame reference values, assigning segment degradation features to each segment according to the maximum permissible indication error, and generating segment degradation state data are as follows: Extract the frame display string from the frame-by-frame reference readings, split the frame display string into individual numeric characters, and write each numeric character into the corresponding character display unit in the instrument template copy to form frame-by-frame character positioning data; Each numeric character is matched with the seven-segment stroke encoding table to determine the corresponding illuminated stroke segment number, and the illuminated stroke segment number is written into the stroke segment field of the corresponding character display unit to form the stroke segment display structure. The maximum permissible indication error is matched with the segment degradation level table to determine the corresponding segment degradation level. The brightness degradation state and edge degradation state are configured for the lit segments in the segment display structure according to the segment degradation level to form segment degradation level data. The base brightness of the illuminated segment is replaced according to the brightness degradation state, and morphological erosion is performed on the outline boundary of the illuminated segment according to the edge degradation state. The brightness replacement result and the edge erosion result are written into the status field of the corresponding illuminated segment to form segment degradation state data.

6. The digital display instrument digital gauge generation system as described in claim 5, characterized in that, The process of generating a display layer based on pen segment degradation data and imaging perturbation parameters, and locking the pixel area corresponding to the decimal point symbol of the numeric character as a reference value locking area to form a reference value protection mask, is as follows: Draw the basic outline of the pen segment according to the coordinates of the lit pen segment in the pen segment degradation state data, fill the pen segment brightness according to the pen segment degradation state data and remove the pen segment gap area to form the initial display layer, and perform sub-pixel anti-aliasing rasterization processing on the initial display layer according to the imaging perturbation parameters to form grayscale transition pixels at the edge of the pen segment to form the display layer. Extract the valid pixel areas corresponding to the numeric characters, decimal points, and unit symbols from the display layer, merge the valid pixel areas into the initial locked area, perform morphological dilation and closing operations on the initial locked area in sequence, mark the processed area as the reference value locked area, and write the reference value locked area into the binary mask layer to form the reference value protection mask.

7. The digital display instrument digital gauge generation system as described in claim 6, characterized in that, The process of defining the unprotected area based on the reference indication protective mask, overlaying scene imaging feature parameters within the unprotected area, and performing generative enhancement processing are as follows: Traverse the pixel markers in the reference value protection mask, keep the pixels corresponding to the protection values ​​locked, remark the pixels corresponding to the non-protection values ​​as upgradable, and merge all the upgradable pixels to form the unprotected area; According to the unprotected area, the image content corresponding to the enhanced state is cropped pixel by pixel in the display layer, and the cropped image content is written to the same coordinate position in the blank layer to form the image area to be enhanced. Lighting gradient, local shadow and background texture noise are sequentially written into the image area to be enhanced, and bilateral filtering is performed on the background texture noise after writing to keep the background texture noise continuous at the edge position, forming a scene feature superposition area. The processing range is limited by the non-protected area. Boundary feathering is performed on the boundary pixels adjacent to the protected area of ​​the scene feature superposition area, while keeping the pixels corresponding to the protected value from participating in the processing, thus forming an enhanced candidate image.

8. The digital display instrument digital gauge generation system as described in claim 7, characterized in that, The steps for simultaneously maintaining consistency between the displayed content within the protected area and the frame-by-frame reference values, and outputting candidate image digital gauges, are as follows: Copy the protected area pixels corresponding to the reference indication protection mask from the display layer, and overwrite the protected area pixels with the same coordinates in the enhanced candidate image to form a reference indication consistent image; The protected area in the reference value consistent image is checked against the frame-by-frame reference value. When the check passes, the reference value consistent image is determined as a candidate image digital gauge.

9. The digital display instrument digital gauge generation system as described in claim 8, characterized in that, The specific steps for combining candidate image digital gauges with frame-by-frame reference values ​​according to frame number to form a video digital gauge are as follows: bind the candidate image digital gauges with frame-by-frame reference values ​​one by one according to frame number, calculate the image verification value for each frame after binding, generate a frame-level verification index table, sort and encode the candidate image digital gauges according to the frame number in the frame-level verification index table, and write the corresponding timestamp and reference value metadata to generate a video digital gauge.

10. The digital display instrument digital gauge generation system as described in claim 1, characterized in that, The specific steps for generating the digital gauge algorithm evaluation report are as follows: The video digital gauge is decoded and verified frame by frame, and when the image verification value of the decoded image frame is consistent with the frame-level verification index table, the video digital gauge is marked as an evaluable state. The video digital gauges in the evaluable state are input frame by frame into the image recognition algorithm to be verified, and the returned recognition strings are aligned according to the decimal position of the corresponding frame number to generate a frame-by-frame recognition value table of the algorithm. The algorithm connects the frame-by-frame recognition value table and the frame-level verification index table according to the frame number. The frame-by-frame recognition values ​​under the same frame number are compared with the standard display values ​​corresponding to the frame-by-frame reference values. Frames with consistent recognition and frames with discrepancies are marked. The algorithm recognition accuracy is calculated based on the proportion of consistent frames in the evaluation frames. An effective deviation sequence is formed based on the difference between the frame-by-frame recognition values ​​and the standard display values. The recognition standard uncertainty is calculated based on the effective deviation sequence. The algorithm recognition accuracy and recognition standard uncertainty are written into the evaluation field according to the algorithm identifier to generate a digital gauge algorithm evaluation report.

Images