An image recognition-based engineering survey data analysis method and system

Abstract

The present application relates to the technical field of measurement data analysis, in particular to an engineering measurement data analysis method and system based on image recognition, comprising the following steps: recognizing structural edges, measurement mark points and calibration scales based on an industrial camera, forming a measurement point coordinate reference set, establishing a corresponding scale quantity according to the direction of the calibration scale, forming a measurement point interval representation set in combination with the displacement relationship of the to-be-measured points, then determining the edge line deviation reference rate, and further forming a boundary positioning correlation result. According to the present application, the edges are defined based on the gray level mutation to construct the measurement point positioning reference, the corresponding scale of the horizontal and vertical displacement of the calibration scale is calculated, the interval representation is formed in combination with the measurement point displacement and the edge trend, the edge line reference is formed by analyzing the normal span and the surface deviation state, the arrangement track correlation section is defined in combination with the boundary extension direction, the manual interpretation interference is reduced, the consistency of the multi-direction conversion reference is maintained, the error accumulation process is controlled, and the coherence and stability of the engineering monitoring data processing are improved.

Description

Technical Field

[0001] This invention relates to the field of measurement data analysis technology, and in particular to an engineering measurement data analysis method and system based on image recognition. Background Technology

[0002] The field of measurement data analysis technology involves technologies related to the acquisition, processing, analysis, and utilization of various types of measurement data. This field primarily focuses on the organization, calculation, modeling, and analysis of data acquired in scenarios such as engineering surveying, industrial inspection, and geographic mapping to support engineering evaluation, quality inspection, and spatial information acquisition. It is widely applied in civil engineering, surveying engineering, construction monitoring, and related information processing industries. Traditional engineering measurement data analysis methods involve processing and analyzing data acquired in engineering measurement scenarios. This typically involves acquiring distance, angle, or elevation data using total stations, levels, or distance measuring devices, organizing the raw data through manual or electronic recording, calculating point positions based on coordinate transformation relationships, and using survey sketches or on-site photographs for auxiliary interpretation. Finally, it utilizes the positions of marker points in the images for manual comparison and reading to complete the process of organizing, calculating, and analyzing engineering measurement data.

[0003] Traditional engineering surveying data analysis relies heavily on manual image comparison for interpretation in practice. When faced with complex sites, reading image markers is easily affected by subjective visual bias and operator fatigue. Obtaining raw data requires complex manual intervention for recording and calculation, which continuously extends the data processing cycle. It is difficult to maintain a unified execution standard when deriving multi-dimensional coordinate transformations, which can easily lead to delays in quality inspection and continuous accumulation of positioning errors. This seriously weakens the objectivity, authenticity, and timeliness of engineering evaluation results, and makes it difficult to fully support the stringent requirements of building construction monitoring for high-precision spatial information acquisition. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of existing technologies and to propose an engineering measurement data analysis method and system based on image recognition.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: an engineering measurement data analysis method based on image recognition, comprising the following steps: S1: Based on an industrial camera, acquire grayscale changes in the captured image, determine the structural edges corresponding to continuous abrupt change areas, compare the relationship between the intersection position and the marker outline, filter the coordinates of the measurement marker points, adjust the positioning sequence of the calibration ruler end, and obtain the reference set of measurement point coordinates; S2: Based on the coordinate reference set of the measuring points, obtain the arrangement direction of the endpoints of the calibration ruler, compare the relationship between the lateral displacement and the longitudinal displacement, calculate the direction of the ruler segment, adjust the corresponding order of the actual length of the calibration ruler and the pixel span, determine the conversion relationship of each direction, and obtain the corresponding quantity of the sub-scale. S3: Based on the corresponding quantities of the directional scale, obtain the positional relationship of the points to be measured, calculate the lateral and longitudinal displacements of the two points to be measured, compare the direction mapping scale segments, adjust the displacement conversion order, determine the relationship between the structural edge direction and the line connecting the measuring points, and obtain the measuring point interval characterization set. S4: Based on the measurement point interval characterization set, analyze the measurement point interval distribution, determine the relationship between the measurement marker point and the normal interval of the structural edge, calculate the deviation state corresponding to the edge line of the same measured component, adjust the edge line reference order, and obtain the edge line deviation reference rate. S5: Based on the deviation rate of the edge line, analyze the deviation direction, compare the relationship between the extension direction of the reference boundary and the position of the edge line on the surface of the measured component, determine the corresponding segment of the trajectory of the measurement marker points, adjust the boundary assignment order, and obtain the boundary positioning association result.

[0006] The present invention improves upon the following: the measuring point coordinate reference set includes the measuring point pixel coordinate distribution, the marker point spatial index, and the calibration ruler endpoint coordinate combination; the directional scale corresponding quantity includes the lateral scale conversion reference, the longitudinal scale conversion reference, and the direction decomposition identifier; the measuring point interval characterization set includes the measuring point spacing expression unit, the connection direction identifier, and the component spacing record item; the edge line deviation reference rate includes the normal offset ratio identifier, the lateral offset classification identifier, and the edge line reference matching item; and the boundary positioning association result includes the boundary matching relationship identifier, the positioning segment division identifier, and the boundary attribution correspondence identifier.

[0007] The present invention is improved in that the steps for obtaining the reference set of the measuring point coordinates are specifically as follows: S111: Based on an industrial camera, analyze the gray-level variation relationship between adjacent pixels in the pixel matrix of the acquired image, determine the continuous variation region formed by the gray-level difference, filter the pixel set with spatial connectivity, and calculate the extension direction of the pixel set boundary to obtain the pixel trajectory of the structural edge. S112: Based on the pixel trajectory of the structure edge, determine the coordinates of the intersection position of the pixel trajectory, calculate the coordinates of the center position of the closed pixel region of the marker contour, compare the spatial position relationship between the intersection coordinates and the center coordinates, filter the matching region pixel coordinates, and obtain the set of marker point position coordinates; S113: Based on the set of coordinates of the marker points, calculate the direction of the line connecting the pixel coordinates of the calibration ruler endpoints, determine the consistency of the endpoint arrangement direction, adjust the spatial position corresponding to the endpoint order, and obtain the reference set of measurement point coordinates.

[0008] The present invention is improved in that the step of obtaining the corresponding quantity of the directional scale is specifically as follows: S211: Based on the reference set of measurement point coordinates, analyze the spatial positional relationship of pixel coordinates at the endpoints of the calibration ruler, calculate the endpoint coordinate difference to form a connecting direction vector, determine the direction category of the connecting direction in the image coordinate system, adjust the arrangement order of endpoint coordinates and the consistency relationship with the direction, and obtain the endpoint direction representation vector. S212: Based on the endpoint direction representation vector, compare the correspondence between the horizontal pixel displacement and the vertical pixel displacement, calculate the proportional relationship of the displacement components under the same calibration scale constraint, and determine the distribution state of the pixel length in each direction to obtain the direction distribution structure identifier. S213: Based on the direction allocation structure identifier, analyze the actual length of the calibration scale and the pixel span, calculate the mapping relationship between the horizontal pixel distance and the actual length, determine the correspondence between the vertical pixel distance and the actual length, adjust the direction conversion order, and obtain the corresponding quantity of the directional scale.

[0009] The present invention is improved in that the step of obtaining the measurement point interval characterization set is specifically as follows: S311: Based on the corresponding quantities of the directional scales, analyze the mapping relationship between the horizontal scale conversion reference and the vertical scale conversion reference, calculate the difference of pixel coordinates of the point to be measured to form the horizontal and vertical displacements, determine the relative positional relationship of the coordinates of the two points in the image coordinate system, and obtain the displacement relationship vector of the point to be measured. S312: Based on the displacement relationship vector of the measuring point, analyze the distribution relationship of lateral displacement and longitudinal displacement in the direction mapping scale section, compare the corresponding relationship of each direction conversion benchmark, determine the direction section to which the displacement component belongs, adjust the conversion order of lateral displacement and longitudinal displacement, and obtain the direction conversion sequence identifier. S313: Based on the direction conversion sequence identifier, analyze the extension direction of the pixel trajectory of the structure edge, calculate the relationship between the direction vector of the measurement point connection and the direction of the structure edge, determine the spatial correspondence between the measurement point connection and the structure edge, adjust the relationship between the measurement point spacing and the direction identifier, and obtain the measurement point interval representation set.

[0010] The present invention is improved in that the step of obtaining the edge deviation reference rate is specifically as follows: S411: Based on the measurement point interval characterization set, calculate the distribution density of the measurement mark point spacing in the component surface area, determine the belonging relationship of adjacent measurement point combinations within the same component range, adjust the expression form of measurement point spacing to unify the spatial structure, and obtain the measurement point spacing distribution density; S412: Based on the distribution density of the measuring point spacing, analyze the projected distance from the measuring marker point to the normal direction of the structural edge, compare the differences in the normal direction distance of different measuring points, and determine the consistency relationship between the measuring point offset direction and the edge normal direction, using the formula: ; The sequence of normal offsets at the measurement points is obtained, where, The edge offset metric indicates the degree of offset of the measuring point relative to the edge of the component. The lateral displacement of the measuring point. The longitudinal displacement of the measuring point. The lateral component of the unit vector normal to the edge. The longitudinal component of the edge normal unit vector. The lateral component of the edge tangential unit vector. The longitudinal component of the edge tangential unit vector. The length of the edge pixel trajectory. Reference scale for the distance between measurement points This refers to the preset balance coefficient. The distribution density coefficient of the measurement points is calculated as follows: , This refers to the average distance between measuring points; S413: Based on the measured point normal offset sequence, analyze its distribution relationship in the edge line of the same component surface, compare the offset changes of different edge line segments, determine the arrangement of the edge line reference order in each edge line segment, adjust the edge line reference order, and obtain the edge line deviation reference rate.

[0011] The present invention is improved in that the step of obtaining the boundary positioning association result is specifically as follows: S511: Based on the deviation rate of the edge line from the reference, analyze the deviation direction, calculate the displacement direction vector of the measurement marker point relative to the reference boundary, determine the orientation category of the displacement direction in the image coordinate system, unify the coordinate structure, and obtain the offset direction representation vector. S512: Based on the offset direction characterization vector, compare the positional relationship between the boundary extension path and the edge line of the measured component surface, determine the segment position of the measurement marker points distributed along the boundary direction, identify the corresponding segment identifier, and obtain the boundary segment association identifier; S513: Based on the boundary segment association identifier, analyze and measure the correspondence of the marker point arrangement trajectory, compare the conflict states of different boundary attribution relationships, determine the priority order of the boundary corresponding to the marker point, adjust the boundary attribution arrangement order, and obtain the boundary positioning association result.

[0012] The present invention is improved in that the intersection position refers to the position of the pixel point where the coordinates of two or more structural edge pixel trajectories coincide, and the endpoint arrangement direction index is the direction vector of the line connecting the pixel coordinates at both ends in the image coordinate system.

[0013] An engineering measurement data analysis system based on image recognition, the system comprising: The image acquisition module is based on an industrial camera. It acquires the grayscale changes of the captured image, determines the structural edges corresponding to continuous abrupt change areas, compares the relationship between the intersection position and the marker outline, filters the coordinates of the measurement marker points, adjusts the positioning order of the calibration ruler end, and obtains the reference set of measurement point coordinates. Based on the coordinate reference set of the measuring points, the scale mapping module obtains the arrangement direction of the endpoints of the calibration ruler, compares the relationship between the lateral and longitudinal displacements, calculates the orientation of the ruler segment, adjusts the corresponding order of the actual length of the calibration ruler and the pixel span, determines the conversion relationship of each direction, and obtains the corresponding quantity of the scale in each direction. The spacing characterization module obtains the positional relationship of the points to be measured based on the corresponding quantities of the directional scale, calculates the lateral and longitudinal displacements of the two points to be measured, compares the direction mapping scale segments, adjusts the displacement conversion order, judges the relationship between the structure edge direction and the line connecting the measuring points, and obtains the measuring point spacing characterization set. The offset determination module analyzes the distribution of measurement point intervals based on the measurement point interval characterization set, determines the relationship between the measurement marker point and the normal interval of the structural edge, calculates the deviation state corresponding to the edge line of the same measured component, adjusts the edge line reference order, and obtains the edge line deviation reference rate. Based on the deviation rate of the edge line reference, the boundary positioning module analyzes the deviation direction, compares the extension direction of the reference boundary with the positional relationship of the edge line on the surface of the measured component, determines the corresponding segment of the trajectory of the measurement marker points, adjusts the boundary assignment order, and obtains the boundary positioning association result.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, the edge is defined based on gray-scale abrupt changes and a reference for measuring point positioning is constructed. The horizontal and vertical displacements of the calibration ruler are calculated to extract the corresponding scales in each direction. The interval characterization is established by coordinating the displacement of measuring points and the direction of the edge. The normal span and surface deviation are analyzed to form a reference for the edge line. Combined with the boundary extension orientation, the associated segments of the arrangement trajectory are defined. This effectively eliminates the drawbacks of subjective visual bias and operational fatigue, realizes automated and precise calculation of structural deformation characteristics, maintains the consistency of the reference height in multi-directional displacement conversion, blocks the error accumulation link caused by complicated intervention, and steadily improves the timeliness and objective accuracy of data output in engineering monitoring operations. Attached Figure Description

[0015] Figure 1 This is a flowchart of the main steps of the present invention; Figure 2 This is a flowchart illustrating the process of obtaining the reference set of measurement point coordinates in this invention. Figure 3 This is a flowchart illustrating the process of obtaining the corresponding quantities of the directional scale in this invention. Figure 4 This is a flowchart illustrating the process of obtaining the measurement point interval characterization set in this invention. Figure 5 This is a flowchart illustrating the process of obtaining the edge deviation reference rate in this invention. Figure 6 This is a flowchart illustrating the process of obtaining boundary positioning association results in this invention. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0017] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0018] All user-related information involved in this invention (including but not limited to biometric information, identity verification information, behavioral data, device information, and other data that can be used for identity verification and personalized services) is collected and processed with the user's full knowledge and voluntary consent. The use of data is limited to purposes necessary for providing the technical services of this invention, and reasonable technical and management measures will be taken to ensure the security and confidentiality of users' personal information in terms of information protection and privacy.

[0019] Example: Please refer to Figure 1 This invention provides a technical solution: an engineering measurement data analysis method based on image recognition, comprising the following steps: S1: Based on an industrial camera, acquire grayscale changes in the captured image, determine the structural edges corresponding to continuous abrupt change areas, compare the correspondence between the intersection position and the marker outline, filter the coordinates of the measurement marker points, and adjust the positioning order of the calibration ruler end to obtain the reference set of measurement point coordinates; S2: Based on the reference set of measuring point coordinates, obtain the arrangement direction of the endpoints of the calibration ruler, compare the correspondence between the lateral displacement and the longitudinal displacement, calculate the direction of the ruler segment, adjust the correspondence order between the actual length of the calibration ruler and the pixel span, determine the conversion relationship of each direction, and obtain the corresponding quantity of the sub-axis scale. S3: Based on the corresponding quantities of the directional scale, obtain the positional relationship of the points to be measured, calculate the lateral and longitudinal displacements of the two points to be measured, compare the scale segments of the direction mapping, adjust the displacement conversion order, determine the relationship between the structural edge direction and the line connecting the measuring points, and obtain the measuring point interval representation set. S4: Based on the measurement point interval characterization set, analyze the measurement point interval distribution, determine the relationship between the measurement marker point and the normal interval of the structural edge, calculate the deviation state corresponding to the edge line of the same measured component, adjust the edge line reference order, and obtain the edge line deviation reference rate. S5: Based on the deviation rate of the edge line reference, analyze the deviation direction, compare the relationship between the extension direction of the reference boundary and the position of the edge line on the surface of the measured component, determine the segment corresponding to the trajectory of the measurement marker points, adjust the boundary assignment order, and obtain the boundary positioning association result.

[0020] The measurement point coordinate reference set includes the measurement point pixel coordinate distribution, the marker point spatial index, and the calibration ruler endpoint coordinate combination. The corresponding quantities of the directional scales include the lateral scale conversion reference, the longitudinal scale conversion reference, and the direction decomposition identifier. The measurement point interval characterization set includes the measurement point spacing expression unit, the connection direction identifier, and the component spacing record item. The edge line deviation reference rate includes the normal offset ratio identifier, the lateral offset classification identifier, and the edge line reference matching item. The boundary positioning association results include the boundary matching relationship identifier, the positioning segment division identifier, and the boundary attribution correspondence identifier.

[0021] In S1, a continuous abrupt change region refers to a set of connected pixels formed after the gray-level difference between adjacent pixels exceeds the edge detection gradient threshold; a structural edge refers to the pixel trajectory of the outer contour line of a component composed of continuous abrupt change regions in the image coordinate system; an intersection position refers to the pixel position where the coordinates of two or more structural edge pixel trajectories coincide; a marker contour refers to the set of closed boundary pixels formed by measurement marker points in the image; marker point coordinates refer to the pixel row and column coordinates corresponding to the center position of the marker contour; and a calibration ruler refers to a ruler reference object with a known actual length and forming a clear boundary in the image.

[0022] In S2, the endpoint arrangement direction index is the direction vector of the line connecting the pixel coordinates at both ends of the ruler in the image coordinate system; the correspondence relationship refers to the proportional relationship between the horizontal pixel displacement and the vertical pixel displacement under the same ruler length constraint; the ruler segment direction assignment refers to the determination result of assigning the ruler pixel length to the horizontal component or the vertical component; the pixel span index is the Euclidean distance or directional distance between the pixel coordinates at both ends of the ruler; the conversion relationship of each direction refers to the proportional mapping between the horizontal pixel distance and the actual length, and the vertical pixel distance and the actual length.

[0023] In S3, the positional relationship of the points to be measured refers to the relative coordinate difference relationship between the two measurement markers in the image coordinate system; the direction mapping ratio refers to the conversion ratio between the horizontal pixel unit and the actual length, and the vertical pixel unit and the actual length; the displacement conversion order refers to the logical order in which the horizontal displacement and the vertical displacement are converted during the distance calculation process; the structural edge direction refers to the extension direction vector of the structural edge pixel trajectory in the image; and the measurement point connection refers to the straight line segment pixel path formed between the coordinates of the two measurement markers.

[0024] In S4, the measurement point interval distribution refers to the spatial distribution of distances between multiple measurement markers within the same component range; measurement markers refer to the pixel representation of manually set marks used for engineering measurement positioning in the image; structural edge refers to the set of boundary pixels of the component outline in the image; the surface of the measured component refers to the set of pixel pixels of the surface area corresponding to the engineering measurement object in the image; deviation state refers to the change in distance between the measurement markers and the normal direction of the structural edge; and edge reference order refers to the priority order of selecting reference edges for distance calculation among multiple structural edges.

[0025] In S5, the deviation direction refers to the direction vector of the displacement of the measurement marker relative to the reference boundary; the reference boundary refers to the representation of the structural edge or calibration boundary that serves as the measurement reference in the image; the edge position relationship refers to the spatial arrangement relationship between different structural edges in the image coordinate system; the segment corresponding to the arrangement trajectory refers to the mapping interval on the reference boundary after the measurement marker is distributed along a certain direction; the boundary assignment order refers to the judgment order used when the measurement marker or edge corresponds to the reference boundary; the boundary positioning association result refers to the positional correspondence information formed between the measurement marker, component edge, and reference boundary.

[0026] Please see Figure 2 The specific steps for obtaining the reference set of measurement point coordinates are as follows: S111: Based on an industrial camera, analyze the gray-level variation relationship between adjacent pixels in the pixel matrix of the acquired image, determine the continuous variation region formed by the gray-level difference, filter the pixel set with spatial connectivity, and calculate the extension direction of the pixel set boundary to obtain the pixel trajectory of the structural edge. Based on the grayscale matrix of the image output from the industrial camera, the grayscale values ​​of pixels are read row by row and column by column. The difference between the current pixel and the pixels to its right and below is calculated to obtain the horizontal and vertical grayscale differences. In the actual acquisition of a 1920×1080 image, the grayscale values ​​of adjacent pixels in a certain area are 120 and 158, respectively, so the difference is 38. This difference is then compared with a preset threshold, which is set based on the grayscale difference statistics of the on-site markings and the background. In the sample statistics, the grayscale difference in the edge area is concentrated between 30 and 50, and the grayscale difference in the non-edge area is concentrated between 5 and 15. Therefore, 22 is selected as the dividing value. When the difference is greater than or equal to 22, it is marked as a change point. Then, the adjacent images are checked according to the four-neighbor relationship (top, bottom, left, and right). If a pixel satisfies the condition, pixels that meet the condition and are adjacent in coordinates are grouped into the same set. For example, if 150 pixels that meet the condition appear consecutively in the edge region of a beam, they are numbered as set 1. If another set of pixels has only 2 pixels and they are not consecutive, they are directly removed. After filtering, the boundary pixels of each set are extracted, and the changes in the row and column directions of the boundary pixels are counted. If the cumulative column change reaches 80 and the row change is only 5 in 10 consecutive points, it is determined to be a horizontal extension. If the opposite is true, it is determined to be a vertical extension. If the column change is 95 and the row change is 8 in a set of edges, it is recorded as a horizontal edge. Finally, the boundary pixels are arranged in order to form a continuous pixel trajectory record, thus obtaining the structural edge pixel trajectory.

[0027] S112: Based on the pixel trajectory of the structural edge, determine the coordinates of the intersection position of the pixel trajectory, calculate the coordinates of the center position of the closed pixel region of the marker outline, compare the spatial position relationship between the intersection coordinates and the center coordinates, filter the matching region pixel coordinates, and obtain the set of marker point position coordinates; Based on the acquired multiple edge pixel trajectories, the pixel coordinates in each trajectory are read one by one. The coordinates of two trajectories are compared one by one. When a point has completely identical row and column coordinates, it is directly recorded as the intersection. For example, if a point (420, 290) exists in one trajectory and the same point exists in another trajectory, it is recorded as the intersection. If they are not completely identical, the row difference and column difference between the two points are compared. When the row difference and column difference do not exceed 1, the average of the two points is taken as the intersection. For example, points (421, 289) and (421, 290) can be recorded as (421, 289.5). Then, the closed marked regions in the image are processed. First, the sum of the row and column coordinates of all pixels in the region is calculated, and then divided by... The center position is obtained by counting the number of pixels. In a certain circular marker, the number of pixels is counted as 200, the total row coordinates are 84000, and the total column coordinates are 58000. The center is approximately (420, 290). Then, the intersection position is compared with the center coordinates. When the row difference is within the range of 0 to 2 and the column difference is within the range of 0 to 2, it is considered a match. For example, the difference between the intersection point (421, 289.5) and the center point (420, 290) is 1 and 0.5 respectively, so the point is kept. If the difference exceeds 5, it is directly rejected. After filtering, the boundary of the area is checked to see if it is complete. If the gap length is no more than 2 pixels, it is kept; otherwise, it is rejected. The coordinates that meet the conditions are sorted and registered to obtain the set of marker point position coordinates.

[0028] S113: Based on the coordinate set of the marker point position, calculate the direction of the line connecting the pixel coordinates of the calibration ruler endpoints, determine the consistency of the endpoint arrangement direction, adjust the spatial position corresponding to the endpoint order, and obtain the reference set of the measuring point coordinates. First, identify the coordinates of the marker points at both ends of the calibration ruler, record their row and column coordinates, and calculate the difference between the two points. In a set of data, the endpoint coordinates are (512, 168) and (514, 468), the column difference is 300, and the row difference is 2, thus indicating a horizontal arrangement. Next, perform a consistency check on multiple consecutive frames, calculating the endpoint difference for each frame. When the column difference for four consecutive frames is between 290 and 310 and the row difference is between 0 and 10, the direction is considered consistent. If the column difference drops to 260 or the row difference rises to 20 in a particular frame, it is marked as abnormal and removed. For example, if the column difference in the fifth frame is 260, it will not participate in subsequent processing. The remaining frames... The endpoint order is standardized. If arranged horizontally, the point with the smaller column coordinate is set as the starting point and the larger one as the ending point. For example, (512, 168) is the starting point and (514, 468) is the ending point. If the order is reversed, they are swapped. Then, all the coordinates of the marker points and the standardized endpoint coordinates are recorded together, sorted by row coordinates from smallest to largest, and then sorted by column coordinates to form a standardized coordinate record table. This record table contains 6 measuring points and 2 endpoints. Each record is checked for duplicate records or missing coordinates. If duplicates are found, the first record is retained. If missing data is found, the data is deleted. This process is used to form a standardized set of measuring point coordinate references.

[0029] Please see Figure 3 The specific steps for obtaining the corresponding quantities at the directional scale are as follows: S211: Based on the reference set of measurement point coordinates, analyze the spatial positional relationship of pixel coordinates at the endpoints of the calibration ruler, calculate the difference between endpoint coordinates to form a connecting direction vector, determine the direction category of the connecting direction in the image coordinate system, adjust the arrangement order of endpoint coordinates and the consistency relationship with the direction, and obtain the endpoint direction representation vector. Based on the coordinates of the measuring points and the coordinates of both ends of the calibration scale in the record, the row and column coordinates of the endpoints were retrieved one by one. The row difference and column difference were obtained by subtracting the front coordinate from the back coordinate. In a set of data, the endpoints were (515, 168) and (517, 469), resulting in a row difference of 2 and a column difference of 301. Then, the absolute values ​​of the two differences were compared within the specified range. Cases with a column difference between 250 and 350 and a row difference between 0 and 15 were classified into the horizontal category; cases with a row difference between 250 and 350 and a column difference between 0 and 15 were classified into the vertical category; and cases with both differences between 80 and 250 were classified into the diagonal category. In this example, the column difference of 301 fell into the horizontal range and the row difference of 2 fell into the low-value range, so it was registered as horizontal. Then, the endpoints were compared sequentially. The sequence is corrected by reading the column coordinates of the two endpoints. If the current direction is horizontal, the point with the smaller column coordinate is set as the front end and the point with the larger column coordinate is set as the back end. If the order is reversed, they are directly swapped. For example, if the original record is (517, 469) first and (515, 168) second, it is rewritten as (515, 168) first and (517, 469) second after the swap. Then, a consistency check is performed on three consecutive frames of data. If two or more of the three frames have the same direction, the direction is retained. Otherwise, the difference is averaged and re-determined. In the sample, the column differences of the three frames are 301, 299 and 304, all of which fall into the horizontal range. Therefore, the horizontal record is retained. Finally, the row difference, column difference, direction category and endpoint order are uniformly registered to obtain the endpoint direction representation vector.

[0030] S212: Based on the endpoint direction representation vector, compare the correspondence between horizontal and vertical pixel displacements, calculate the proportional relationship of displacement components under the same calibration scale constraint, and determine the distribution state of pixel length in each direction to obtain the direction distribution structure identifier. Based on the row and column differences in the endpoint direction records, the two values ​​are first compared to determine the primary and secondary displacements. In a set of records, the column difference is 301 and the row difference is 2. Therefore, 301 is recorded as the primary displacement and 2 as the secondary displacement. The ratio between the two is then calculated by dividing 301 by 2, resulting in 150.5. The data is then categorized according to the ratio range: a ratio greater than 20 is classified as a unidirectional dominant range; a ratio between 3 and 20 is classified as a biased range; and a ratio less than 3 is classified as a bidirectional convergent range. In this example, 150.5 is greater than 20, therefore it is registered as unidirectional dominant. The total span of this set of data is then read, and the two differences are summed to obtain approximately 303. The proportion of the primary displacement to the total span is calculated as 301 divided by 303, approximately equal to 0.99. Then... Based on the proportion range, when the proportion is greater than 0.85, it is registered as concentrated in the main direction; when the proportion is between 0.60 and 0.85, it is registered as partially concentrated; and when the proportion is less than 0.60, it is registered as dispersed. In this example, 0.99 falls into the concentration range, so it is marked as concentrated in the main direction. Then, the attribution judgment is performed based on the direction category. If the preceding record is horizontal and the main displacement comes from the column difference, it is registered as horizontal priority allocation, and the auxiliary direction is registered as vertical supplementation. In another set of data, the column difference is 180, the row difference is 120, and the proportion is 1.5, so it is classified as bidirectional proximity. At the same time, the proportion of the main displacement is about 0.6, so it is registered as partially concentrated. Finally, the main displacement value, auxiliary displacement value, proportion result and direction attribution order are uniformly written into the record to form the direction allocation structure identifier.

[0031] S213: Based on the orientation allocation structure identifier, analyze the actual length of the calibration scale and the pixel span, calculate the mapping relationship between the horizontal pixel distance and the actual length, determine the correspondence between the vertical pixel distance and the actual length, adjust the orientation conversion order, and obtain the corresponding quantity of the orientation scale. Based on the direction allocation results and the actual length data of the calibration ruler, the known length of the calibration ruler is read, for example, the length of the field ruler is 300 mm. Then, the corresponding pixel span is read. In the horizontal dominant record, the column difference is 301 and the row difference is 2. First, 300 and 301 are mapped to obtain a length of approximately 0.997 mm per pixel. This value is then recorded as the horizontal conversion benchmark. Subsequently, the row direction is checked. Since the row difference of 2 is in the range of 0 to 5, the horizontal conversion value is directly used synchronously for the row direction as the initial conversion value. In another set of diagonal data, the column difference is 180 and the row difference is 120. First, the bidirectional processing order is determined according to the direction allocation identifier. Then, the column and row conversion values ​​are calculated separately. The column direction is approximately 1 when 300 is divided by 180. The pixel length is 67 mm. The row length is calculated by dividing 300 by 120 to obtain approximately 2.5 mm per pixel, which are then recorded as the horizontal and vertical conversion bases. Subsequently, the order is adjusted. When the direction allocation is dominated by a single direction, the conversion value of the main direction is recorded first, followed by the conversion value of the auxiliary direction. When the two directions are close, the horizontal direction is prioritized and then the vertical direction is recorded. In the measurement point calculation example, the column difference between two points is 30 pixels and the row difference is 4 pixels. After converting with 0.997, the results are approximately 29.9 mm and 4.0 mm. In another example, the column difference is 24 pixels and the row difference is 18 pixels. Therefore, the results are approximately 40 mm and 45 mm after converting with 1.67 and 2.5, respectively. Finally, the actual length, pixel span, and corresponding conversion values ​​are recorded in order to obtain the corresponding quantities of the directional scale.

[0032] Please see Figure 4 The specific steps for obtaining the measurement point interval characterization set are as follows: S311: Based on the corresponding quantities of the directional scales, analyze the mapping relationship between the horizontal scale conversion benchmark and the vertical scale conversion benchmark, calculate the difference of the pixel coordinates of the point to be measured to form the horizontal and vertical displacements, determine the relative positional relationship of the coordinates of the two points in the image coordinate system, and obtain the displacement relationship vector of the point to be measured. Retrieve the row and column coordinates of the two points to be measured from the measurement point record table. Then, subtract the coordinates of the previous point from the coordinates of the second point to obtain the column difference and row difference. In the example of two measurement marks on the steel beam surface, the coordinates of the first point are (420, 286), and the coordinates of the second point are (424, 352). The column difference is 66, and the row difference is 4. Then, 66 is assigned to the lateral displacement, and 4 is assigned to the longitudinal displacement. The numerical values ​​are checked by calling the previously determined conversion benchmark. If the lateral conversion value is 0.997 and the longitudinal conversion value is 0.997, the actual lateral displacement is recorded as 65.8, and the actual longitudinal displacement is recorded as 4.0. Then, the relative positions of the two points in the image coordinate system are judged. First, compare the column coordinates. If the column coordinate of the second point is greater than that of the first point, it is recorded as a right offset. If the column coordinate of the second point is less than that of the first point, it is recorded as a left offset. The offset is then compared with the row coordinates. If the row coordinate of the later point is greater than that of the previous point, it is recorded as a downward offset; if the row coordinate of the later point is less than that of the previous point, it is recorded as an upward offset. In this example, the later point relative to the previous point shows a rightward offset accompanied by a downward offset. If the column difference is in the range of 0 to 5, it is recorded as a small horizontal displacement; if the column difference is in the range of 6 to 50, it is recorded as a normal horizontal displacement; if the column difference is greater than 50, it is recorded as a large horizontal displacement. In this example, 66 falls into the large displacement range, and the row difference 4 falls into the small displacement range. Then, the horizontal displacement value, vertical displacement value, left and right position indicators, and up and down position indicators are written into the same record in a unified order. In another set of points, if the column difference is 12 and the row difference is 28, it is written as a normal rightward displacement and a normal downward displacement, respectively. The displacement value and position category of each set of points to be measured are uniformly recorded to obtain the displacement relationship vector of the measuring points.

[0033] S312: Based on the displacement relationship vector of the measuring point, analyze the distribution relationship of lateral and longitudinal displacement in the direction mapping scale section, compare the corresponding relationship of conversion benchmarks in each direction, determine the direction section to which the displacement component belongs, adjust the conversion order of lateral and longitudinal displacement, and obtain the direction conversion sequence identifier. The conversion datum corresponding to each of the two displacements is read in groups, and then the two displacements are placed into preset scale ranges for classification. Scale ranges can be categorized as follows: 0 to 5 are small ranges, 6 to 30 are regular ranges, 31 to 80 are extended ranges, and greater than 80 are large-span ranges. In a set of data, the lateral displacement is 65.8 and the longitudinal displacement is 4.0. First, 65.8 is registered in the extended range, and then 4.0 is registered in the small range. Then, the conversion datum sources corresponding to these two displacements are compared. If both use the same conversion value, they are registered as having the same datum relationship. If the lateral and longitudinal displacements use different conversion values, they are registered as having separate datum relationships. In the example of an oblique component, each pixel in the lateral direction corresponds to 1.67, and each pixel in the longitudinal direction corresponds to 2.5. The column difference between two points is 24, and the row difference is 18. After conversion, the lateral value is 40.0 and the longitudinal value is 45.0. If both values ​​fall into the extended segment, they are registered as sub-base relationships. Then, displacement attribution judgment is performed. If the lateral value is greater than the longitudinal value by more than 5, it is written as a lateral priority segment. If the longitudinal value is greater than the lateral value by more than 5, it is written as a longitudinal priority segment. If the difference between the two is in the range of 0 to 5, it is written as a bidirectional approach segment. In this oblique example, the difference between 40.0 and 45.0 is 5, and it is registered as a bidirectional approach segment. In the beam example, the difference between 65.8 and 4.0 is 61.8, and it is registered as a lateral priority segment. Then, the conversion order is adjusted according to the attribution result. When lateral priority is used, the lateral is recorded first and then the longitudinal is recorded. When longitudinal priority is used, the longitudinal is recorded first and then the lateral is recorded. When bidirectional approach is used, it is written in the order of column first and then row in the image. Finally, the segment category, base relationship, priority direction and conversion order are sorted and written into the record table to form a direction conversion sequence identifier.

[0034] S313: Based on the direction conversion sequence identifier, analyze the extension direction of the pixel trajectory of the structure edge, calculate the relationship between the direction vector of the measurement point connection and the direction of the structure edge, determine the spatial correspondence between the measurement point connection and the structure edge, adjust the expression of the measurement point spacing and the correlation with the direction identifier, and obtain the measurement point spacing representation set. Retrieve the structural edge pixel sequence adjacent to the test point from the edge trajectory data. Sequentially read the row and column changes of consecutive points in this sequence. If the cumulative column change of 20 consecutive edge points is above 60 and the cumulative row change is below 10, it is registered as a lateral extension. If the cumulative row change is above 60 and the cumulative column change is below 10, it is registered as a longitudinal extension. If both are between 20 and 60, it is registered as a diagonal extension. In a set of beam edge records, the cumulative column change is 92 and the cumulative row change is 7, so it is initially written as a lateral extension. Then, read the coordinates of two test points and calculate the column difference and row difference of the line connecting the test points. In this example, the coordinates of the two test points are (420, 286) and (424, 352), and the line shows a column difference of 66 and a row difference of 4. Therefore, the direction of the line is recorded as lateral dominance. Then, put the line direction and the edge direction into the same comparison sequence. If the two direction categories are... If they are the same, they are registered as a same-direction relationship. If one is horizontal and the other is vertical, they are registered as an intersecting relationship. If one is diagonal and the other is horizontal or vertical, they are registered as a deflection relationship. In this example, the connecting line and the edge are both horizontal, so it is written as a same-direction relationship. In another set of data, the edge extends vertically, the column difference of the measuring point is 8, and the row difference is 54, so it is also written as a same-direction relationship. If the cumulative change of the edge column direction is 10 and the cumulative change of the row direction is 88, while the column difference of the measuring point connecting line is 43 and the row difference is 6, it is written as an intersecting relationship. Then the expression of the measuring point spacing is adjusted. When encountering a same-direction relationship, the main direction spacing value is written first and then the auxiliary direction spacing value is written. When encountering an intersecting relationship, the connecting line direction is written first and then the edge direction is written. When encountering a deflection relationship, the connecting line value is written first and then the deflection mark is added. Finally, the measuring point number, horizontal spacing, vertical spacing, connecting line direction, edge relationship, and spatial corresponding category are uniformly registered to obtain the measuring point interval representation set.

[0035] Please see Figure 5 The specific steps for obtaining the edge deviation reference rate are as follows: S411: Based on the measurement point interval characterization set, calculate the distribution density of the measurement mark point spacing in the component surface area, determine the belonging relationship of adjacent measurement point combinations within the same component range, adjust the expression form of measurement point spacing to unify the spatial structure, and obtain the measurement point spacing distribution density; The horizontal and vertical spacing values ​​of each pair of measuring points are read one by one. The two values ​​are added together to obtain the total spacing of the measuring point pair. Then, the spatial range is defined according to the area where the measuring point is located. For example, a 100×100 mm area on the surface of the component is selected as the statistical unit. All measuring point pairs are retrieved within this area. In this example, there are 4 pairs of measuring points in this area, with total spacings of 60, 45, 70, and 35 respectively, resulting in a cumulative value of 210. The area of ​​the region (10000) is then compared with the cumulative spacing of 210 to obtain a density value of approximately 0.021. The density value is then compared with a preset range. Values ​​between 0 and 0.02 are registered as sparse areas, values ​​between 0.02 and 0.05 are registered as medium areas, and values ​​greater than 0.05 are registered as dense areas. In this example, 0.021 is classified as a medium area. Check each pair of measuring points to see if they belong to the same component range. First, read the component boundary coordinate range, for example, row range 400 to 600 and column range 200 to 500. Then, compare the coordinates of each measuring point with this range item by item. If both points meet the range conditions, they are registered as a combination of the same component. If any point exceeds the range, it is registered as a combination of different components. In the example, points (420, 286) and (424, 352) meet the range conditions, so they are registered as a combination of the same component. Then, unify the order of expressing the measuring point spacing. When the horizontal spacing is more than 10 greater than the vertical spacing, record the horizontal spacing first. When the vertical spacing is more than 10 greater than the horizontal spacing, record the vertical spacing first. When the difference is in the range of 0 to 10, write it in the column-to-row order. Finally, record the density value, interval category, component affiliation and expression order in a unified manner to obtain the measuring point spacing distribution density.

[0036] S412: Based on the distribution density of measuring point spacing, analyze the projected distance from the measuring marker point to the normal direction of the structural edge, compare the differences in the normal direction distance of different measuring points, and determine the consistency relationship between the measuring point offset direction and the edge normal direction, using the formula: ; The sequence of normal offsets at the measurement points is obtained, where, The edge offset metric indicates the degree of offset of the measuring point relative to the edge of the component. The lateral displacement of the measuring point refers to the difference in horizontal coordinates between the actual and reference positions of the measuring point. The longitudinal displacement of the measuring point refers to the vertical coordinate difference between the actual position and the reference position of the measuring point. The horizontal component of the edge normal unit vector represents the projection of the unit vector perpendicular to the edge of the component onto the horizontal axis. The longitudinal component of the edge normal unit vector represents the projection of the unit vector perpendicular to the edge direction onto the vertical axis. The lateral component of the edge tangential unit vector represents the projection of the unit vector parallel to the edge direction onto the horizontal axis. The longitudinal component of the edge tangential unit vector represents the projection of the unit vector parallel to the edge direction onto the vertical axis. It refers to the length of the pixel trajectory at the edge, that is, the total length of the pixel trajectory occupied by the edge of the component in the image. The reference scale for the spacing between measuring points is used as a benchmark length for normalization. It is usually taken as the standard spacing or average spacing between measuring points. This refers to a preset balance coefficient used to adjust the weights between the normal offset term and the tangential-edge coupling term. It is dimensionless and determined through engineering experience. The measurement point distribution density coefficient is used to dynamically adjust the influence of edge length based on the actual density of measurement points. Its calculation method is as follows: , The average distance between measuring points refers to the arithmetic mean of the distances between all adjacent measuring points within the same component area; The edge offset metric is used to characterize the degree of normal offset of the measurement marker point relative to the edge of the structure at a uniform scale and its coupling relationship with the edge orientation. It also serves as the basis for constructing the normal offset sequence of measurement points and calculating the edge deviation reference rate. Based on the distribution density of the measuring points, two measuring points corresponding to the edge line of the same component surface are selected for calculation. First, the original displacement data and edge direction data of the measuring points are retrieved, and the lateral displacement of the measuring points is calculated. Longitudinal displacement The minimum-maximum normalization process is used to calculate the edge pixel trajectory length. Reference scale for measuring point spacing and average measuring point spacing The preset engineering range is normalized, where the original data corresponding to measuring point 1 is... , , , , The original data corresponding to measuring point 2 is , , , , The original displacements of the two measuring points are used as the minimum-maximum normalized interval: The minimum value is 28, the maximum value is 30, and the normalized result for measuring point 1 is: The normalized result for measurement point 2 is ; The minimum value is 2, the maximum value is 3, and the normalized result for measurement point 1 is: The normalized result for measurement point 2 is ; Normalized within a preset range of 100 to 300 ; and Normalized within a preset range of 20 to 40 ; edge normal unit vector components , ; Edge tangential unit vector components , ; Balance coefficient ,calculate ; Substitute the normalized result into the formula, and substitute measurement point 1 into the formula. , Then, the normal term is calculated as follows: ; The tangential term is calculated as follows: ; The square of the tangential term is Marginal terms are calculated as The result within the square root is: The result of the prescription is ; Multiply by the balance factor to ; Molecular results Divide by ,get: ; Substitute measurement point 2 , Then, the normal term is calculated as follows: ; The tangential term is calculated as follows: ; The square of the tangential term is The marginal term remains at 0.25, and the result within the square root is: The result of the prescription is ; Multiply by the balance factor to ; Molecular results Divide by ,get: ; The calculation results from the two measuring points are combined into a sequence. The normal offset sequence of the measuring points is obtained.

[0037] Will Compare with a preset interval, where: Dividing the measurement point into segments close to the edge line indicates that the normal offset is in a slight range; Dividing it into a boundary transition section indicates that the normal offset of the measuring point is in a gradually changing range; Dividing it into edge line offset segments indicates that the normal offset of the measuring point is within a defined range; The boundary line is divided into anomaly segments, indicating that the normal offset of the measuring point is in a concentrated range; This measurement point 1 corresponds to Measurement point 2 corresponds to All satisfy The results indicate that both measuring points are located in the edge offset segment relative to the structural edge, and the offset of measuring point 2 is greater than that of measuring point 1. This numerical result, used as the single-point normal offset, is incorporated into subsequent sequence processing, connecting the corresponding measuring points within the same component area. Formed by combining the measuring points in the order of arrangement. Then, the input quantities are adjusted by grouping the edge segment numbers and using them as the reference order of the edge lines, thereby obtaining the normal offset sequence of the measurement points.

[0038] S413: Based on the normal offset sequence of the measuring points, analyze their distribution relationship in the edge lines of the same component surface, compare the changes in the offset of different edge line segments, determine the arrangement of the edge line reference order in each edge line segment, adjust the edge line reference order, and obtain the edge line deviation reference rate. The offset values ​​of each measuring point are grouped according to the edge line number. The offset value sequence is read within one edge line, for example, recorded as 3, 4, 6, 5, 7. This sequence is arranged in the sampling order, and the difference between adjacent values ​​is calculated one by one, resulting in differences of 1, 2, 1, 2. These differences are then compared with a preset interval. A difference between 0 and 1 is recorded as a stable change, between 2 and 4 as a gradual change, and greater than 4 as an abrupt change. In this example, a stable, gradually changing, stable, gradually changing sequence is formed. The same process is then performed on the other edge line. If the offset values ​​are 10, 14, 18, 25, the difference is 4, 4, 7, corresponding to gradually changing, gradually changing, and abrupt change. The average offset value of the two edge lines is then calculated: 5 for the first line and 16.75 for the second. The difference between the two values, 11.75, is compared with a threshold of 5. When the difference is greater than 5, it is registered as a difference zone; when it is less than or equal to 5, it is registered as a proximity zone. In this example, it is classified as a difference zone. Then, the reference order is determined according to the size of the average value. The edge with the smaller average value is set as the priority reference, and the one with the larger average value is set as the secondary reference. In this example, the first edge has priority. Then, the offset value of each point of the secondary edge is compared with the average value of the priority edge. For example, an offset of 18 corresponds to a ratio of approximately 3.6. The ratio is then compared with the intervals 1 to 2, 2 to 4, and greater than 4, and registered as mild, moderate, and severe deviations, respectively. In this example, it is registered as a moderate deviation. The edge deviation classification, reference order, and ratio interval are recorded uniformly to obtain the edge deviation reference rate.

[0039] Please see Figure 6 The specific steps for obtaining the boundary localization association results are as follows: S511: Based on the deviation rate of the edge line from the reference, analyze the deviation direction, calculate the displacement direction vector of the measurement marker point relative to the reference boundary, determine the orientation category of the displacement direction in the image coordinate system, unify the coordinate structure, and obtain the offset direction representation vector. The coordinates of the measurement marker points and the nearest reference boundary point are read point by point. The coordinates of the marker points and the reference boundary points are then subtracted to obtain the column difference and row difference. In a set of steel component surface data, the marker point coordinates are (428, 315), and the reference boundary point coordinates are (424, 302). Therefore, the column difference is 13, and the row difference is 4. Then, the direction is determined according to the image coordinates: a column difference greater than 0 is recorded as rightward, and a column difference less than 0 is recorded as leftward; a row difference greater than 0 is recorded as downward, and a row difference less than 0 is recorded as upward. In this example, it is written as downward rightward. The difference values ​​are then divided into intervals; column or row differences between 0 and 3 are recorded as slight deviation. The offset is recorded as a regular offset between 4 and 15, and a large offset greater than 15. In this example, column 13 is recorded as regular, and row 4 is recorded as regular. Then, the coordinate structure is unified, and all points are written into the record in the order of column first, then row, and left and right first, then top and bottom. If the original record order is row first, then column, it is directly rewritten. In another set of data, the marker point is (410, 288) and the reference point is (418, 290). The column difference is -2 and the row difference is -8. It is recorded as the top left direction and written as a combination of micro offset and regular offset. Finally, the left and right categories, top and bottom categories, offset levels and unified coordinate order of each marker point are merged and recorded to obtain the offset direction representation vector.

[0040] S512: Based on the offset direction characterization vector, compare the positional relationship between the boundary extension path and the edge line of the measured component surface, determine the segment position of the measurement marker points distributed along the boundary direction, identify the corresponding segment identifier, and obtain the boundary segment association identifier; Obtain the coordinates of continuous edge points of the reference boundary. Calculate the cumulative column and row changes of the boundary extension path according to the sampling order. If the cumulative column change is greater than 60 and the cumulative row change is less than 10, it is registered as a lateral extension. If the cumulative row change is greater than 60 and the cumulative column change is less than 10, it is registered as a longitudinal extension. In a set of boundary data, the cumulative column change of 20 consecutive edge points is 88 and the cumulative row change is 6, so it is initially recorded as a lateral extension. Then, read the edge line position of the measured component surface, convert the projection position of the marker point on the boundary into the boundary length sequence, and then divide the boundary into segments according to the total boundary length, with lengths from 0 to 100 recorded as the starting point. The segments are divided into middle segments (101 to 300) and end segments (greater than 300). In the example, the cumulative position of a certain marker point on the boundary is 126, so it is registered as a middle segment. The offset direction of the marker point is then checked in parallel with the extension direction of the boundary. If the boundary extends horizontally and the position of the marker point along the boundary increases continuously, the current segment identifier is retained. If there is a jump of more than 50, the adjacent segment is checked. In another set of data, the cumulative position changes from 92 to 148, a difference of 56, so the check is performed and it is rewritten as a middle segment. Finally, the boundary extension category, segment position, continuous status of adjacent points and segment number are written into the record in a unified format to obtain the boundary segment association identifier.

[0041] S513: Based on the boundary segment association identifier, analyze and measure the correspondence between the arrangement trajectory of the marker points, compare the conflict status of different boundary attribution relationships, determine the priority order of the boundary corresponding to the marker points, adjust the boundary attribution arrangement order, and obtain the boundary positioning association result; Read the trajectory of adjacent measuring points according to the marker point number, and compare the continuity of the boundary segment to which each point belongs. If three or more consecutive marker points in the same group fall into the same boundary and the segment numbering is progressive from the start segment to the middle segment to the end segment, it is registered as a continuous single boundary. In a set of component edge data, P1, P2, and P3 fall into the start segment, middle segment, and middle segment of boundary A, respectively. Therefore, they are first written as continuous boundary A. Then, compare the conflict status of different boundaries. If the same marker point corresponds to both boundary A and boundary B, read the deviation reference rate of the two boundaries. The smaller value takes priority. If the difference is between 0 and 0.2, continue to compare the closest distance between the point and the two boundaries. Prioritize points with smaller distances. In this example, the reference rate of boundary A for a certain point is 1.3, and the reference rate of boundary B is 1.6. Therefore, boundary A is directly set as the priority. If the reference rate of boundary C for another point is 1.4 and the reference rate of boundary D is 1.5, with a difference of 0.1, then the distance is compared again. The boundary with a distance of 8 takes precedence over the boundary with a distance of 12. Then, the boundary assignment order is adjusted according to the priority result, with the priority boundary written first and the conflict boundary written last. If more than half of the points in the same trajectory belong to the same boundary, then the entire group of trajectories is uniformly arranged according to that boundary. Finally, the marker number, boundary number, segment order, conflict status and priority order are merged and registered to obtain the boundary positioning association result.

[0042] An engineering measurement data analysis system based on image recognition, the system comprising: The image acquisition module is based on an industrial camera. It acquires the grayscale changes of the captured image, determines the structural edges corresponding to continuous abrupt change areas, compares the relationship between the intersection position and the marker outline, filters the coordinates of the measurement marker points, adjusts the positioning order of the calibration ruler end, and obtains the reference set of measurement point coordinates. The scale mapping module obtains the alignment direction of the calibration ruler endpoints based on the measurement point coordinate reference set, compares the relationship between the lateral and longitudinal displacements, calculates the orientation of the ruler segment, adjusts the corresponding order of the actual length of the calibration ruler and the pixel span, determines the conversion relationship of each direction, and obtains the corresponding quantity of the sub-scale. The spacing characterization module obtains the positional relationship of the points to be measured based on the corresponding quantities of the directional scale, calculates the lateral and longitudinal displacements of the two points to be measured, compares the scale segments of the direction mapping, adjusts the displacement conversion order, judges the relationship between the direction of the structure edge and the connection of the measuring points, and obtains the measuring point spacing characterization set. The offset determination module analyzes the distribution of measurement point intervals based on the measurement point interval characterization set, determines the relationship between the measurement marker point and the normal interval of the structural edge, calculates the deviation state corresponding to the edge line of the same measured component, adjusts the edge line reference order, and obtains the edge line deviation reference rate. The boundary positioning module analyzes the deviation direction based on the edge deviation reference rate, compares the positional relationship between the extension direction of the reference boundary and the edge line of the measured component surface, determines the corresponding segment of the trajectory of the measurement marker points, adjusts the boundary assignment order, and obtains the boundary positioning association result.

[0043] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A method for analyzing engineering measurement data based on image recognition, characterized in that, Includes the following steps: S1: Based on an industrial camera, acquire grayscale changes in the captured image, determine the structural edges corresponding to continuous abrupt change areas, compare the relationship between the intersection position and the marker outline, filter the coordinates of the measurement marker points, adjust the positioning sequence of the calibration ruler end, and obtain the reference set of measurement point coordinates; The specific steps for obtaining the reference set of the measuring point coordinates are as follows: S111: Based on an industrial camera, analyze the gray-level variation relationship between adjacent pixels in the pixel matrix of the acquired image, determine the continuous variation region formed by the gray-level difference, filter the pixel set with spatial connectivity, and calculate the extension direction of the pixel set boundary to obtain the pixel trajectory of the structural edge. S112: Based on the pixel trajectory of the structure edge, determine the coordinates of the intersection position of the pixel trajectory, calculate the coordinates of the center position of the closed pixel region of the marker contour, compare the spatial position relationship between the intersection coordinates and the center coordinates, filter the matching region pixel coordinates, and obtain the set of marker point position coordinates; S113: Based on the set of coordinates of the marker points, calculate the direction of the line connecting the pixel coordinates of the calibration ruler endpoints, determine the consistency of the endpoint arrangement direction, adjust the spatial position corresponding to the endpoint order, and obtain the reference set of measurement point coordinates. S2: Based on the coordinate reference set of the measuring points, obtain the arrangement direction of the endpoints of the calibration ruler, compare the relationship between the lateral displacement and the longitudinal displacement, calculate the direction of the ruler segment, adjust the corresponding order of the actual length of the calibration ruler and the pixel span, determine the conversion relationship of each direction, and obtain the corresponding quantity of the sub-scale. The specific steps for obtaining the corresponding quantities of the directional scale are as follows: S211: Based on the reference set of measurement point coordinates, analyze the spatial positional relationship of pixel coordinates at the endpoints of the calibration ruler, calculate the endpoint coordinate difference to form a connecting direction vector, determine the direction category of the connecting direction in the image coordinate system, adjust the arrangement order of endpoint coordinates and the consistency relationship with the direction, and obtain the endpoint direction representation vector. S212: Based on the endpoint direction representation vector, compare the correspondence between the horizontal pixel displacement and the vertical pixel displacement, calculate the proportional relationship of the displacement components under the same calibration scale constraint, and determine the distribution state of the pixel length in each direction to obtain the direction distribution structure identifier. S213: Based on the orientation allocation structure identifier, analyze the actual length of the calibration scale and the pixel span, calculate the mapping relationship between the horizontal pixel distance and the actual length, determine the correspondence between the vertical pixel distance and the actual length, adjust the orientation conversion order, and obtain the corresponding quantity of the orientation scale. S3: Based on the corresponding quantities of the directional scale, obtain the positional relationship of the points to be measured, calculate the lateral and longitudinal displacements of the two points to be measured, compare the direction mapping scale segments, adjust the displacement conversion order, determine the relationship between the structural edge direction and the line connecting the measuring points, and obtain the measuring point interval characterization set. The specific steps for obtaining the measurement point interval characterization set are as follows: S311: Based on the corresponding quantities of the directional scales, analyze the mapping relationship between the horizontal scale conversion reference and the vertical scale conversion reference, calculate the difference of pixel coordinates of the point to be measured to form the horizontal and vertical displacements, determine the relative positional relationship of the coordinates of the two points in the image coordinate system, and obtain the displacement relationship vector of the point to be measured. S312: Based on the displacement relationship vector of the measuring point, analyze the distribution relationship of lateral displacement and longitudinal displacement in the direction mapping scale section, compare the corresponding relationship of each direction conversion benchmark, determine the direction section to which the displacement component belongs, adjust the conversion order of lateral displacement and longitudinal displacement, and obtain the direction conversion sequence identifier. S313: Based on the direction conversion sequence identifier, analyze the extension direction of the pixel trajectory of the structure edge, calculate the relationship between the direction vector of the measurement point connection and the direction of the structure edge, determine the spatial correspondence between the measurement point connection and the structure edge, adjust the measurement point spacing expression and the correlation with the direction identifier, and obtain the measurement point spacing representation set. S4: Based on the measurement point interval characterization set, analyze the measurement point interval distribution, determine the relationship between the measurement marker point and the normal interval of the structural edge, calculate the deviation state corresponding to the edge line of the same measured component, adjust the edge line reference order, and obtain the edge line deviation reference rate. The specific steps for obtaining the edge deviation reference rate are as follows: S411: Based on the measurement point interval characterization set, calculate the distribution density of the measurement mark point spacing in the component surface area, determine the belonging relationship of adjacent measurement point combinations within the same component range, adjust the expression form of measurement point spacing to unify the spatial structure, and obtain the measurement point spacing distribution density; S412: Based on the distribution density of the measuring point spacing, analyze the projected distance from the measuring marker point to the normal direction of the structural edge, compare the differences in the normal direction distance of different measuring points, and determine the consistency relationship between the measuring point offset direction and the edge normal direction, using the formula: ； The sequence of normal offsets at the measurement points is obtained, where, The edge offset metric indicates the degree of offset of the measuring point relative to the edge of the component. The lateral displacement of the measuring point. The longitudinal displacement of the measuring point. The lateral component of the unit vector normal to the edge. The longitudinal component of the edge normal unit vector. The lateral component of the edge tangential unit vector. The longitudinal component of the edge tangential unit vector. The length of the edge pixel trajectory. Reference scale for the distance between measurement points This refers to the preset balance coefficient. The distribution density coefficient of the measurement points is calculated as follows: , This refers to the average distance between measuring points; S413: Based on the measured point normal offset sequence, analyze its distribution relationship in the edge line of the same component surface, compare the offset changes of different edge line segments, determine the arrangement of the edge line reference order in each edge line segment, adjust the edge line reference order, and obtain the edge line deviation reference rate. S5: Based on the deviation rate of the edge line, analyze the deviation direction, compare the relationship between the extension direction of the reference boundary and the position of the edge line on the surface of the measured component, determine the corresponding segment of the trajectory of the measurement marker points, adjust the boundary assignment order, and obtain the boundary positioning association result; The specific steps for obtaining the boundary positioning association result are as follows: S511: Based on the deviation rate of the edge line from the reference, analyze the deviation direction, calculate the displacement direction vector of the measurement marker point relative to the reference boundary, determine the orientation category of the displacement direction in the image coordinate system, unify the coordinate structure, and obtain the offset direction representation vector. S512: Based on the offset direction characterization vector, compare the positional relationship between the boundary extension path and the edge line of the measured component surface, determine the segment position of the measurement marker points distributed along the boundary direction, identify the corresponding segment identifier, and obtain the boundary segment association identifier; S513: Based on the boundary segment association identifier, analyze and measure the correspondence between the arrangement trajectory of the marker points, compare the conflict states of different boundary attribution relationships, determine the priority order of the boundary corresponding to the marker points, adjust the boundary attribution arrangement order, and obtain the boundary positioning association result; The measurement point coordinate reference set includes the measurement point pixel coordinate distribution, the marker point spatial index, and the calibration ruler endpoint coordinate combination. The directional scale corresponding quantity includes the lateral scale conversion benchmark, the longitudinal scale conversion benchmark, and the direction decomposition identifier. The measurement point interval characterization set includes the measurement point spacing expression unit, the connection direction identifier, and the component spacing record item. The edge line deviation reference rate includes the normal offset ratio identifier, the lateral offset classification identifier, and the edge line reference matching item. The boundary positioning association result includes the boundary matching relationship identifier, the positioning segment division identifier, and the boundary belonging correspondence identifier.

2. The engineering measurement data analysis method based on image recognition according to claim 1, characterized in that, The intersection point refers to the pixel position where the coordinates of two or more structural edge pixel trajectories coincide, and the endpoint arrangement direction index is the direction vector of the line connecting the pixel coordinates at both ends in the image coordinate system.

3. An engineering measurement data analysis system based on image recognition, characterized in that, The system is used to implement the engineering measurement data analysis method based on image recognition as described in any one of claims 1-2, and the system comprises: The image acquisition module is based on an industrial camera. It acquires the grayscale changes of the captured image, determines the structural edges corresponding to continuous abrupt change areas, compares the relationship between the intersection position and the marker outline, filters the coordinates of the measurement marker points, adjusts the positioning order of the calibration ruler end, and obtains the reference set of measurement point coordinates. Based on the coordinate reference set of the measuring points, the scale mapping module obtains the arrangement direction of the endpoints of the calibration ruler, compares the relationship between the lateral and longitudinal displacements, calculates the orientation of the ruler segment, adjusts the corresponding order of the actual length of the calibration ruler and the pixel span, determines the conversion relationship of each direction, and obtains the corresponding quantity of the scale in each direction. The spacing characterization module obtains the positional relationship of the points to be measured based on the corresponding quantities of the directional scale, calculates the lateral and longitudinal displacements of the two points to be measured, compares the direction mapping scale segments, adjusts the displacement conversion order, judges the relationship between the structure edge direction and the line connecting the measuring points, and obtains the measuring point spacing characterization set. The offset determination module analyzes the distribution of measurement point intervals based on the measurement point interval characterization set, determines the relationship between the measurement marker point and the normal interval of the structural edge, calculates the deviation state corresponding to the edge line of the same measured component, adjusts the edge line reference order, and obtains the edge line deviation reference rate. Based on the deviation rate of the edge line reference, the boundary positioning module analyzes the deviation direction, compares the extension direction of the reference boundary with the positional relationship of the edge line on the surface of the measured component, determines the corresponding segment of the trajectory of the measurement marker points, adjusts the boundary assignment order, and obtains the boundary positioning association result.

Images