A method and system for evaluating the quality of prefabricated construction based on image processing

Abstract

This invention relates to the field of image processing technology, and more particularly to a method and system for assessing the quality of prefabricated construction based on image processing. The method includes the following steps: after removing the EVA transport protective padding that came with the prefabricated construction components, a panoramic image of the component splicing surface is captured using an industrial camera to obtain an original image containing the natural adhesion marks of the padding; multi-scale edge enhancement processing is performed on the original image to extract the contour data of the adhesion marks, and the global flatness of the component splicing surface is calculated based on the deformation features of the contour data; an industrial camera is used in a ring array to capture images of the pre-reserved grouting holes in the components, and the grouting hole boundaries are extracted after contour sharpening processing of the images; the three-dimensional axis of the grouting holes is constructed by combining the reflective features of the inner sidewalls of the holes. This invention achieves accurate detection and standardized assessment of the quality of prefabricated components through multi-polarized light sources and ring camera imaging technology, ultimately improving the accuracy of construction quality assessment.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to a method and system for evaluating the quality of prefabricated construction based on image processing. Background Technology

[0002] Prefabricated construction achieves efficient construction through factory-prefabricated components and on-site assembly. It features standardized component production, short on-site operation cycles, and low resource consumption. Quality assessment requires that the flatness of component splicing surfaces and the accuracy of the axis of reserved grouting holes meet standards to ensure structural connection stability and overall safety. However, existing construction quality assessment technologies are difficult to adapt to this requirement and have significant shortcomings: traditional flatness detection relies on manual straightedges or single image acquisition, without optimizing the processing of EVA pad bonding marks, resulting in blurred edge feature extraction and easy neglect of abrupt deformation areas, leading to assessment bias; grouting hole detection often uses single-angle shooting and two-dimensional imaging, which can easily cause boundary breakage due to hole wall reflection and local damage; the axis fitting lacks a graded weighting mechanism, and the deviation parameters do not match the actual working conditions. Summary of the Invention

[0003] Therefore, it is necessary to provide a method and system for evaluating the quality of prefabricated construction based on image processing to solve at least one of the above-mentioned technical problems.

[0004] To achieve the above objectives, a prefabricated construction quality assessment method based on image processing is provided. The prefabricated construction components include EVA transport protective padding. The method includes the following steps:

[0005] Step S1: After removing the EVA transport protective padding that came with the prefabricated components at the factory, use an industrial camera to take a panoramic photo of the splicing surface of the components to obtain the original image containing the natural adhesion marks of the padding.

[0006] Step S2: Perform multi-scale edge enhancement processing on the original image, extract the contour data of the bonding traces, and calculate the global flatness of the component splicing surface based on the deformation features of the contour data;

[0007] Step S3: Use an industrial camera in a circular array to capture images of the reserved grouting holes in the component. After contour sharpening of the images, extract the boundaries of the grouting holes and construct the three-dimensional axis of the grouting holes by combining the reflection characteristics of the inner sidewall of the holes. Determine the overall axis deviation parameters of the component based on the three-dimensional axis of all reserved grouting holes.

[0008] Step S4: Compare the global flatness and axis deviation parameters with the preset qualified construction quality thresholds respectively. If both the global flatness and axis deviation parameters meet the standards, the assembly quality is deemed qualified. If either exceeds the standard, the assembly quality is deemed unqualified.

[0009] The present invention also provides an image processing-based prefabricated construction quality assessment system for performing the above-described image processing-based prefabricated construction quality assessment method. The image processing-based prefabricated construction quality assessment system includes:

[0010] The padding trace image acquisition module is used to take panoramic photos of the splicing surface of prefabricated construction components after removing the EVA transport protective padding that came with the components at the factory, and to obtain the original image containing the natural adhesion traces of the padding.

[0011] The flatness analysis module is used to perform multi-scale edge enhancement processing on the original image, extract the contour data of the bonding traces, and calculate the global flatness of the component splicing surface based on the deformation features of the contour data.

[0012] The grouting hole axis detection module is used to capture images of the reserved grouting holes of the component using an industrial camera in a circular array. After contour sharpening processing of the image, the boundary of the grouting hole is extracted, and the three-dimensional axis of the grouting hole is constructed by combining the reflection characteristics of the inner sidewall of the hole. The axis deviation parameters of the entire component are determined based on the three-dimensional axis of all reserved grouting holes.

[0013] The quality assessment module compares the global flatness and axis deviation parameters with preset qualified construction quality thresholds. If both the global flatness and axis deviation parameters meet the standards, the assembly quality is deemed qualified; if either exceeds the standard, the assembly quality is deemed unqualified.

[0014] The beneficial effects of the present invention are as follows:

[0015] I. This invention achieves high-precision detection through multi-dimensional technical design. It employs multi-polarization direction light sources to enhance the edge features of bonding marks and proportionally fuses images to accurately extract contour data. It also adjusts global flatness by combining deformation ratio ranges, simultaneously detecting abrupt deformation areas and adding correction amounts to avoid misjudgments of flatness. For grouting hole detection, it uses a ring array camera to capture images from multiple angles and stitches boundary data at different depths. Combined with multi-angle light sources to capture reflective bands and construct a three-dimensional axis, it then calculates axis deviation parameters using hierarchical weights. This multi-stage technical collaboration significantly reduces evaluation errors.

[0016] Second, this invention simplifies the traditional complex testing process, achieving efficient testing through standardized equipment operation and parameter settings. For example, it uses an industrial camera to capture panoramic images, collects data using a fixed-angle light source and camera rotation parameters, and defines weights and thresholds to complete the evaluation and judgment, eliminating the need for repeated manual adjustments or complex preprocessing. From image acquisition to parameter calculation and quality judgment, a standardized process is formed, reducing manual intervention. Compared with traditional manual measurement or single-equipment testing, it significantly shortens the evaluation cycle and adapts to the needs of efficient prefabricated construction.

[0017] Third, this invention possesses precise defect identification and traceability capabilities, enabling it to locate localized abrupt deformation zones and quantify relevant parameters. For borehole wall damage, virtual boundary connections ensure continuous evaluation and prevent detection interruptions. In the quality assessment stage, correlation coefficient analysis of parameter relationships dynamically adjusts thresholds for secondary assessment. When standards are not met, detailed records of exceeding parameters and related data are kept, and the defect type is marked. This not only clarifies the location of the defect but also allows for tracing related factors. Attached Figure Description

[0018] Figure 1 This is a schematic diagram illustrating the steps of a prefabricated construction quality assessment method based on image processing.

[0019] Figure 2 Drawings of prefabricated construction components;

[0020] Figure 3 This is a schematic diagram of multi-scale edge enhancement in an image;

[0021] Figure 4 To construct a three-dimensional axis diagram of the grouting hole;

[0022] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0023] The technical method of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0024] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0025] It should be understood that although the terms "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0026] To achieve the above objectives, please refer to Figures 1 to 4 A method for quality assessment of prefabricated construction based on image processing, wherein the prefabricated construction components include EVA transport protective pads, the method comprising the following steps:

[0027] Preferably, in step S1: after removing the EVA transport protective padding that came with the prefabricated construction components from the factory, use an industrial camera to take a panoramic photo of the splicing surface of the components to obtain the original image containing the natural adhesion marks of the padding.

[0028] In one embodiment, when removing the EVA transport protective pad that comes with the prefabricated construction component, a plastic peeling tool is used. The edge thickness of the tool is controlled at 1.5-2.5mm, and the front end radius of curvature is 4-6mm. The tool peels off at a uniform speed along the edge of the pad at an angle of 12°-18° to the splicing surface of the component. The peeling thrust is limited to 0.2-0.3MPa by a pressure control device.

[0029] In another embodiment, the component after the padding is removed is fixed on a workbench with a leveling function, and calibrated using a level measuring instrument with an accuracy of not less than 0.02 mm / m. The workbench support structure is adjusted so that the horizontal deviation of the component splicing surface is controlled within ±0.05 mm / m.

[0030] It should be noted that an industrial camera with a resolution of no less than 30 million pixels is used to take panoramic photos of the splicing surface of the components. The camera is equipped with a fixed-focus lens with a focal length of 20-30mm and is mounted on a height-adjustable bracket. The center of the lens is vertically aligned with the geometric center of the splicing surface through scale positioning. The vertical distance from the lens to the splicing surface is set to 0.8-1.2m.

[0031] Specifically, a ring-shaped LED fill light system was used during the shooting process. This system contains no fewer than 96 LED beads. The light intensity was adjusted to 500-1100 lux and the color temperature was controlled at 4800-5200K through a dimming device. The angle between the light group and the splicing surface was fixed at 25°-35°.

[0032] In another embodiment, the shooting zones are divided according to the actual size of the component splicing surface, with the longitudinal and transverse intervals not exceeding 0.6m. When shooting each zone, the camera exposure time is set to 1 / 150-1 / 180s and the ISO value is set to 100-200. The overlap rate of adjacent zone images is ensured to be no less than 10% through camera linkage control.

[0033] It should be noted that all partitioned images are combined into a complete original image using an image stitching processing device. The output image format is TIFF, with a resolution of no less than 7000×5000 pixels. The grayscale contrast of the natural bonding marks of the padding is verified to be ≥25 gray levels using an image quality inspection tool.

[0034] Preferably, step S2: perform multi-scale edge enhancement processing on the original image, extract the contour data of the bonding traces, and calculate the global flatness of the component splicing surface based on the deformation features of the contour data;

[0035] Optionally, the multi-scale edge enhancement processing of the original image in step S2 specifically involves:

[0036] In the first stage, a 0° polarized light source is used to illuminate the splicing surface of the components, and the first set of edge images is obtained by synchronously taking pictures. This light source only enhances the lateral edge features of the bonding marks.

[0037] In the second stage, a 90° polarized light source is used to illuminate the same area, and a second set of edge images is obtained by synchronously taking pictures. This light source only enhances the longitudinal edge features of the bonding traces.

[0038] The third stage uses a 45° polarized light source to illuminate the third set of edge images, which enhances the features of obliquely intersecting edges.

[0039] The three sets of images are superimposed and fused according to the ratio of 40% for horizontal edges, 40% for vertical edges, and 20% for diagonal edges to form a complete edge-enhanced image.

[0040] Please see Figure 3 In this embodiment, when performing multi-scale edge enhancement processing on the original image, the component splicing surface is first fixed on an optical platform with a polarization adjustment device. The horizontal accuracy of the platform is controlled within ±0.01mm / m to ensure that the parallelism error between the component splicing surface and the light source emission surface is ≤0.02mm.

[0041] It should be noted that in the first stage, a 50W linearly polarized light source is used, the polarizer angle is adjusted to 0° so that the polarization direction is consistent with the length direction of the component, the light source intensity is set to 650 lux, the illumination angle is 45° with the splicing surface, and a 12-megapixel industrial camera that is triggered synchronously with the light source is started, the exposure time is set to 1 / 120s and the ISO value is 100, and the first set of edge images is captured. In this image, the grayscale gradient of the horizontal edge of the bonding mark is ≥40.

[0042] It should be noted that in the second stage, the camera position and parameters are kept unchanged, the polarizer angle is adjusted to 90° so that the polarization direction is consistent with the width direction of the component, the light source intensity is maintained at 650 lux, the illumination angle is kept unchanged, and the second set of edge images is captured. The gray level gradient of the longitudinal edge of the bonding trace in the image is ≥40.

[0043] It should be noted that in the third stage, the polarizer angle is adjusted to 45°, the polarization direction is at a 45° angle with the length direction of the component, the light source intensity is adjusted to 500 lux, the illumination angle remains unchanged, and the third set of edge images is captured. In this image, the grayscale gradient of the oblique cross edge of the bonding trace is ≥30.

[0044] In another embodiment, three sets of images are imported into an image fusion processing system. The system first extracts edge features from each set of images. During extraction, the horizontal edge detection threshold is set to 35, the vertical edge detection threshold is set to 35, and the diagonal edge detection threshold is set to 25. Then, pixel-level overlay fusion is performed according to the weights of the edge features of the first set of images (40%), the second set of images (40%), and the third set of images (20%).

[0045] It should be noted that the Gaussian pyramid fusion algorithm is used in the fusion process, with 6 pyramid levels. The fusion coefficient of each level is consistent with the proportion of the corresponding edge features. The final output is a complete edge-enhanced image with the same resolution as the original image and a signal-to-noise ratio of ≥30dB for the edge features.

[0046] Optionally, the calculation of global flatness based on the deformation features of the contour data in step S2 is specifically as follows:

[0047] The bonding trace contour data is divided into continuous trace segments along the length direction, with each segment containing the same number of contour points.

[0048] Calculate the straightness deviation of each contour segment; segments with straightness deviation exceeding the basic threshold are marked as deformed segments.

[0049] The proportion of the total length of the deformed segment to the total length of the contour is denoted as the deformation percentage.

[0050] When the deformation percentage is in the first interval, the average plane deviation of all contour points is taken as the global flatness; when the deformation percentage is in the second interval, the average plane deviation is adjusted in combination with the deformation percentage to obtain the final global flatness.

[0051] In one embodiment, when calculating the global flatness based on the deformation features of the contour data, the contour data of the bonding trace is extracted from the edge enhancement image. The contour data contains 1200 contour points continuously distributed along the length direction of the trace, and each contour point has three-dimensional coordinate information (X-axis is the length direction, Y-axis is the width direction, and Z-axis is the height direction).

[0052] In another embodiment, the contour data is divided into 20 continuous trace segments along the length direction, each segment containing 60 contour points. The X-axis coordinates of the starting and ending points of the segments are equally spaced. For each contour segment, a reference straight line is fitted using the least squares method. The deviation values ​​of the Z-axis coordinates of all contour points in the segment from the corresponding Z-axis coordinates of the reference straight line are calculated. The maximum value among the deviation values ​​is taken as the straightness deviation of the contour segment. The basic threshold is set to 0.3 mm. Segments with a straightness deviation exceeding 0.3 mm are marked as deformed segments. The total length of all deformed segments is calculated, and the ratio of this to the total length of the contour (the sum of the lengths of all segments) is recorded as the deformation ratio. The calculation accuracy of the deformation ratio is retained to two decimal places.

[0053] It should be noted that the first interval is set as deformation percentage ≤ 15%, and the second interval is set as 15% < deformation percentage ≤ 40%.

[0054] Specifically, when the deformation percentage is in the first interval, the deviation values ​​of the Z-axis coordinates of all 1200 contour points from the preset reference plane are calculated, and the arithmetic mean of each deviation value is taken as the global flatness. When the deformation percentage is in the second interval, the average plane deviation is multiplied by the adjustment coefficient of (1 + deformation percentage / 2), and the result is taken as the final global flatness. The calculation result of the adjustment coefficient is retained to three decimal places.

[0055] Optionally, after calculating the global flatness of the component splicing surface in step S2, the following steps are also included:

[0056] Within the deformation segment, feature points are selected at intervals according to the matching accuracy, and the height change rate between adjacent feature points is calculated.

[0057] When the height change rate exceeds the gradient threshold for multiple consecutive segments, the region is determined to be a region of abrupt deformation.

[0058] Measure the maximum height difference and area of ​​the abrupt deformation zone. If both the maximum height difference and area meet the set conditions, add the abrupt correction amount to the global flatness result.

[0059] The original calculated global smoothness is superimposed with the mutation correction amount to obtain the corrected global smoothness.

[0060] In this embodiment, after calculating the global flatness of the component splicing surface, feature points are selected within the marked deformation segment at intervals of the matching accuracy. The matching accuracy interval is set to 2mm, that is, a feature point is selected every 2mm. The three-dimensional coordinates (X-axis, Y-axis, Z-axis) of each feature point are recorded. The height change rate between two adjacent feature points is calculated. The height change rate is obtained by the ratio of (the Z-axis coordinate of the later feature point minus the Z-axis coordinate of the earlier feature point) to the distance between the two points along the X-axis direction. The result is retained to three decimal places. The gradient threshold is set to 0.05mm / mm. When the height change rate exceeds 0.05mm / mm for three or more consecutive intervals, the continuous area is determined to be a sudden deformation zone.

[0061] It should be noted that the maximum height difference and area of ​​the abrupt deformation zone are measured using coordinate measurement tools. The maximum height difference is the difference between the maximum and minimum Z-axis coordinates of all feature points in the region. The area is calculated using the X-axis and Y-axis coordinates of the feature points at the region boundary. The conditions are set as follows: maximum height difference ≥ 0.5 mm and area ≥ 100 mm². If both conditions are met, an abrupt correction amount is added to the global flatness result. The abrupt correction amount is calculated as (maximum height difference × 0.3 + area × 0.001), and the result is retained to three decimal places. The original calculated global flatness and the abrupt correction amount are numerically superimposed, and the sum is used as the corrected global flatness.

[0062] Of particular importance is that step S2, which involves extracting the contour data of the bonding marks, includes:

[0063] The fluorescence excitation source was determined, and the irradiation angle of the fluorescence excitation source was adjusted so that the residual components of EVA could be selectively excited, while avoiding reflection interference from the component body.

[0064] The image after multi-scale edge enhancement is subjected to fluorescence excitation processing, and the EVA material residue area is continuously irradiated to emit characteristic fluorescence of stable intensity.

[0065] A narrowband filter matching the characteristic fluorescence wavelength is installed in front of the imaging device lens to filter out ambient light and stray light interference, and only capture the fluorescence image of the EVA residue area. The area with stable fluorescence signal in the image is marked as the core contour area of ​​the bonding trace.

[0066] The outer boundary of the core contour region is identified by line-by-line scanning: scanning from the image edge to the center, when pixels that meet the fluorescence intensity threshold are detected continuously, the starting point is recorded as the outer boundary point, and all outer boundary points are connected to form the initial outer contour line;

[0067] The detection process gradually shrinks from the outer contour line towards the core area. When the fluorescence intensity drops to a set threshold, the shrinkage stops, and the boundary point at this point is recorded to form the inner boundary line.

[0068] Calculate the midpoint of the distance between the inner and outer boundary lines at corresponding positions, and connect all the midpoints in sequence to form the final fitting trace contour data.

[0069] In one embodiment, an ultraviolet fluorescence excitation light source with a wavelength of 365nm is used, the angle between the light source and the splicing surface of the component is adjusted to 30°, the power of the light source is set to 80W, and the irradiation range covers the entire splicing surface. By adjusting the angle, it is ensured that the residual EVA components are selectively excited, while the reflectivity of the component body is controlled to be below 15% of the fluorescence signal intensity.

[0070] In another embodiment, the image after multi-scale edge enhancement is subjected to fluorescence excitation processing and continuously irradiated for 60 seconds until the EVA material residue area emits a stable intensity of characteristic fluorescence with a peak wavelength of 450nm. A narrow-band filter with a center wavelength of 450nm and a bandwidth of 20nm is installed in front of the lens of a 16-megapixel imaging device to filter out ambient light and stray light interference, and only capture the fluorescence image of the EVA residue area. The image resolution is 4928×3264 pixels, and the area with a fluorescence intensity ≥5000 count in the image is marked as the core contour area of ​​the bonding trace.

[0071] It should be noted that a line-by-line scanning method is used to identify the outer boundary of the core contour region, with a scanning step size of 1 pixel. The scan proceeds from the image edge towards the center. When five or more pixels with fluorescence intensities ≥4500 are detected consecutively, the starting point of the sequence is recorded as the outer boundary point. All outer boundary points are connected to form the initial outer contour line. The detection gradually shrinks from the outer contour line towards the core region with a step size of 0.5 pixels. When the fluorescence intensity drops to 2000, the shrinkage stops, and the boundary point at this point is recorded to form the inner boundary line. The midpoint of the distance between the inner boundary line and the outer boundary line at the same X-axis coordinate position is calculated. All midpoints are connected sequentially according to the X-axis coordinate order to form the final bonding trace contour data containing 2000 consecutive coordinate points. Each coordinate point contains X-axis and Y-axis pixel position information and the corresponding fluorescence intensity value.

[0072] Of particular importance is that, before the multi-scale edge enhancement processing in step S2, a trace preprocessing step is added, specifically as follows:

[0073] A special colorimetric reagent composed of a specific solvent and an EVA-affinity component was prepared, which reacts specifically with EVA residues only.

[0074] Atomized spraying equipment is used to evenly cover the splicing surfaces of the components with color developer. The spraying path is distributed in a grid pattern to cover all areas with bonding marks.

[0075] The component is placed in a constant temperature and humidity environment and left to stand until the color developer and EVA residues react completely to form a reaction layer with a significant color difference from the surrounding area.

[0076] Use a low-pressure airflow to remove unreacted colorimetric reagent residue;

[0077] The color difference contrast between the reacting and non-reacting areas is detected. When the contrast reaches the preset standard, subsequent multi-scale edge enhancement processing is performed to amplify the distinction between the bonding marks and the background through directional color development reaction.

[0078] In this embodiment, in the trace pretreatment step before multi-scale edge enhancement treatment, a special color developer is prepared, consisting of 30% by mass of ethyl acetate solvent and 70% by mass of ethylene-vinyl acetate copolymer affinity component. This color developer only undergoes a specific chemical reaction with EVA residue. A WS-500 atomizing spraying device is used to evenly cover the splicing surface of the components with the color developer. The device working pressure is set to 0.2 MPa, the droplet diameter is controlled at 50-80 μm, the spraying path is distributed in a 50 mm × 50 mm grid, the horizontal spraying speed is 200 mm / s, and the vertical spacing is 30 mm to cover all bonding marks. The area with the traces was identified; the component was placed in a constant temperature and humidity chamber with the temperature set at 25℃ and the relative humidity controlled at 60% for 15 minutes until the color developer and EVA residues completely reacted, forming a reaction layer with a significant color difference from the surrounding area; clean compressed air at a pressure of 0.1MPa was used to remove unreacted color developer residues at a 45° angle to the splicing surface of the component, with the airflow action time being 30 seconds; the color difference contrast between the reaction area and the non-reaction area was detected using a CR-400 colorimeter, with one detection point set per 100mm², for a total of no less than 50 detection points. The preset standard was a ΔE value ≥30 in the CIE LAB color space. When the ΔE value of all detection points reached 30 or above, subsequent multi-scale edge enhancement processing was performed to amplify the distinction between the bonding traces and the background through directional color development reaction.

[0079] Optionally, in step S3, the specific steps of using an industrial camera in a circular array to capture images of the pre-reserved grouting holes on the component are as follows:

[0080] The industrial cameras in the ring array rotate synchronously around the axis of the grouting hole, with a rotation angle range of 0° to 90°;

[0081] A synchronous shot is taken every 15° of rotation to obtain 6 sets of images from different angles;

[0082] Among them, the images of the 0° and 45° angle groups are used to extract the upper boundary of the hole, the 15° and 60° angle groups are used to extract the middle boundary of the hole, and the 30° and 90° angle groups are used to extract the lower boundary of the hole.

[0083] The boundary data extracted from each group of images are stitched together along the depth direction to form a complete imaging image.

[0084] In one embodiment, when using an industrial camera ring array to capture images of the pre-reserved grouting holes in a component, an array of eight 20-megapixel industrial cameras is used. The cameras are evenly distributed on a circumference with a diameter of 500mm. The focal length of the lenses is 25mm, and the vertical distance between the center of the lens and the axis of the grouting hole is 250mm. The optical axes of all cameras converge at the same point on the axis of the grouting hole, which is 300mm away from the plane of the hole opening. The ring array is driven by a servo rotating platform to rotate synchronously around the axis of the grouting hole. The rotation angle range is set from 0° to 90°, and the rotation accuracy is controlled within ±0.1°. Synchronous shooting is triggered every 15° of rotation, resulting in six sets of images from different angles. Each set of images contains eight partial images captured simultaneously by the eight cameras.

[0085] In another embodiment, the camera's built-in LED coaxial light source is used during shooting, with the light source brightness adjusted to 800 lux, the exposure time set to 1 / 100s, and the ISO value set to 200. The images from the 0° and 45° angle groups are used to extract the upper boundary of the hole (from the hole opening to 1 / 3 of the hole depth), the 15° and 60° angle groups are used to extract the middle boundary of the hole (from 1 / 3 to 2 / 3 of the hole depth), and the 30° and 90° angle groups are used to extract the lower boundary of the hole (from 2 / 3 of the hole depth to the bottom of the hole). The boundary data extracted from each group of images are stitched together along the depth direction, with the hole axis as the reference. The overlap length of boundary data in adjacent depth segments is not less than 5mm, ultimately forming a complete imaging image with a resolution of 6000×4000 pixels and a hole wall detail sharpness of 0.05mm / pixel.

[0086] Preferably, step S3: use an industrial camera in a circular array to capture images of the reserved grouting holes in the component; after contour sharpening of the images, extract the boundaries of the grouting holes; and construct the three-dimensional axis of the grouting holes by combining the reflective features of the inner sidewall of the holes; determine the overall axis deviation parameters of the component based on the three-dimensional axes of all reserved grouting holes.

[0087] Optionally, step S3, which involves extracting the grouting hole boundary after contour sharpening of the imaging image, includes:

[0088] Identify the prefabricated annular groove on the end face of the grouting hole, which is a positioning structure preset when the component leaves the factory;

[0089] Locate the inner and outer edges of the groove in the imaging image, calculate the centerline between the inner and outer edges, and use this centerline as the reference boundary of the aperture.

[0090] The extraction boundary is extended from the reference boundary at the orifice into the orifice. The extraction accuracy level is switched every time the depth is increased to the appropriate depth. The extraction accuracy of the subsequent level is higher than that of the previous level.

[0091] When the inner wall of the hole is damaged, a virtual boundary of the damaged area is generated by extending along the axis of the hole body, with the hole opening reference boundary as the symmetry reference, and the virtual boundary is connected with the boundary of the complete area.

[0092] In one embodiment, when extracting the boundary of the grouting hole after contour sharpening of the imaging image, the image grayscale threshold segmentation technology is first used to identify the prefabricated annular groove on the end face of the grouting hole. The groove is a positioning structure preset when the component leaves the factory. The groove width is set to 5mm and the depth is set to 2mm. In the imaging image, the groove range is located by the difference in grayscale values ​​(the grayscale value of the groove area is 40-60 grayscale levels lower than that of other areas on the end face of the hole).

[0093] In another embodiment, the inner and outer edges of the groove are located in the imaging image. The inner edge is the boundary of the groove on the side closer to the center of the hole, and the outer edge is the boundary of the groove on the side farther from the center of the hole. The distance between the inner and outer edges in the same radial direction is calculated using pixel coordinates. The pixel point corresponding to 1 / 2 of the distance is taken to form a centerline. This centerline is used as the reference boundary of the hole opening. The pixel coordinate error of the reference boundary of the hole opening is controlled within ±1 pixel. The extraction boundary is extended from the reference boundary of the hole opening into the hole along the depth direction. The adaptation depth is set to 20mm, that is, the extraction accuracy level is switched every 20mm. The extraction accuracy of the initial level (from the hole opening to a depth of 20mm) is set to 0.1mm / pixel, the extraction accuracy of the second level (from 20mm to a depth of 40mm) is set to 0.08mm / pixel, and the extraction accuracy of the third level (from 40mm to the bottom of the hole) is set to 0.05mm / pixel. The extraction accuracy of subsequent levels is higher than that of the previous level.

[0094] Additionally, it should be noted that when damage is found on the inner wall of the hole through image comparison, a virtual boundary of the damaged area is generated along the hole body axis, using the hole opening reference boundary as a symmetry reference, according to the depth range of the damaged area (from the starting depth of the damage to the ending depth of the damage). The curvature of the virtual boundary is consistent with the curvature of the complete area boundary at the starting point of the damage. The virtual boundary and the complete area boundary are smoothly connected at the starting point and the ending point of the damage, and the pixel coordinate deviation at the connection point does not exceed ±2 pixels.

[0095] Optionally, in step S3, constructing the three-dimensional axis of the grouting hole by combining the reflective features of the inner sidewall of the hole is specifically as follows:

[0096] The first step uses a light source at a 30° angle to the hole axis to illuminate the inner wall of the hole and capture the reflective band in the hole opening area;

[0097] The second step uses a light source at a 60° angle to the hole axis to capture the reflective band in the middle area of ​​the hole;

[0098] The third step uses a light source at an angle of 80° to the hole axis to capture the reflective band in the bottom area of ​​the hole;

[0099] Connect the center points of the three stepped reflective strips in sequence to form the initial axis through the hole;

[0100] The deviation of the initial axis from each step of the reflective strip is calculated, and the final three-dimensional axis is obtained by correcting it using the least squares method.

[0101] Please see Figure 4 In this embodiment, when constructing the three-dimensional axis of the grouting hole by combining the reflective features of the inner sidewall of the hole, three sets of adjustable-angle laser light sources (wavelength 650nm, output power 10mW) are used to carry out three-step irradiation operations respectively.

[0102] In one embodiment, the first step adjusts the angle between the laser light source and the hole axis to 30°, and focuses the light source illumination point on the hole opening area (0-15mm depth range from the hole opening plane). A 12-megapixel industrial camera (lens focal length 35mm, exposure time 1 / 120s, ISO100) captures the reflective band in the hole opening area. The width of the reflective band needs to be controlled within the range of 0.5-1mm. The pixel coordinates of the reflective band in the image are recorded and its geometric center point is calculated.

[0103] In another embodiment, the second step keeps the camera parameters unchanged, adjusts the angle between the laser source and the hole axis to 60°, focuses the illumination point on the middle area of ​​the hole (within a depth range of 15-30mm from the hole opening plane), and captures the reflective band in this area, and calculates the geometric center point of the reflective band.

[0104] In another embodiment, the third step adjusts the angle between the laser source and the hole axis to 80°, focuses the irradiation point on the bottom area of ​​the hole (within a depth range of 30-45mm from the hole opening plane), and calculates the geometric center point after capturing the reflective band;

[0105] It should be noted that the geometric center points of the reflective strips obtained from the three steps are connected sequentially along the depth direction to form the initial axis through the hole. The coordinate data of the initial axis is based on the center of the hole opening plane as the origin, and the Z-axis is established along the depth direction of the hole.

[0106] Specifically, the deviation of the initial axis from each reflective band is calculated. The deviation is the average vertical distance of all pixels on each reflective band to the initial axis. The calculation accuracy of the first step deviation needs to be ±0.01mm, and the second and third steps need to be ±0.008mm. The initial axis is corrected by the least squares method, with each step deviation as a weight (20% for the first step, 50% for the second step, and 30% for the third step). The axis coordinates are iteratively optimized until the average deviation of the corrected axis from each reflective band is ≤0.005mm, thus obtaining the final three-dimensional axis. The final three-dimensional axis needs to output the X, Y, and Z coordinate values ​​of each depth node (5mm interval).

[0107] Optionally, the determination of the overall axial deviation parameters of the component based on the three-dimensional axes of all reserved grouting holes in step S3 includes:

[0108] All reserved grouting holes are divided into primary holes, secondary holes and final holes according to their distance from the end of the component; among them, primary holes are located in the front third of the component length direction, secondary holes are located in the middle third of the component length direction, and final holes are located in the rear third of the component length direction.

[0109] The axial deviation weight for the first-stage hole is set to 30%, for the second-stage hole to 50%, and for the final-stage hole to 20%.

[0110] Calculate the average axial deviation value of each hole, multiply the average deviation value of each hole by the corresponding weight, and add the product results to obtain the weighted sum;

[0111] The weighted sum is compared with the maximum deviation value in each level of hole, and the larger of the two values ​​is taken as the final axis deviation parameter.

[0112] In this embodiment, when determining the overall axial deviation parameters of the component based on the three-dimensional axes of all reserved grouting holes, the total length of the component in the length direction is first measured. The component length is set to 6000mm. Taking the front end face of the component in the length direction as the starting point, the range of 0-2000mm is divided into the front third region, the range of 2000-4000mm is divided into the middle third region, and the range of 4000-6000mm is divided into the rear third region. The reserved grouting holes located in the front third region are marked as primary holes (3 in total), those located in the middle third region are marked as secondary holes (4 in total), and those located in the rear third region are marked as final holes (3 in total).

[0113] It should be noted that the weighting for the axial deviation of the first-stage hole is set at 30%, the weighting for the axial deviation of the second-stage hole is 50%, and the weighting for the axial deviation of the last-stage hole is 20%.

[0114] For example, the deviation values ​​of the three-dimensional axis of each grouting hole in each stage of the hole and the component design reference axis were calculated using coordinate measuring software. The axis deviation values ​​of the three grouting holes in the first stage were 0.12mm, 0.15mm, and 0.13mm, respectively. The average axis deviation value was calculated as follows: The axial deviations of the four grouting holes in the secondary holes were 0.08 mm, 0.10 mm, 0.09 mm, and 0.11 mm, respectively, with an average axial deviation of [value missing]. The axial deviations of the three grouting holes in the final stage were 0.14 mm, 0.16 mm, and 0.15 mm, respectively, with an average axial deviation of [value missing]. Multiply the average deviation value of each stage by its corresponding weight. The contribution value of the first-stage hole is 0.133mm × 30% = 0.0399mm, and the contribution value of the second-stage hole is... The contribution value of the final-stage hole is Adding the products of the three together gives the weighted sum. ;

[0115] The maximum deviation values ​​for each stage of the hole were statistically analyzed. The maximum deviation value for the first stage hole was 0.15 mm, the maximum deviation value for the second stage hole was 0.11 mm, and the maximum deviation value for the last stage hole was 0.16 mm. The maximum deviation value for each stage of the hole was 0.16 mm. The weighted sum of 0.1174 mm was compared with the maximum deviation value of 0.16 mm for each stage of the hole, and the larger value of 0.16 mm was taken as the final axis deviation parameter.

[0116] Preferably, step S4: compare the global flatness and axis deviation parameters with the preset qualified construction quality threshold respectively. If both the global flatness and axis deviation parameters meet the standard, the assembly quality is judged to be qualified; if either exceeds the standard, the assembly quality is judged to be unqualified.

[0117] Optionally, step S4 includes the following steps:

[0118] Step S41: Calculate the correlation coefficient between global flatness and axis deviation parameter. When the absolute value of the correlation coefficient is greater than 0.7, it is determined that there is a strong correlation between the two.

[0119] Step S42: If there is a strong correlation and one parameter exceeds the standard by less than 10%, then adjust the qualified threshold of the other parameter to 1.05 times the original threshold, and make a second judgment based on the adjusted threshold.

[0120] Step S43: If there is no strong correlation, the original qualified threshold shall be used for judgment directly;

[0121] Step S44: If the second judgment still fails to meet the standard, record the specific value of the parameter that exceeds the standard and the value of the associated parameter, and mark it as a parameter association defect.

[0122] In this embodiment, the Pearson correlation coefficient between the global flatness of the component splicing surface and the axial deviation parameter is calculated using data statistics tools. The calculated global flatness is set to 0.32 mm and the axial deviation parameter to 0.16 mm. Substituting these values ​​into the correlation coefficient calculation formula yields a correlation coefficient of 0.82. Since the absolute value of this value is greater than 0.7, a strong correlation is determined between the two. The original acceptable thresholds for global flatness are preset to 0.30 mm and for axial deviation parameter to 0.15 mm. At this point, the global flatness exceeds the acceptable range by [missing value]. If the deviation is less than 10%, the acceptable threshold for the axis deviation parameter will be adjusted to 1.05 times the original threshold. A second judgment is made based on the adjusted threshold. The axis deviation parameter of 0.16mm is compared with the adjusted threshold of 0.1575mm. If the axis deviation parameter still exceeds the standard, the correlation coefficient is calculated to be 0.55 (absolute value less than 0.7). If the two are not strongly correlated, the original qualified threshold (global flatness 0.30mm, axis deviation 0.15mm) is used for judgment. If the second judgment still fails to meet the standard, the specific values ​​of the exceeding parameter (axis deviation parameter 0.16mm) and the related parameter (global flatness 0.32mm) are recorded through the quality record system. The component is marked as a "parameter-related defect" in the system. The defect record must include the inspection time, component number and corresponding parameter calculation results.

[0123] Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of the equivalents of the application are intended to be included within the invention.

[0124] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A method for evaluating the quality of prefabricated construction based on image processing, characterized in that, The method for constructing prefabricated components, which includes EVA transport protective padding, comprises the following steps: Step S1: After removing the EVA transport protective padding that came with the prefabricated construction components from the factory, use an industrial camera to take a panoramic photo of the splicing surface of the components to obtain the original image containing the natural adhesion marks of the padding. Step S2: Perform multi-scale edge enhancement processing on the original image, extract the contour data of the bonding traces, and calculate the global flatness of the component splicing surface based on the deformation features of the contour data; Specifically, in step S2, calculating the global flatness based on the deformation features of the contour data is as follows: The bonding trace contour data is divided into continuous trace segments along the length direction, with each segment containing the same number of contour points. Calculate the straightness deviation of each contour segment; segments with straightness deviation exceeding the basic threshold are marked as deformed segments. The proportion of the total length of the deformed segment to the total length of the contour is denoted as the deformation percentage. When the deformation percentage is in the first interval, the average plane deviation of all contour points is taken as the global flatness; when the deformation percentage is in the second interval, the average plane deviation is adjusted in combination with the deformation percentage to obtain the final global flatness. Step S3: Use an industrial camera in a circular array to capture images of the reserved grouting holes in the component. After contour sharpening of the images, extract the boundaries of the grouting holes and construct the three-dimensional axis of the grouting holes by combining the reflection characteristics of the inner sidewall of the holes. Determine the overall axis deviation parameters of the component based on the three-dimensional axis of all reserved grouting holes. Specifically, in step S3, constructing the three-dimensional axis of the grouting hole by combining the reflective features of the inner sidewall of the hole is as follows: The first step uses a light source at a 30° angle to the hole axis to illuminate the inner wall of the hole and capture the reflective band in the hole opening area; The second step uses a light source at a 60° angle to the hole axis to capture the reflective band in the middle area of ​​the hole; The third step uses a light source at an angle of 80° to the hole axis to capture the reflective band in the bottom area of ​​the hole; Connect the center points of the three stepped reflective strips in sequence to form the initial axis through the hole; The deviation of the initial axis from each step of the reflective strip is calculated, and the final three-dimensional axis is obtained by correcting it using the least squares method. Step S4: Compare the global flatness and axis deviation parameters with the preset qualified construction quality thresholds respectively. If both the global flatness and axis deviation parameters meet the standards, the assembly quality is deemed qualified. If either exceeds the standard, the assembly quality is deemed unqualified.

2. The image processing-based prefabricated construction quality assessment method according to claim 1, characterized in that, Step S2 involves performing multi-scale edge enhancement processing on the original image, specifically as follows: In the first stage, a 0° polarized light source is used to illuminate the splicing surface of the components, and the first set of edge images is obtained by synchronously taking pictures. This light source only enhances the lateral edge features of the bonding marks. In the second stage, a 90° polarized light source is used to illuminate the same area, and a second set of edge images is obtained by synchronously taking pictures. This light source only enhances the longitudinal edge features of the bonding traces. The third stage uses a 45° polarized light source to illuminate the third set of edge images, which enhances the features of obliquely intersecting edges. The three sets of images are superimposed and fused according to the ratio of 40% for horizontal edges, 40% for vertical edges, and 20% for diagonal edges to form a complete edge-enhanced image.

3. The image processing-based prefabricated construction quality assessment method according to claim 1, characterized in that, Step S2, after calculating the global flatness of the component splicing surface, also includes: Within the deformation segment, feature points are selected at intervals according to the matching accuracy, and the height change rate between adjacent feature points is calculated. When the height change rate exceeds the gradient threshold for multiple consecutive segments, the deformed segment is determined to be an abrupt deformation zone. Measure the maximum height difference and area of ​​the abrupt deformation zone. If both the maximum height difference and area meet the set conditions, add the abrupt correction amount to the global flatness result. The original calculated global smoothness is superimposed with the mutation correction amount to obtain the corrected global smoothness.

4. The prefabricated construction quality assessment method based on image processing according to claim 1, characterized in that, Step S3, which involves using an industrial camera in a circular array to capture images of the pre-reserved grouting holes in the component, specifically involves: The industrial cameras in the ring array rotate synchronously around the axis of the grouting hole, with a rotation angle range of 0° to 90°; A synchronous shot is taken every 15° of rotation to obtain 6 sets of images from different angles; Among them, the images of the 0° and 45° angle groups are used to extract the upper boundary of the hole, the 15° and 60° angle groups are used to extract the middle boundary of the hole, and the 30° and 90° angle groups are used to extract the lower boundary of the hole. The boundary data extracted from each group of images are stitched together along the depth direction to form a complete imaging image.

5. The image processing-based prefabricated construction quality assessment method according to claim 1, characterized in that, Step S3, which involves extracting the grouting hole boundary after contour sharpening of the imaging image, includes: Identify the prefabricated annular groove on the end face of the grouting hole, which is a positioning structure preset when the component leaves the factory; Locate the inner and outer edges of the groove in the imaging image, calculate the centerline between the inner and outer edges, and use this centerline as the reference boundary of the aperture. The extraction boundary is extended from the reference boundary at the orifice into the orifice. The extraction accuracy level is switched every time the depth is increased to the appropriate depth. The extraction accuracy of the subsequent level is higher than that of the previous level. When the inner wall of the hole is damaged, a virtual boundary of the damaged area is generated by extending along the axis of the hole body, with the hole opening reference boundary as the symmetry reference, and the virtual boundary is connected with the boundary of the complete area.

6. The prefabricated construction quality assessment method based on image processing according to claim 1, characterized in that, Step S3, which determines the overall axial deviation parameters of the component based on the three-dimensional axes of all reserved grouting holes, includes: All reserved grouting holes are divided into primary holes, secondary holes and final holes according to their distance from the end of the component; among them, primary holes are located in the front third of the component length direction, secondary holes are located in the middle third of the component length direction, and final holes are located in the rear third of the component length direction. The axial deviation weight for the first-stage hole is set to 30%, for the second-stage hole to 50%, and for the final-stage hole to 20%. Calculate the average axial deviation value of each hole, multiply the average deviation value of each hole by the corresponding weight, and add the product results to obtain the weighted sum; The weighted sum is compared with the maximum deviation value in each level of hole, and the larger of the two values ​​is taken as the final axis deviation parameter.

7. The image processing-based prefabricated construction quality assessment method according to claim 1, characterized in that, Step S4 includes the following steps: Step S41: Calculate the correlation coefficient between global flatness and axis deviation parameter. When the absolute value of the correlation coefficient is greater than 0.7, it is determined that there is a strong correlation between the two. Step S42: If there is a strong correlation and one parameter exceeds the standard by less than 10%, then adjust the qualified threshold of the other parameter to 1.05 times the original threshold, and make a second judgment based on the adjusted threshold. Step S43: If there is no strong correlation, the original qualified threshold shall be used for judgment directly; Step S44: If the second judgment still fails to meet the standard, record the specific value of the parameter that exceeds the standard and the value of the associated parameter, and mark it as a parameter association defect.

8. A prefabricated construction quality assessment system based on image processing, characterized in that, For performing the image processing-based prefabricated construction quality assessment method as described in claim 1, the image processing-based prefabricated construction quality assessment system includes: The padding trace image acquisition module is used to take panoramic photos of the splicing surface of prefabricated construction components after removing the EVA transport protective padding that came with the components at the factory, and to obtain the original image containing the natural adhesion traces of the padding. The flatness analysis module is used to perform multi-scale edge enhancement processing on the original image, extract the contour data of the bonding traces, and calculate the global flatness of the component splicing surface based on the deformation features of the contour data. Specifically, the calculation of global smoothness based on the deformation features of contour data is as follows: The bonding trace contour data is divided into continuous trace segments along the length direction, with each segment containing the same number of contour points. Calculate the straightness deviation of each contour segment; segments with straightness deviation exceeding the basic threshold are marked as deformed segments. The proportion of the total length of the deformed segment to the total length of the contour is denoted as the deformation percentage. When the deformation percentage is in the first interval, the average plane deviation of all contour points is taken as the global flatness; when the deformation percentage is in the second interval, the average plane deviation is adjusted in conjunction with the deformation percentage to obtain the final global flatness. The grouting hole axis detection module is used to capture images of the reserved grouting holes of the component using an industrial camera in a circular array. After contour sharpening processing of the image, the boundary of the grouting hole is extracted, and the three-dimensional axis of the grouting hole is constructed by combining the reflection characteristics of the inner sidewall of the hole. The axis deviation parameters of the entire component are determined based on the three-dimensional axis of all reserved grouting holes. Specifically, the three-dimensional axis of the grouting hole is constructed by combining the reflective characteristics of the inner sidewall of the hole as follows: The first step uses a light source at a 30° angle to the hole axis to illuminate the inner wall of the hole and capture the reflective band in the hole opening area; The second step uses a light source at a 60° angle to the hole axis to capture the reflective band in the middle area of ​​the hole; The third step uses a light source at an angle of 80° to the hole axis to capture the reflective band in the bottom area of ​​the hole; Connect the center points of the three stepped reflective strips in sequence to form the initial axis through the hole; The deviation of the initial axis from each step of the reflective strip is calculated, and the final three-dimensional axis is obtained by correcting it using the least squares method. The quality assessment module compares the global flatness and axis deviation parameters with preset qualified construction quality thresholds. If both the global flatness and axis deviation parameters meet the standards, the assembly quality is deemed qualified; if either exceeds the standard, the assembly quality is deemed unqualified.

Images