Floor wear resistance detection method and system

Abstract

The invention provides a floor wear resistance detection method and system, is applied to the technical field of floor wear detection, and reflects the wear resistance of a floor coating by obtaining the variation of porosity before and after floor wear. When the porosity after abrasion is obtained, a two-dimensional optical image of a to-be-detected surface is collected through multi-direction illumination, and the image is subjected to alignment processing. By analyzing the light and shade change information and the shadow form information of the dark area in different illumination directions, when the light and shade change information of the dark area changes along with the change of the position of the light source and the shadow form information shows the dynamic change of rotation or movement, it is judged that the hole is the floor hole. The discrimination mechanism based on the illumination dynamic change effectively solves the problem that a two-dimensional optical microscope scheme in the prior art cannot distinguish real pores and shallow scratches, and avoids misjudgment of scratches generated by a large amount of wear as pores, thereby significantly improving the accuracy of porosity calculation after wear.

Description

Technical Field

[0001] The present application relates to the technical field of floor wear detection, and in particular to a method and system for detecting floor wear resistance. Background Art

[0002] In modern industrial production, evaluating the wear resistance of ultra-thin functional coatings applied to new composite flooring surfaces is a critical step in ensuring product quality. Traditional wear testing methods face challenges in evaluating these ultra-thin coatings, as they are easily worn through, resulting in test results that fail to truly reflect the coating's performance. Even by adjusting test parameters to accommodate ultra-thin coatings and minimizing wear, traditional macroscopic measurements are often difficult to accurately quantify due to the small magnitude of the change. For example, microgram-level mass loss or nanometer-level wear depth may be below the resolvable reading of the measuring instrument, resulting in an extremely low signal-to-noise ratio and difficulty in reliably and accurately quantifying the degree of wear.

[0003] Faced with the macroscopic measurement dilemma, researchers turned to the microscopic level and discovered that even extremely slight wear can significantly alter the pore morphology of the coating surface, such as increasing the average pore size or making the pore profile irregular. This provides a new approach for indirectly assessing wear resistance: quantifying the change in porosity before and after wear to reflect the wear resistance of the coating.

[0004] However, this detection method based on porosity changes faces the challenge of production efficiency in actual industrial applications. To meet the needs of rapid detection, the technical team has tried to use three-dimensional profile measurement equipment, such as confocal microscopes or white light interferometers, which can effectively distinguish pores and shallow scratches through depth thresholds. However, obtaining high-precision three-dimensional data usually takes tens of minutes, which is too time-consuming and impractical. Therefore, in order to meet production efficiency requirements, the technical team had to return to a two-dimensional optical microscope solution that can achieve second-level imaging.

[0005] In these rapidly acquired two-dimensional optical images, the shallow and fine scratches produced by extremely mild wear exhibit image features highly similar to those of real pores. For example, they may appear as dark areas, and their sizes and shapes may overlap or be difficult to distinguish in the two-dimensional projection. This makes it impossible for conventional image processing algorithms, such as those based on simple grayscale threshold segmentation or basic morphological operations, to effectively distinguish these shallow scratches from real pores. As a result, when calculating porosity, image analysis algorithms will mistakenly identify a large number of scratches caused by wear as pores, thereby seriously overestimating the surface porosity after wear. This inaccurate measurement result makes it impossible to guarantee the accuracy of the method of evaluating wear resistance through porosity changes, and thus cannot meet the data reliability requirements of industrial quality inspection.

[0006] In view of the above problems, the existing technology needs to be improved urgently. Summary of the Invention

[0007] In view of the above-mentioned shortcomings of the existing technology, the present application provides a floor wear resistance detection method and system, which aims to solve the problem in the existing technology that when evaluating the wear resistance of ultra-thin functional coatings, traditional macro measurement methods are difficult to accurately quantify the degree of wear, and the porosity detection method based on two-dimensional optical images cannot effectively distinguish between real pores and shallow scratches caused by wear, resulting in inaccurate detection results.

[0008] In a first aspect, a method for testing the wear resistance of a floor comprises the following steps: S1: Obtain the initial porosity of the floor before wear; S2: Obtain two-dimensional optical images of the surface to be measured under multiple different lighting directions after the floor is worn, and align the two-dimensional optical images to obtain aligned images; S3: Identifying a dark area in the aligned image, and analyzing light and dark change information and shadow morphology information of the dark area under multiple different lighting directions; S4: When the brightness and darkness change information changes with the change of the light source position, and the shadow shape information presents a dynamic change of rotation or movement with the change of the light source position, it is determined that the corresponding dark area is a floor hole; S5: Calculating the porosity of the worn floor according to the total area of ​​the floor pores; S6: Compare the porosity of the floor after wear with the initial porosity to obtain a change in the porosity of the floor before and after wear, wherein the change in the porosity of the floor reflects the wear resistance of the floor coating.

[0009] This application proposes a method for detecting the wear resistance of flooring, which evaluates the wear resistance of floor coatings by analyzing the changes in the porosity of the floor before and after wear. In particular, at the microscopic level, the changes in image features under multi-directional illumination are used to accurately distinguish between real pores and wear scratches, thereby overcoming the limitations of traditional macroscopic measurement methods and the problem of misjudgment in two-dimensional optical image analysis, and improving the accuracy and reliability of detection.

[0010] Furthermore, step S1 includes: S11: Select multiple floor samples and select multiple detection areas on the surface of each floor sample; S12: For each of the detection areas, a two-dimensional optical image is collected using multi-directional illumination, and light and dark variation information and shadow morphology information of a dark area in the two-dimensional optical image under multiple different illumination directions are analyzed to identify and quantify the porosity of the detection area; S13: Calculate the average porosity of all the detection areas to obtain the initial porosity that is representative of the batch.

[0011] This application proposes a method for testing the wear resistance of flooring. By quantifying the porosity of multiple flooring samples under multi-region and multi-directional lighting and calculating the average value, a more representative and accurate initial porosity is obtained, providing reliable benchmark data for subsequent wear assessment and effectively reducing the accidental error of single-point or single measurement.

[0012] Furthermore, step S2 includes: S21: performing noise reduction processing on the two-dimensional optical image, and selecting an image in the two-dimensional optical image after the noise reduction processing as a reference image; S22: Detecting key feature points in the reference image and the remaining two-dimensional optical images after noise reduction processing; S23: Eliminate incorrect matching points from the key feature points according to the key feature points to obtain matching points; S24: Calculating a geometric transformation matrix between the reference image and the remaining two-dimensional optical images after noise reduction according to the matching points; S25: Using the geometric transformation matrix, geometrically correct the remaining two-dimensional optical images after the noise reduction processing, so that the remaining two-dimensional optical images after the noise reduction processing are accurately overlapped with the reference image to obtain an aligned image.

[0013] A floor wear resistance detection method proposed in this application ensures the precise alignment of images under different lighting directions by performing precise noise reduction, key feature point detection, error matching point elimination, and geometric correction on two-dimensional optical images under multi-directional lighting. This lays a solid foundation for the subsequent accurate identification and analysis of light and dark changes and shadow morphology information in dark areas, and effectively avoids misjudgment caused by image misalignment.

[0014] Furthermore, step S22 includes: S221: Analyzing brightness change information of local areas of the reference image and the remaining two-dimensional optical images after noise reduction processing; S222: Identifying pixels with high contrast or significant edge features based on the brightness change information; S223: Pixels with high contrast or significant edge features are used as key feature points.

[0015] A floor wear resistance detection method proposed in this application analyzes the brightness change information of the local area of ​​the image and identifies pixels with high contrast or significant edge features as key feature points, thereby improving the accuracy and robustness of key feature point detection and providing a more reliable matching basis for subsequent image alignment.

[0016] Furthermore, in step S3, identifying the dark area in the aligned image includes the steps of: S31: performing statistical analysis on the brightness value distribution of each pixel in the aligned image to generate a brightness histogram; S32: Obtaining brightness distribution characteristics of the brightness histogram, and calculating a bright / dark segmentation threshold according to the brightness distribution characteristics; S33: dividing the pixels whose brightness values ​​are less than or equal to the bright-dark segmentation threshold into dark area pixels; S34: performing a morphological operation on the dark area pixels to optimize the boundary of the area formed by the dark area pixels to obtain the optimized dark area.

[0017] Furthermore, in step S3, analyzing the light and dark change information and shadow shape information of the dark area under multiple different lighting directions includes the following steps: S35: For each dark area, calculating the grayscale centroid coordinates of the dark area in the image captured under different illumination directions; S36: Compare the grayscale centroid coordinates in the multiple images to obtain a displacement distance and a displacement direction of the grayscale centroid coordinates, wherein the shadow morphology information includes the displacement distance and the displacement direction; S37: For each pixel point in the dark area, calculate its grayscale value in the image collected under different lighting directions; S38: Compare the grayscale values ​​of the same pixel in multiple images to obtain a grayscale value change, wherein the brightness change information includes the grayscale value change.

[0018] Furthermore, step S5 includes: S51: Calculating the areas of all the dark regions, excluding the dark regions with areas smaller than a preset threshold, to obtain the dark regions used for calculation; S52: Accumulating the areas of the dark regions used for calculation to obtain the total area of ​​the floor pores; S53: Calculating the porosity of the worn floor according to the total area of ​​the floor pores.

[0019] Furthermore, step S6 includes: S61: Obtaining the floor porosity of the plurality of floor samples after wear, and calculating the average floor porosity of the plurality of floor samples; S62: Compare the average floor porosity with the initial porosity to obtain an average porosity change; S63: Evaluate the wear resistance of the floor coating according to the average porosity change.

[0020] Furthermore, step S63 includes: S631: Compare the average porosity change with a plurality of preset wear resistance level thresholds; S632: Determine the wear resistance level of the floor coating according to the comparison result.

[0021] In a second aspect, a floor wear resistance detection system is provided, for implementing any of the above methods, the system comprising: The first acquisition module: obtains the initial porosity of the floor before wear; The second acquisition module is used to obtain two-dimensional optical images of the surface to be measured under multiple different lighting directions after the floor is worn, and align the two-dimensional optical images to obtain aligned images; Recognition module: identifies the dark area in the aligned image and analyzes the light and dark change information and shadow morphology information of the dark area under multiple different lighting directions; A judgment module: when the brightness and darkness change information changes with the change of the light source position, and the shadow shape information presents a dynamic change of rotation or movement with the change of the light source position, it is judged that the corresponding dark area is a floor hole; Calculation module: calculates the porosity of the worn floor according to the total area of ​​the floor pores; Performance detection module: compares the porosity of the floor after wear with the initial porosity to obtain the change in porosity of the floor before and after wear, and the change in porosity of the floor reflects the wear resistance of the floor coating.

[0022] Beneficial Effects: This application proposes a method and system for testing floor wear resistance. By incorporating image feature analysis under multi-directional illumination, this method can effectively distinguish between true pores and shallow scratches caused by wear in two-dimensional optical images. Specifically, this method reflects the wear resistance of the floor coating by measuring the change in porosity before and after wear. To determine the post-wear porosity, the method uses multi-directional illumination to capture a two-dimensional optical image of the surface under test and aligns the images. Key to this is the analysis of the brightness and shadow morphology of dark areas under different illumination directions (such as the displacement of the grayscale centroid coordinates and the change in grayscale value) to accurately determine which dark areas are true pores. When the brightness and shadow morphology of dark areas changes with the position of the light source, and the shadow morphology exhibits dynamic changes such as rotation or movement, the dark areas are identified as floor pores. This identification mechanism, based on dynamic illumination changes, effectively addresses the problem of existing two-dimensional optical microscopy solutions that cannot distinguish between true pores and shallow scratches, preventing the misclassification of numerous wear-induced scratches as pores, thereby significantly improving the accuracy of post-wear porosity calculations. Compared with traditional macroscopic measurement methods, the method of the present application can accurately quantify extremely slight degrees of wear, overcoming the dilemma of traditional methods that are difficult to accurately quantify due to the small amount of change. At the same time, compared with the time-consuming three-dimensional profile measurement equipment, the present application adopts a two-dimensional optical microscope solution, combined with an innovative image analysis algorithm, to achieve second-level imaging and rapid detection, meeting the efficiency requirements of industrial production. In summary, the method of the present application can provide an accurate, efficient and reliable solution for evaluating the wear resistance of ultra-thin functional coatings, effectively solving the detection accuracy and efficiency problems existing in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 This is a flow chart of a floor wear resistance testing method proposed in this application.

[0024] Figure 2 This is a structural diagram of a floor wear resistance detection system proposed in this application.

[0025] Figure 3 This is a simplified schematic diagram of a floor wear resistance detection system proposed in this application.

[0026] Description of reference numerals: 201, first acquisition module; 202, second acquisition module; 203, identification module; 204, judgment module; 205, calculation module; 206, performance detection module. DETAILED DESCRIPTION

[0027] The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application generally described and marked in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application for protection, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work fall within the scope of protection of the present application.

[0028] It should be noted that similar reference numerals and letters represent similar items in the following figures. Therefore, once an item is defined in one figure, it does not need to be further defined or explained in subsequent figures. At the same time, in the description of this application, the terms "first", "second", etc. are only used to distinguish the description and should not be understood as indicating or implying relative importance.

[0029] Please refer to Figure 1 A method for detecting the wear resistance of a floor comprises the following steps: S1: Obtain the initial porosity of the floor before wear; S2: Obtain two-dimensional optical images of the surface to be measured under multiple different lighting directions after the floor is worn, and align the two-dimensional optical images to obtain aligned images; S3: Identify dark areas in the aligned images and analyze the brightness and shadow changes and shadow morphology of the dark areas under multiple different lighting directions; S4: When the brightness and darkness change information changes with the change of the light source position, and the shadow shape information presents a dynamic change of rotation or movement with the change of the light source position, the corresponding dark area is determined to be a floor hole; S5: Calculate the porosity of the floor after wear based on the total area of ​​the floor pores; S6: Compare the porosity of the floor after wear with the initial porosity to obtain the change in floor porosity before and after wear. The change in floor porosity reflects the wear resistance of the floor coating.

[0030] Specifically, in step S1, the initial porosity of the floor before wear can be obtained in a variety of ways. For example, the initial porosity can be obtained by performing surface testing on an unworn floor sample, capturing a two-dimensional optical image, and using image processing techniques to identify and quantify the surface pores. As a preferred embodiment, multiple floor samples can be selected, and porosity measurements can be taken at multiple test areas on each sample surface. The average of these measurements is then calculated as the initial porosity to improve representativeness.

[0031] In step S2, the acquisition of a two-dimensional optical image can be achieved by a microscope system equipped with an adjustable light source, which can illuminate the surface to be measured from different angles. The purpose of image alignment is to eliminate the slight displacement or rotation that may exist during the image acquisition process and ensure that images under different lighting directions can accurately overlap, so as to facilitate the subsequent accurate analysis of the changes in light and shadow in the same area. The alignment process can adopt a method based on image feature point matching. For example, a number of easily identifiable key feature points are selected from multiple images, and the relative position relationship of these key feature points in different images is calculated to infer the geometric transformation parameters between the images, and these parameters are used to correct the images.

[0032] In step S3, dark areas can be identified by analyzing the image's brightness histogram, setting an appropriate threshold for light-dark segmentation, and identifying pixels with brightness values ​​below the threshold as dark area pixels. Analysis of light-dark variation can be performed by comparing the grayscale value changes of pixels within the same dark area under different illumination directions. Analysis of shadow morphology can be performed by tracking the displacement and rotation of the geometric center or grayscale center of gravity of the dark area under different illumination directions.

[0033] In step S4, the core of this application lies in exploiting the difference in optical response between real pores and shallow scratches under different lighting conditions. Real pores have a certain depth and three-dimensional structure. When the position of the light source changes, the reflection and scattering properties within them will change, causing the brightness of the pores in the image to change, and the shadow area formed by them will rotate or move due to changes in the light source angle. In contrast, shallow scratches are usually shallow, and their optical response is insensitive to changes in the light source position, or the shadow shape does not change significantly. Therefore, by setting appropriate judgment criteria, it is possible to effectively distinguish between real pores and shallow scratches.

[0034] In step S5, the post-wear porosity of the floor is calculated based on the total area of ​​the identified floor pores. This typically involves summing the pixel areas of all dark regions identified as pores and then dividing the sum by the total area of ​​the measured surface to obtain the post-wear porosity. To improve calculation accuracy, dark regions that are too small or too large can be excluded to avoid misidentifying noise or atypical features as pores.

[0035] In step S6, the change in floor porosity directly reflects the degree of change in the surface pore structure of the floor coating during wear, and thus indirectly reflects the wear resistance of the floor coating. A greater change in floor porosity generally indicates more severe coating wear and poorer wear resistance; conversely, a smaller change in floor porosity indicates better wear resistance.

[0036] The solution of this application effectively solves the problem that traditional two-dimensional optical detection methods cannot distinguish between real pores and shallow scratches by introducing multi-directional illumination acquisition and dynamic optical response analysis. Compared with time-consuming three-dimensional profiling equipment, the method of this application can use quickly acquired two-dimensional optical images to achieve high-precision detection, significantly improving detection efficiency and meeting the needs of rapid detection in industrial production. As a result, this application can provide an accurate, efficient, and practical solution for evaluating the wear resistance of ultra-thin functional coatings, thereby ensuring product quality.

[0037] Furthermore, step S1 includes: S11: Select multiple floor samples and select multiple testing areas on the surface of each floor sample; S12: For each inspection area, a two-dimensional optical image is collected using multi-directional illumination, and the light and dark variation information and shadow morphology information of the dark area in the two-dimensional optical image under multiple different illumination directions are analyzed to identify and quantify the porosity of the inspection area; S13: Calculate the average porosity of all test areas to obtain an initial porosity representative of the batch.

[0038] In step S11, by selecting multiple detection areas, it is ensured that the obtained initial porosity can reflect the overall characteristics of a batch of floor products, rather than just the local characteristics of a single area or a single floor, thereby improving the representativeness of the data.

[0039] In step S12, the analysis process accurately distinguishes true pores from surface textures or impurities by observing the optical response of the dark areas under different lighting conditions, quantifying their area proportions and ensuring accurate pore identification. By analyzing the changes in brightness and shadow shape of the dark areas in the two-dimensional optical image under multiple different lighting directions and the changes in the lighting position, it can be determined that the dark areas are true voids, not scratches.

[0040] In step S13 , by averaging the porosities of multiple detection areas, the influence of single measurement errors can be effectively reduced, and the accuracy and reliability of the initial porosity can be further improved.

[0041] The solution of this application ensures that the initial porosity data obtained is highly representative and accurate by introducing a multi-sample, multi-area detection strategy when obtaining the initial porosity, combined with image analysis technology under multi-directional lighting. Specifically, the selection of multiple floor samples and multiple detection areas aims to cover the variability of floor product batches and avoid measurement deviations caused by local non-uniformity. At the same time, the use of multi-directional lighting to capture images and analyze light and dark variation information and shadow morphology information makes pore identification more accurate and can effectively eliminate interference from non-porous features. Therefore, by calculating the average value, the measurement data is further smoothed, so that the initial porosity can truly reflect the original state of the floor coating before wear.

[0042] Furthermore, step S2 includes: S21: performing noise reduction processing on the two-dimensional optical image, and selecting an image in the noise-reduced two-dimensional optical image as a reference image; S22: detecting key feature points in the reference image and the remaining two-dimensional optical images after noise reduction processing; S23: Eliminate incorrect matching points from the key feature points based on the key feature points to obtain matching points; S24: Calculating a geometric transformation matrix between the reference image and the remaining denoised two-dimensional optical images based on the matching points; S25: performing geometric correction on the remaining two-dimensional optical images after the noise reduction processing using a geometric transformation matrix, so that the remaining two-dimensional optical images after the noise reduction processing are accurately overlapped with the reference image, thereby obtaining an aligned image.

[0043] Specifically, in step S21, during the image acquisition process, due to various factors such as the limitations of the sensor itself, changes in lighting conditions, and environmental interference, different types of noise are often introduced into the image, such as Gaussian noise, salt and pepper noise, Poisson noise, etc. These noises will reduce the signal-to-noise ratio of the image, blur image details, and may even cause errors in subsequent feature extraction and image recognition. Therefore, the purpose of the solution in this application to first perform noise reduction processing on the two-dimensional optical image is to eliminate the noise that may be introduced during the image acquisition process, thereby improving the image quality and providing a clear data basis for subsequent feature detection and image alignment. Noise reduction processing can use a variety of image processing algorithms, such as Gaussian filtering.

[0044] Gaussian filtering is a commonly used linear smoothing filtering method. It is implemented by convolving a two-dimensional Gaussian kernel (or Gaussian template) with a two-dimensional optical image. This Gaussian kernel is a matrix in which each element represents the weight of the pixel at the corresponding position. As the Gaussian kernel slides across the image, it calculates the weighted sum of the pixels within the kernel and uses this sum as the new grayscale value of the center pixel. The size of the Gaussian kernel (for example, 3x3, 5x5) and the standard deviation (σ) are two key parameters of Gaussian filtering. The kernel size determines the filtering range, while the standard deviation controls the degree of smoothing. A larger standard deviation results in a flatter Gaussian function curve, a stronger filtering effect, and a more blurred two-dimensional optical image. Conversely, a smaller standard deviation results in a weaker filtering effect, preserving more detail in the two-dimensional optical image. This application utilizes Gaussian filtering to achieve noise reduction, effectively preserving edge details in two-dimensional optical images.

[0045] After noise reduction is complete, select one of the processed 2D optical images as a reference image. This reference image will serve as the basis for alignment of all other images. Typically, the image with the best quality, the richest features, or the image located in the middle of the sequence is selected as the reference image to optimize alignment.

[0046] Furthermore, in step S22, key feature points are local image features in the image that are unique, repeatable, and robust to changes in illumination, scale, rotation, and so on. These feature points can be corner points, edge points, or texture-rich regions. The SIFT (Scale-Invariant Feature Transform) algorithm can be used to detect key feature points.

[0047] The SIFT algorithm is a classic feature detection algorithm that detects local features in images at different scales and generates descriptors that are invariant to scale and rotation. Its core concept is to simulate images at different scales by constructing a Difference of Gaussian pyramid and to identify extreme points in these scale spaces as key points. These key points are then assigned directions and 128-dimensional descriptors are generated that effectively capture the local image information surrounding the key points. SIFT features are widely used in fields such as image matching and object recognition due to their robustness to image transformations.

[0048] Therefore, those skilled in the art can establish a correspondence between the reference image and the image to be aligned by detecting and matching these key feature points. These matching points are then used to calculate the geometric transformation matrix between the images, thereby achieving precise alignment of the images. For example, a two-dimensional optical image captured under multi-directional illumination may experience slight displacement or deformation due to changes in the light source angle. The stable feature points detected by the SIFT algorithm can effectively overcome these changes, ensuring that images under different illumination directions can accurately overlap, providing an accurate image foundation for subsequent dark area identification and analysis.

[0049] In step S23, when matching between the key feature points initially detected, a large number of false matches may be generated due to factors such as image noise, repeated textures, or changes in viewing angles. In order to ensure the accuracy of the alignment, these matching points need to be screened. The method for removing false matching points can use the RANSAC (Random Sample Consensus) algorithm. The RANSAC (Random Sample Consensus) algorithm is widely used to estimate the correct model parameters from a data set containing a large number of outliers. It can estimate model parameters from a data set containing a large number of outliers, thereby identifying and removing false matching points that do not conform to the geometric transformation model, and retaining the correct matching points.

[0050] Specifically, the implementation process of those skilled in the art using the RANSAC algorithm to remove incorrect matching points is roughly as follows: First, a minimum sample set is randomly selected: a minimum number of matching point pairs are randomly selected from all preliminary matching points (for example, for affine transformation, at least 3 matching point pairs are usually required; for perspective transformation, at least 4 matching point pairs are usually required). This minimum number of matching point pairs is assumed to be "inliers", that is, they are correct matches.

[0051] Secondly, estimate the model parameters: use these randomly selected matching point pairs to calculate the affine transformation model, and then use the coordinates of these pairs of matching points to solve the parameters of the transformation matrix.

[0052] Next, validate the model and count inliers: Use the estimated affine transformation model to transform all remaining matching points. For each matching point, calculate the distance (e.g., Euclidean distance) between its corresponding points before and after the transformation. If this distance is less than a preset error threshold, the matching point is considered consistent with the current model and marked as an "inlier." Otherwise, it is marked as an "outlier." Count the number of inliers supported by the current model.

[0053] Finally, iteration and optimal model selection are performed: the above random selection, model estimation, and inlier count process are repeated multiple times. After all iterations are complete, the model with the greatest number of inliers is selected as the final radiometric transformation model. All matching points that agree with this optimal model are considered correct, while those that disagree are identified and discarded as incorrect matches.

[0054] Thus, in step S24, the geometric transformation matrix describes the spatial mapping relationship between the images, such as translation, rotation, scaling, or perspective transformation. By using the accurate matching point pairs obtained in step S23, the least squares method can be used to calculate the geometric transformation matrix that can accurately transform one image to the coordinate system of another image. Among them, the least squares method is a commonly used optimization algorithm that can be used to estimate the geometric transformation matrix. Its basic idea is to find a geometric transformation matrix that minimizes the sum of the squared distances between all matching point pairs after the transformation.

[0055] Specifically, given a set of known point coordinates (x_s, y_s) in the reference image and the corresponding coordinates (x_t, y_t) in the remaining denoised 2D optical image, and assuming a radial transformation model, we construct an overdetermined system of linear equations and solve them using least squares to obtain the parameters of the geometric transformation matrix. This geometric transformation model minimizes the sum of the squared distances between all matching point pairs after the transformation.

[0056] Finally, in step S25, the geometric correction process applies the calculated geometric transformation matrix to the 2D optical image to be corrected. Using techniques such as pixel interpolation, each pixel in the image is mapped from its original position to the corresponding position in the reference image. This eliminates spatial misalignment between images captured under different illumination directions, ensuring precise spatial overlap of all images and providing a unified coordinate system for subsequent image analysis.

[0057] The solution of the present application ensures the purity of the image data by performing noise reduction processing on the original two-dimensional optical image, laying the foundation for subsequent processing. Subsequently, by detecting and accurately matching key feature points between images, the spatial correspondence between images is effectively captured. Furthermore, by eliminating erroneous matching points, the reliability of the matching is greatly improved, and the alignment deviation caused by erroneous matching is avoided. Based on these reliable matching points, the transformation matrix describing the geometric relationship between images can be accurately calculated. Finally, the geometric transformation matrix is ​​used to perform geometric correction on the image so that all images can be accurately overlapped in space. It is through this series of logically rigorous steps that the solution of the present application can overcome the slight displacement, rotation or deformation that may exist in image acquisition under different lighting directions, ensuring the high spatial consistency of multiple images.

[0058] Furthermore, step S22 includes: S221: Analyzing brightness change information of a local area of ​​the reference image and the remaining two-dimensional optical images after noise reduction processing; S222: Identifying pixels with high contrast or significant edge features based on brightness change information; S223: Pixels with high contrast or significant edge features are used as key feature points.

[0059] Analyzing brightness changes in local areas of an image involves examining each pixel and its surrounding area in the reference image and the remaining denoised 2D optical image, calculating the degree of change in brightness relative to surrounding pixels. For example, a gradient operator (such as the Sobel operator) can be used to calculate the image's gradient magnitude and direction to quantify the severity of the brightness change.

[0060] Specifically, the gradient of an image represents the rate of change of image brightness over space. The gradient magnitude reflects the intensity of the brightness change, or the strength of the edge, while the gradient direction indicates the direction of the brightness change, or the direction of the edge.

[0061] The Sobel operator is widely used in image processing for tasks such as edge detection, feature extraction, and image segmentation. It effectively extracts key information from images, such as object contours and textures. Those skilled in the art use the Sobel operator to approximate the horizontal and vertical gradients of an image (including a reference image and the remaining denoised two-dimensional optical image) by convolving a convolution kernel (typically a 3x3 matrix) with the image. For example, the Sobel operator contains two convolution kernels: one for detecting horizontal edges (Gx) and the other for detecting vertical edges (Gy). By convolving the image with these two convolution kernels, the horizontal and vertical gradient components for each pixel are obtained. The gradient magnitude (typically sqrt(Gx^2 + Gy^2)) and direction (typically arctan(Gy / Gx)) can then be calculated from these two components. A larger gradient magnitude calculated by the Sobel operator indicates a more dramatic brightness change at that pixel, and a higher likelihood of an image edge.

[0062] Furthermore, identifying pixels with high contrast or significant edge features based on the brightness change information involves filtering out pixels with the most significant changes from the brightness change information based on a preset threshold. High-contrast pixels are typically located at the boundary between light and dark in an image, while pixels with significant edge features form the image's outline. These pixels exhibit good stability under image transformations (such as rotation and scaling) and are less susceptible to changes in relative position due to minor perturbations. For example, in image processing, analyzing brightness changes in local areas of an image is necessary to identify key features. Brightness change information is typically obtained by calculating the image gradient, reflecting the degree of spatial variation in pixel brightness. After obtaining the brightness change information, pixels can be filtered by setting a preset threshold. For example, a gradient amplitude threshold can be set; only when the gradient amplitude of a pixel exceeds this threshold is the pixel considered to have a significant brightness change.

[0063] Therefore, the pixels with high contrast or significant edge features are identified as key feature points. These key feature points are the basis for subsequent image alignment processing, and they can provide enough information to accurately calculate the geometric transformation relationship between images.

[0064] The solution of the present application can effectively capture the distinctive structural features in the image by analyzing the brightness change information of the local area of ​​the image. It is precisely because the edges and high-contrast areas in the image are relatively stable under different lighting and viewing angles that the pixel points identified based on these areas are used as key feature points to provide a reliable benchmark for subsequent image alignment. When the image is translated, rotated or scaled, the position and relative relationship of these key feature points change with strong regularity, making the calculation of the geometric transformation matrix more accurate. Therefore, by identifying these pixel points with high contrast or significant edge features, it can be ensured that the selected feature points have high compatibility between different images, thereby laying the foundation for accurate image alignment.

[0065] Furthermore, in step S3, identifying the dark area in the aligned image includes the following steps: S31: performing statistical analysis on the brightness value distribution of each pixel in the aligned image to generate a brightness histogram; S32: Obtaining brightness distribution features of the brightness histogram, and calculating the bright and dark segmentation threshold according to the brightness distribution features; S33: dividing the pixels whose brightness values ​​are less than or equal to the bright-dark segmentation threshold into dark area pixels; S34: performing a morphological operation on the dark area pixels to optimize the boundary of the area formed by the dark area pixels to obtain an optimized dark area.

[0066] In step S31, the brightness histogram can intuitively reflect the distribution of the number of pixels at different brightness levels in the image, thereby providing basic data for subsequent image segmentation. Specifically, each pixel in the aligned image can be traversed to obtain its grayscale value or brightness value, and these values ​​are counted. The count results are mapped to a preset brightness range to ultimately form a brightness histogram.

[0067] Furthermore, in step S32, the brightness distribution characteristics include, but are not limited to, the peak value, valley value, distribution range, skewness, and kurtosis of the brightness histogram. In image processing, in order to effectively separate bright and dark areas, it is generally necessary to determine an appropriate threshold value to divide image pixels into foreground (bright area) and background (dark area). For example, the Otsu algorithm can be used to automatically determine the optimal threshold value.

[0068] The Otsu algorithm is a global threshold-based image segmentation method. Its core concept is to find an optimal threshold that can separate the pixels in the image into two categories (foreground and background) and maximize the variance between these two categories of pixels. Specifically, the Otsu algorithm iterates through all possible grayscale levels as thresholds. For each candidate threshold, it divides the pixels of the aligned image into two groups: one group consists of pixels with grayscale values ​​less than or equal to the candidate threshold, and the other group consists of pixels with grayscale values ​​greater than the candidate threshold. The algorithm then calculates the intra-class variance and inter-class variance of these two groups of pixels. The goal of the Otsu algorithm is to find a threshold that maximizes the inter-class variance. When the inter-class variance is maximized, the difference between the foreground and background is most significant, thus achieving the best segmentation effect. This method is particularly effective for images with a bimodal histogram because it can effectively separate the two main grayscale peaks.

[0069] In step S33 , the dark area pixels preliminarily constitute the dark area in the image, which generally corresponds to pores, depressions or other low reflectivity areas on the floor surface.

[0070] In step S34, morphological operations are performed on the delineated dark area pixels to further improve recognition accuracy and morphological integrity of the dark area. Morphological operations include, but are not limited to, erosion, dilation, opening, or closing operations, which can effectively remove noise points in the image, fill small breaks, smooth area boundaries, and connect adjacent dark area pixels.

[0071] Specifically, morphological operations are a series of nonlinear operations based on shape analysis in image processing. They are primarily applied to binary images, but can also be extended to grayscale images. These operations define a structuring element (or kernel), slide it across the image, and modify pixel values ​​based on the interaction between the structuring element and the local image region.

[0072] In this application's solution, the erosion operation shrinks foreground objects in the aligned images or removes pixels from the edges. When the structuring element slides across the aligned images, center pixels are retained only if the structuring element is completely contained within the foreground object. This operation effectively removes isolated noise points in the aligned images, as these noise points are typically small and cannot fully contain the structuring element, thus being eliminated after erosion.

[0073] Dilation is the opposite of erosion, enlarging foreground objects in aligned images or filling holes. As the structuring element slides across the image, any overlap between the structuring element and foreground objects causes the center pixel to be moved to the foreground. This operation can be used to fill small gaps in an image or connect adjacent dark pixels, creating a seamless whole.

[0074] The opening operation is a combination of an erosion operation followed by a dilation operation. It first erodes the aligned image to remove small noise points and fine connections, and then dilates to restore the main shape of the object. The main function of the opening operation is to smooth the object's outline, remove small protrusions, and break fine connections.

[0075] Closing is a combination of a dilation operation followed by an erosion operation. It first dilates the image to fill small holes and breaks, then erodes to restore the primary shape of the object. Closing smoothes the object's outline, fills small holes, and connects adjacent broken areas.

[0076] In summary, morphological operations can effectively remove noise points in images, fill small breaks, smooth area boundaries, and connect adjacent dark area pixels through basic operations such as erosion, dilation, opening and closing operations, thereby optimizing the morphology of dark areas in the image and making it more accurately reflect the actual pore structure.

[0077] Through these operations, the boundaries of the area formed by the dark area pixels can be optimized to make it more consistent with the actual pore morphology, thereby obtaining the optimized dark area and providing more accurate input for subsequent pore analysis.

[0078] Furthermore, in step S3, analyzing the light and dark change information and shadow shape information of the dark area under multiple different lighting directions includes the following steps: S35: For each dark area, calculate the grayscale centroid coordinates in the image collected under different illumination directions; S36: comparing the grayscale centroid coordinates in the multiple images to obtain the displacement distance and displacement direction of the grayscale centroid coordinates, where the shadow morphology information includes the displacement distance and displacement direction; S37: For each pixel point in the dark area, calculate its grayscale value in the image collected under different illumination directions; S38: Compare the grayscale values ​​of the same pixel in the multiple images to obtain a grayscale value variation. The brightness change information includes the grayscale value variation.

[0079] In step S35, for each identified dark region, the grayscale centroid coordinates of the region are calculated in images captured under different illumination directions. The grayscale centroid coordinates are the weighted average of the grayscale values ​​of the pixels within the image region and reflect the center of the brightness distribution within that region. These coordinates can be calculated by taking the weighted average of the grayscale values ​​and corresponding coordinates of all pixels within the dark region.

[0080] Specifically, assume that a dark area is composed of a series of pixels, each of which has its own specific coordinates (x, y) and grayscale value I(x', y'). Then, the grayscale centroid coordinates (Xc, Yc) of the dark area can be calculated using the following formula: Xc = Σ [x * I(x', y')] / Σ I(x', y') Yc = Σ [y * I(x', y')] / Σ I(x', y').

[0081] Here, Σ represents the sum of all pixels within the dark area. In this formula, each pixel's coordinates (x, y) are multiplied by its grayscale value I(x', y') as a weighting factor. All weighted coordinate values ​​are then added together and divided by the total grayscale value of all pixels within the area. This weighted averaging method gives brighter pixels a greater influence on the center of gravity, thereby more accurately reflecting the center of brightness distribution within the area.

[0082] In step S36, by comparing the grayscale centroid coordinates of the same dark area obtained under multiple different lighting directions, the displacement distance and direction of these grayscale centroid coordinates can be obtained. This displacement information collectively constitutes the shadow morphological information. For example, when the light source position changes, the shadow of the actual aperture will move or rotate accordingly, causing its grayscale centroid coordinates to shift. By quantifying this displacement, the dynamic characteristics of the shadow can be effectively captured.

[0083] In addition, in step S37, for each pixel in the dark area, its grayscale value in the image captured under different lighting directions needs to be calculated. The grayscale value reflects the brightness information of the pixel.

[0084] In step S38, by comparing the grayscale values ​​of the same pixel across multiple images, grayscale value variations are obtained. These grayscale value variations constitute information about brightness and shading variations. For example, when the light source position changes, the grayscale values ​​of pixels within a real pore will change significantly due to differences in light obstruction and reflection characteristics. By analyzing these grayscale value variations, it is possible to determine whether the area is a real pore, as non-pore surface defects (such as scratches and stains) typically do not exhibit similar brightness and shading patterns under different lighting conditions.

[0085] By calculating and comparing the grayscale centroid coordinates of dark areas under different lighting directions, the solution in this application can quantify the distance and direction of shadow displacement, thereby capturing the dynamic morphological changes of shadows. Furthermore, by comparing the grayscale values ​​of pixels within the dark areas, information about light and dark variations can be obtained. This information is used to determine whether the dark areas are actual floor pores. Under different lighting directions, the shadows within real floor pores will noticeably shift or rotate, and their brightness values ​​will change accordingly. However, other surface defects (such as stains and scratches) generally do not exhibit this regular light and shadow variation. This allows for more accurate distinction between real pores and non-pore dark areas.

[0086] Furthermore, step S5 includes: S51: Calculating the areas of all dark regions, excluding dark regions with areas smaller than a preset threshold, and obtaining dark regions for calculation; S52: Accumulate the areas of the dark regions used for calculation to obtain the total area of ​​the floor pores; S53: Calculate the porosity of the floor after wear based on the total area of ​​the floor pores.

[0087] Calculating the area of ​​all dark regions in step S51 refers to quantifying the area of ​​each independent dark region identified as floor pores in step S4. Excluding dark regions with an area smaller than a preset threshold is intended to filter out tiny noise points, artifacts, or minor non-pore defects that may have occurred during image acquisition or processing. These tiny regions may be too small to be representative or may not be actual floor pores.

[0088] The preset threshold can be determined through experiments to ensure that only dark areas that meet the real pore characteristics are included in subsequent calculations.

[0089] In step S52, the areas of the dark regions determined for calculation are summed to obtain the total area of ​​all real pores on the floor surface to be measured. This total area is a key indicator for measuring the porosity of the floor surface.

[0090] In step S53, the floor porosity can be defined as the ratio of the total area of ​​the floor pores to the total area of ​​the surface to be measured. For example, the porosity can be calculated as a percentage by dividing the total area of ​​the floor pores by the total area of ​​the surface to be measured and multiplying the result by 100%. This porosity directly reflects the surface integrity of the floor coating after wear.

[0091] The above technical solution effectively improves the accuracy and reliability of porosity calculations for worn floors. This improvement allows the calculated porosity to more accurately reflect the actual wear condition of the floor coating, providing a more precise data basis for subsequent wear resistance evaluations. This helps avoid errors introduced by noise or irrelevant features, making wear resistance assessments more objective.

[0092] Furthermore, step S6 includes: S61: Obtain floor porosity of multiple floor samples after wear, and calculate an average floor porosity of the multiple floor samples; S62: Compare the average floor porosity with the initial porosity to obtain the average porosity change; S63: Evaluate the wear resistance of floor coatings based on the change in average porosity.

[0093] In step S61, the porosity of each worn floor sample is calculated. These individually calculated porosities are then aggregated and averaged to obtain the average porosity of the multiple floor samples. This average provides a more comprehensive reflection of the overall porosity of a batch of worn floor samples.

[0094] In step S62, an average porosity change reflecting the overall wear degree can be obtained by subtracting the average porosity of the floor after wear from the initial porosity.

[0095] In step S63, the calculated average porosity change is used as a core indicator to quantify and evaluate the floor coating's ability to resist wear. For example, a larger average porosity change generally indicates poorer wear resistance; conversely, a smaller average porosity change indicates better wear resistance.

[0096] By testing multiple flooring samples and calculating the average porosity change, this proposed approach effectively addresses the potential for randomness and underrepresentation in single-sample test results. This statistical analysis of multiple samples allows the resulting average porosity change to more accurately and consistently reflect the true wear resistance of the entire batch of floor coatings. This multi-sample average-based evaluation method effectively reduces the impact of local defects or testing errors, thereby improving the reliability and accuracy of the evaluation results.

[0097] Furthermore, step S63 includes: S631: comparing the average porosity change with a plurality of preset wear resistance level thresholds; S632: Determine the wear resistance level of the floor coating based on the comparison results.

[0098] The preset wear resistance thresholds are numerical boundaries pre-set based on industry standards, product specifications, or empirical data to define different wear resistance levels. These thresholds can be discrete numerical points, such as 0.01, 0.03, 0.05, and so on, corresponding to different wear resistance ranges. The calculated average porosity change is compared with these preset thresholds to determine the specific range within which the change falls.

[0099] Furthermore, determining the wear resistance grade of the floor coating based on the comparison results involves categorizing it into a corresponding wear resistance grade based on the threshold range within which the average porosity change falls. For example, if the average porosity change is less than the first threshold, it may be classified as "Excellent"; if it is between the first and second thresholds, it may be classified as "Good," and so on. This allows for a quantitative and graded assessment of the wear resistance of floor coatings.

[0100] This application's solution implements a standardized, quantitative assessment of the wear resistance of floor coatings by introducing preset wear resistance thresholds and comparing average porosity changes with these thresholds. This threshold-based comparison mechanism makes the previously abstract "assessment" process concrete and actionable, mapping continuous porosity changes to discrete, easily understood grade classifications. This approach provides a clear basis for quality control, performance verification, and market classification of flooring products.

[0101] Please refer to Figure 2 、 Figure 3 A floor wear resistance detection system is used to implement any of the above methods, and the system includes: First acquisition module 201: used to obtain the initial porosity of the floor before wear. Specifically, this module can be configured to execute step S1 of the above method, namely, by selecting multiple floor samples and multiple detection areas on the surface of each floor sample, using multi-directional lighting to capture a two-dimensional optical image, and analyzing the light and dark changes and shadow morphology information of the dark areas in the two-dimensional optical image under multiple different lighting directions, thereby identifying and quantifying the porosity of the detection areas, and finally calculating the average porosity of all detection areas to obtain a representative initial porosity for the batch.

[0102] Second acquisition module 202: used to obtain two-dimensional optical images of the surface to be tested after floor wear under multiple different lighting directions, and align the two-dimensional optical images to obtain aligned images. Specifically, this module can be configured to execute step S2 of the above method, namely, performing noise reduction processing on the two-dimensional optical images, selecting an image as a reference image, detecting key feature points between the reference image and the remaining two-dimensional optical images after noise reduction, eliminating incorrect matching points based on the key feature points to obtain matching points, and calculating a geometric transformation matrix based on the matching points. Finally, using the geometric transformation matrix, geometrically correct the remaining two-dimensional optical images after noise reduction to ensure that they accurately overlap with the reference image, thereby obtaining an aligned image.

[0103] Identification module 203 is configured to identify dark regions in the aligned images and analyze the light-dark variation and shadow morphology information of the dark regions under multiple different illumination directions. Specifically, this module can be configured to execute step S3 of the above method, namely, statistically analyzing the brightness value distribution of each pixel in the aligned images to generate a brightness histogram, obtaining the brightness distribution characteristics of the brightness histogram and calculating the bright-dark segmentation threshold, classifying pixels with brightness values ​​less than or equal to the bright-dark segmentation threshold as dark region pixels, and performing morphological operations on the dark region pixels to optimize the boundaries of the dark region pixel-forming region, thereby obtaining optimized dark region regions. Furthermore, this module is responsible for calculating the grayscale centroid coordinates of each dark region in images captured under different illumination directions, comparing the grayscale centroid coordinates across multiple images to obtain displacement distances and displacement directions as shadow morphology information, and calculating the grayscale values ​​of each pixel in the dark region in images captured under different illumination directions, comparing the grayscale values ​​of the same pixel across multiple images to obtain grayscale value changes as light-dark variation information.

[0104] Determination Module 204 is configured to determine that the corresponding dark area is a floor void when the light-darkness variation information changes with the light source position, and the shadow shape information exhibits dynamic rotation or movement with the light source position. This module is configured to execute step S4 of the above method and accurately distinguish between true voids and surface texture or scratches by comprehensively analyzing the light-darkness variation information and shadow shape information.

[0105] Calculation module 205 is configured to calculate the post-wear floor porosity based on the total area of ​​the floor pores. Specifically, this module can be configured to execute step S5 of the above method, namely, calculating the areas of all dark regions, excluding dark regions with areas smaller than a preset threshold, obtaining dark regions for calculation, accumulating the areas of the dark regions for calculation to obtain the total area of ​​the floor pores, and calculating the post-wear floor porosity based on the total area of ​​the floor pores.

[0106] Performance Testing Module 206 is configured to compare the post-wear porosity of the floor with the initial porosity to determine the change in porosity before and after wear. This change in porosity reflects the wear resistance of the floor coating. This module is configured to execute step S6 of the aforementioned method, namely, obtaining the post-wear porosity of multiple floor samples, calculating the average porosity of the multiple floor samples, comparing the average porosity with the initial porosity to determine the average porosity change, and evaluating the wear resistance of the floor coating based on the average porosity change, for example by comparing it with multiple preset wear resistance thresholds to determine the wear resistance level of the floor coating.

[0107] The solution of the present application modularizes each step of the above-mentioned floor wear resistance detection method and gives it a specific hardware or software implementation carrier, thereby realizing the automation and standardization of the method. The various modules work together to ensure the accuracy of data acquisition, the robustness of image processing, the accuracy of pore identification and the reliability of the final wear resistance performance evaluation. Specifically, the first acquisition module 201 and the second acquisition module 202 serve as data input terminals, responsible for providing key image data of the floor surface before and after wear; the recognition module 203 and the judgment module 204 serve as core processing units, and accurately extract and identify floor pores from complex images through image analysis algorithms; the calculation module 205 quantifies the identified pores and converts them into comparable porosity data; finally, the performance detection module 206 directly reflects the wear resistance of the floor coating by comparing the changes in porosity before and after wear. This modular design makes the entire detection process efficient and repeatable, effectively solving the automation and efficiency problems that the method may face in practical applications.

[0108] In this document, relational terms such as first and second, etc. are used merely to distinguish one entity or operation from another entity or operation, but do not necessarily require or imply any actual relationship or order between these entities or operations.

[0109] The foregoing is merely an embodiment of the present application and is not intended to limit the scope of protection of the present application. Persons skilled in the art will readily appreciate that the present application may be modified and altered in various ways. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present application shall be included within the scope of protection of the present application.

Claims

1. A method for detecting floor wear resistance, characterized in that: The method comprises the steps of: S1: Obtain the initial porosity of the floor before wear; S2: Obtain two-dimensional optical images of the surface to be measured under multiple different lighting directions after the floor is worn, and align the two-dimensional optical images to obtain aligned images; S3: Identifying a dark area in the aligned image, and analyzing light and dark change information and shadow morphology information of the dark area under multiple different lighting directions; S4: When the brightness and darkness change information changes with the change of the light source position, and the shadow shape information presents a dynamic change of rotation or movement with the change of the light source position, it is determined that the corresponding dark area is a floor hole; S5: Calculating the porosity of the worn floor according to the total area of ​​the floor pores; S6: Compare the porosity of the floor after wear with the initial porosity to obtain a change in the porosity of the floor before and after wear, wherein the change in the porosity of the floor reflects the wear resistance of the floor coating.

2. A floor wear resistance detection method according to claim 1, characterized in that: Step S1 includes: S11: Select multiple floor samples and select multiple detection areas on the surface of each floor sample; S12: For each of the detection areas, a two-dimensional optical image is collected using multi-directional illumination, and light and dark variation information and shadow morphology information of a dark area in the two-dimensional optical image under multiple different illumination directions are analyzed to identify and quantify the porosity of the detection area; S13: Calculate the average porosity of all the detection areas to obtain the initial porosity that is representative of the batch.

3. A method for detecting floor wear resistance according to claim 1, characterized in that: Step S2 includes: S21: performing noise reduction processing on the two-dimensional optical image, and selecting an image in the two-dimensional optical image after the noise reduction processing as a reference image; S22: Detecting key feature points in the reference image and the remaining two-dimensional optical images after noise reduction processing; S23: Eliminate incorrect matching points from the key feature points according to the key feature points to obtain matching points; S24: Calculating a geometric transformation matrix between the reference image and the remaining two-dimensional optical images after noise reduction according to the matching points; S25: Using the geometric transformation matrix, geometrically correct the remaining two-dimensional optical images after the noise reduction processing, so that the remaining two-dimensional optical images after the noise reduction processing are accurately overlapped with the reference image to obtain an aligned image.

4. A method for detecting floor wear resistance according to claim 3, characterized in that: Step S22 includes: S221: Analyzing brightness change information of a local area of ​​the reference image and the remaining two-dimensional optical images after noise reduction processing; S222: Identifying pixels with high contrast or significant edge features based on the brightness change information; S223: Pixels with high contrast or significant edge features are used as key feature points.

5. A method for detecting floor wear resistance according to claim 3, characterized in that: In step S3, identifying the dark area in the aligned image includes the steps of: S31: performing statistical analysis on the brightness value distribution of each pixel in the aligned image to generate a brightness histogram; S32: Obtaining brightness distribution characteristics of the brightness histogram, and calculating a bright / dark segmentation threshold according to the brightness distribution characteristics; S33: dividing the pixels whose brightness values ​​are less than or equal to the bright-dark segmentation threshold into dark area pixels; S34: performing a morphological operation on the dark area pixels to optimize the boundary of the area formed by the dark area pixels to obtain the optimized dark area.

6. A method for detecting floor wear resistance according to claim 3, characterized in that: In step S3, analyzing the light and dark change information and shadow shape information of the dark area under multiple different lighting directions includes the following steps: S35: For each dark area, calculating the grayscale centroid coordinates of the dark area in the image captured under different illumination directions; S36: Compare the grayscale centroid coordinates in multiple images to obtain a displacement distance and a displacement direction of the grayscale centroid coordinates, wherein the shadow morphology information includes the displacement distance and the displacement direction; S37: For each pixel point in the dark area, calculate its grayscale value in the image collected under different lighting directions; S38: Compare the grayscale values ​​of the same pixel in multiple images to obtain a grayscale value change, wherein the brightness change information includes the grayscale value change.

7. A method for detecting floor wear resistance according to claim 1, characterized in that: Step S5 includes: S51: Calculating the areas of all the dark regions, excluding the dark regions with areas smaller than a preset threshold, to obtain the dark regions used for calculation; S52: Accumulating the areas of the dark regions used for calculation to obtain the total area of ​​the floor pores; S53: Calculating the porosity of the worn floor according to the total area of ​​the floor pores.

8. A method for detecting floor wear resistance according to claim 2, characterized in that: Step S6 includes: S61: Obtaining the floor porosity of the plurality of floor samples after wear, and calculating the average floor porosity of the plurality of floor samples; S62: Compare the average floor porosity with the initial porosity to obtain an average porosity change; S63: Evaluate the wear resistance of the floor coating according to the average porosity change.

9. A method for detecting floor wear resistance according to claim 8, characterized in that: Step S63 includes: S631: Compare the average porosity change with a plurality of preset wear resistance level thresholds; S632: Determine the wear resistance level of the floor coating according to the comparison result.

10. A floor wear resistance detection system, characterized in that: For implementing the method according to any one of claims 1 to 9, the system comprises: The first acquisition module: obtains the initial porosity of the floor before wear; The second acquisition module is used to obtain two-dimensional optical images of the surface to be measured under multiple different lighting directions after the floor is worn, and align the two-dimensional optical images to obtain aligned images; Recognition module: identifies the dark area in the aligned image and analyzes the light and dark change information and shadow morphology information of the dark area under multiple different lighting directions; A judgment module: when the brightness and darkness change information changes with the change of the light source position, and the shadow shape information presents a dynamic change of rotation or movement with the change of the light source position, it is judged that the corresponding dark area is a floor hole; Calculation module: calculates the porosity of the worn floor according to the total area of ​​the floor pores; Performance detection module: compares the porosity of the floor after wear with the initial porosity to obtain the change in porosity of the floor before and after wear, and the change in porosity of the floor reflects the wear resistance of the floor coating.

Images