Early warning system and method for aging degree of curtain wall aluminum veneer

Abstract

The invention discloses an early warning system and method for the aging degree of a curtain wall aluminum veneer, and relates to the technical field of image analysis, and the method comprises the steps: obtaining a to-be-detected aluminum veneer image, and dividing the to-be-detected aluminum veneer image into a plurality of pixel windows; the method comprises the following steps: acquiring an aging-free aluminum veneer with the same specification as a to-be-detected aluminum veneer and aluminum veneers with various aging characteristics, and establishing an influence model of the aging characteristics on the aging of the aluminum veneers; according to an influence model, obtaining an influence weight corresponding to each aging characteristic, and further screening out a first aging window and a second aging window; and combining the aging windows to obtain a first aging area and a second aging area in the to-be-detected aluminum single plate, and further performing early warning prompt and processing on the to-be-detected aluminum single plate. According to the method, different aging areas are obtained by carrying out image analysis on various aluminum veneers, the detection accuracy is improved, the aging trend of the aluminum veneers is predicted, treatment is carried out in advance, and the maintenance safety and the overall attractiveness of a building curtain wall are guaranteed.

Description

Technical Field

[0001] The invention relates to the technical field of image analysis, and in particular to an early warning system and method for the aging degree of curtain wall aluminum veneer. Background Art

[0002] Curtain wall aluminum veneer, also known as exterior wall aluminum veneer, is a single piece of aluminum plate made of aluminum alloy through cutting, bending, spraying and other processing techniques. It is mainly composed of panels, reinforcing ribs and corners. Curtain wall aluminum veneer is widely used in construction, including the exterior decoration of various buildings such as commercial buildings, office buildings, hotels, exhibition halls, theaters, etc. Through careful design and selection of aluminum veneers with different colors and textures, various styles and effects can be created to meet the overall design requirements of the building. However, due to seasonal changes, temperature differences between day and night, and long-term exposure to sunlight, the curtain wall aluminum veneer will age. The aging phenomenon is reflected in the characteristics of fading, deformation and corrosion. Aging will lead to safety hazards and impaired functions, causing significant direct or indirect losses and endangering personal safety.

[0003] The existing technologies for detecting aging phenomena include visual inspection and ultrasonic inspection. Although the above methods have their own advantages in terms of measurement speed and measurement method, they also have some shortcomings, such as high subjectivity and poor detection accuracy. Therefore, according to the influence of each aging characteristic on the aging degree, a comprehensive analysis is conducted to obtain the aging degree of the curtain wall aluminum veneer, and different early warnings and treatments are carried out to improve the detection accuracy, predict the aging trend of the aluminum veneer, and make treatments in advance, which is of great significance to the maintenance, safety and overall aesthetics of the building curtain wall. Summary of the invention

[0004] The purpose of the present invention is to provide an early warning system and method for the aging degree of curtain wall aluminum veneer to solve the problems raised in the prior art.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] An early warning method for the aging degree of curtain wall aluminum veneer comprises the following steps:

[0007] Step S100: obtaining an aluminum veneer image to be detected, and performing preprocessing, dividing the aluminum veneer image to be detected into a plurality of pixel windows according to each pixel point in the aluminum veneer image to be detected;

[0008] Step S200: obtaining non-aging aluminum veneers of the same specifications as the aluminum veneer to be tested, and aluminum veneers with various aging characteristics, and performing performance tests on them respectively; and analyzing the pre-shot images of the non-aging aluminum veneers and the images of the aluminum veneers with various aging characteristics, and establishing a model for the influence of various aging characteristics on the aging of the aluminum veneers;

[0009] Step S300: Obtaining the influence weight corresponding to each aging characteristic according to the influence model, obtaining the pixel points in each pixel window, obtaining the aging value of each pixel window according to the influence weight, and screening out the first aging window and the second aging window;

[0010] Step S400: According to the first aging window and the second aging window, the aging windows are merged to obtain the first aging area and the second aging area in the aluminum veneer to be detected, and then the aluminum veneer to be detected is given an early warning and processed.

[0011] Furthermore, step S100 includes:

[0012] Step S110: Take the image of the aluminum veneer to be inspected and convert it into a square grayscale image. Take the lower left corner of the grayscale image as the origin and establish a plane coordinate system of the grayscale image. The pixel window is composed of a number of pixels, and the minimum number of pixels in the pixel window is set to M. 1 *M 1 , the maximum number of pixels is M 2 *M 2 , set the feature window size equal to M 0 *M 0 , 1 <M 1 <M 0 <M 2 , and the lower left corner corresponds to the origin of the plane coordinate system, the upper right corner of the feature window is used as the target point, and the initial size of the window to be changed is set to M 0 *M 0 , and the lower left corner corresponds to the target point;

[0013] Among them, if the length and width of the aluminum veneer are not much different, it can be converted into a square by cutting or stretching. However, when the length and width of the aluminum veneer are significantly different, the aluminum veneer can be divided into multiple pieces, and each piece can be converted into a square. The specific size is determined based on actual conditions.

[0014] Step S120: Degrade the grayscale value of each pixel in the grayscale image to obtain a grayscale level of Q; based on each pixel in the feature window, the total number of pixels with a grayscale level of Q is recorded as V q , we get the probability that the gray level is level q, p(q) = V q / M 0 *M 0 , and then get the entropy value of the feature window Among them, log 2 The logarithm with base 2 is obtained by the same method. The entropy value H(X) of the window to be changed is obtained by 2 ; If H(X) 1 <H(X) 2, the size of the window to be changed is max() is to find the maximum value, To round up, if H(X) 1 >H(X) 2 , the size to be changed to When H(X) 2 =0, M = M 2 , To round down, if H(X) 1 =H(X) 2 , M=M 0 , and then the actual size of the window to be changed is M*M;

[0015] The larger the entropy value, the higher the disorder and chaos of the system. In information theory, entropy is a measure of uncertainty. The larger the amount of information, the smaller the uncertainty, and the smaller the entropy; the smaller the amount of information, the greater the uncertainty, and the greater the entropy. If there are three windows a, b and c, the entropy values ​​are H(a) = 1.271, H(b) = 1.843, and H(c) = 1 respectively; take a as the feature window, if b is the window to be changed, since the entropy value of a is smaller, it means that the grayscale value distribution on a is more uniform, and b is not as uniform as a, so the window situation of b is more complicated. In order to make a more detailed judgment on window b, the window of b should not be set too large, so the size of b should be relatively small; similarly, if c is the window to be changed, since the entropy value of c is smaller, it means that the grayscale value distribution on c is more uniform, and a is not as uniform as c, so the window situation of c is more uniform than a, so the window setting of window c should be larger;

[0016] Step S130: record the upper right corner of the window to be changed as the target point P. If the number of pixels in the grayscale image above and to the right of the target point P is less than M 2 *M 2 , then all the pixels in the upper right corner are gathered into a waiting window. If the number of pixels in the upper right corner is not less than M 2 *M 2 According to step S100, another window to be changed is obtained again, and the entropy values ​​of the two windows to be changed are obtained, and the actual size of the other window to be changed is obtained. The process is repeated in sequence to obtain several windows to be changed. The four sides of each window to be changed are extended to the edge of the grayscale image, and the minimum area surrounded by the extension lines is used as the pixel window, thereby obtaining all pixel windows.

[0017] Further, step S200 includes:

[0018] Step S210: pressure tests are performed on the non-aging aluminum veneer and the aluminum veneers with various aging characteristics, and the parts of the aluminum veneer that are subjected to pressure tests are used as test parts. The pressure test is used to determine the maximum pressure value that the aluminum veneer can withstand when it is subjected to pressure perpendicular to its surface. The maximum pressure value of the non-aging aluminum veneer is used as the target pressure value, the maximum pressure value of the aluminum veneer with aging characteristics is used as the aging pressure value, and the difference between the target pressure value and the aging pressure value is used as the pressure difference value, which is normalized;

[0019] Step S220: Obtain RGB images, grayscale images, and depth images of the non-aging aluminum veneer and each aging characteristic aluminum veneer, where the aging characteristics include fading, corrosion, and deformation; randomly obtain F pixels from the test part of the aluminum veneer AL with a certain aging characteristic as aging pixels, and randomly obtain a certain pixel a from the non-aging aluminum veneer. If the aging characteristic of the aluminum veneer AL is fading, according to the RGB value (R ^ ,G ^ ,B ^ ), and obtain the characteristic value of the test part Where e is the natural logarithm, R f , G f and B f are the R, G and B values ​​of the fth aged pixel respectively; if the aging characteristic is corrosion, according to the gray value V of a pixel a a , get the characteristic value V f is the gray value of the fth aged pixel; if the aging characteristic is deformation, according to the depth value d of a certain pixel a a , get the characteristic value d f is the depth value of the f-th aged pixel; and then according to the aging characteristics of the aluminum veneer AL, the aging characteristic difference of the aluminum veneer AL is obtained: n represents the type of aging characteristics, n = 1, 2 or 3, h 1 is the fading characteristic coefficient, h 2 is the corrosion characteristic coefficient, h 3 is the deformation characteristic coefficient;

[0020] Step S230: Establish a two-dimensional coordinate system with the aging characteristic difference as the horizontal coordinate and the pressure difference as the vertical coordinate; obtain the corresponding coordinates of each aluminum single plate in the two-dimensional coordinate system according to the aging characteristic difference and pressure difference of all aluminum single plates with a certain aging characteristic, and obtain the influence model of a certain aging characteristic as Y(n)=k according to the least square method. n *T+b n , where n represents the type of aging characteristics, Y(n) is the pressure difference, T is the aging characteristic difference, and k n is the slope, b n is the intercept, and then the influence model of each aging characteristic is obtained.

[0021] Further, step S300 includes: obtaining the influence weight of the nth aging characteristic according to the slope in each influence model. where k n is the slope of the nth influence model, and N is the total number of influence models; then, according to the average RGB value, average grayscale value, and average depth value of the pixels in a certain pixel window, and based on step S200, the difference of each aging characteristic corresponding to a certain pixel window and the corresponding pressure difference of each are obtained, and normalization is performed. Furthermore, the aging value of a certain pixel window is obtained as: Y n is the nth pressure difference; set the first aging degree threshold D 1 and the second aging degree threshold D 2 , 0 < D 1 < D 2 , if D 1 < D ≤ D 2 , it is determined that a certain pixel window is the first aging window. If D 2 < D, it is determined that a certain pixel window is the second aging window.

[0022] It should be noted that the slope of the model is used here to represent the weight of each aging characteristic, only because the slope of the model here represents the influence degree of the change of the aging characteristic on the pressure value. Moreover, when the slope is larger, it indicates that the corresponding aging characteristic has a greater influence on the pressure value, and the pressure value is a specific manifestation directly reflecting the aging degree. Therefore, the slope of the model is used in this solution to represent the weight of each aging characteristic. In this solution, the first aging window is a pixel window with a smaller aging degree, and the second aging window is a pixel window with a larger aging degree. The type distinction of the aging window is mainly for the rationality of merging regions in the following steps.

[0023] Further, step S400 includes:

[0024] Step S410: Establish a two-dimensional coordinate system for the image of the aluminum single plate to be detected, and obtain the central coordinates of each pixel window therein; randomly obtain an aging window, the type of the aging window is L, and obtain all pixel windows in the eight neighborhoods around the aging window, all of which are used as eight-neighborhood windows. If the number of eight-neighborhood windows of type L is greater than the preset number threshold, the window types of all eight-neighborhood windows are set as L, and the aging value remains unchanged, and the aging window and all eight-neighborhood windows are marked; if the number is not greater than the number threshold, randomly obtain another aging window, and repeat this step S410;

[0025] In this solution, the eight-domain windows are eight adjacent pixel windows around a pixel window in image processing, which is a conventional term in the prior art, namely, adjacent pixel windows in eight directions of up, down, left, right, upper left, lower left, upper right and lower right, and the number threshold is 6. Because when the number threshold is 6, all the up, down, left, right and any one of the pixel windows can be taken into account. If the aging degree in any one of the up, down, left, right and any one of the directions is L, in order to facilitate merging, it is determined that the types of all the surrounding pixel windows are L at this time, so the types of all the pixel windows adjacent to the eight-domain windows are L.

[0026] Step S420: The center coordinate of a certain aging window is used as P 0 , the aging value is D 0 , the center coordinate of the gth eight-domain window is taken as P g , the aging value is D g , if D 0 <D g , with P 0 As the starting point, point to P g Direction, size |D g -D 0 |, get the vector If D 0 >D g , with P 0 As the starting point, point to P g The opposite direction of |D g -D 0 |, get the vector Then, according to each eight-domain window, all vectors are obtained, and the vectors are added to obtain the pointing vector V corresponding to a certain pixel window 0 ; Then get the distance pointing vector V 0 The center coordinates closest to the end point are the corresponding pixel window WIN; if the pixel window WIN is not an aging window, according to step S400, a certain unmarked aging window and the surrounding eight-domain windows are randomly obtained; if the pixel window WIN is an aging window, the eight-domain windows around the pixel window WIN are obtained, and so on until all the aging windows are marked, and the final multiple adjacent aging windows of the same type are taken as an aging area.

[0027] This plan does not make an overall judgment on the curtain wall aluminum veneer, but only judges the local aging degree, and divides the aging degree into mild and severe degrees. Mild is the first aging area, and severe is the second aging area. What we hope to achieve in the future is to implement different treatments for different aging areas, such as cleaning and waxing maintenance for mildly aged areas, and coating repairs for severe areas, to ensure its adhesion and durability, extend the service life of the curtain wall aluminum veneer, and ensure the safety and aesthetics of the building.

[0028] An early warning system for the aging degree of curtain wall aluminum veneer, comprising a pixel window division module, an impact model establishment module, an aging window screening module and an aging window merging module;

[0029] Pixel window division module: used to obtain the aluminum veneer image to be detected, and perform preprocessing, and divide the aluminum veneer image to be detected into multiple pixel windows according to each pixel point in the aluminum veneer image to be detected;

[0030] Impact model building module: used to obtain non-aging aluminum veneers of the same specifications as the aluminum veneers to be tested, as well as aluminum veneers with various aging characteristics, and perform performance tests on them respectively; and analyze the pre-shot images of non-aging aluminum veneers and images of aluminum veneers with various aging characteristics, and establish an impact model of various aging characteristics on the aging of aluminum veneers;

[0031] Aging window screening module: used to obtain the influence weight corresponding to each aging characteristic according to the influence model, obtain the pixel points in each pixel window, obtain the aging value of each pixel window according to the influence weight, and screen out the first aging window and the second aging window;

[0032] Aging window merging module: used to merge the aging windows according to the first aging window and the second aging window, obtain the first aging area and the second aging area in the aluminum veneer to be detected, and then provide early warning prompts and processing for the aluminum veneer to be detected.

[0033] Further, the pixel window division module includes a pixel window setting unit, a window to be changed determination unit and a pixel window division unit;

[0034] Pixel window setting unit: used to capture the image of the aluminum veneer to be inspected and convert it into a square grayscale image. The pixel window is composed of several pixels. The minimum pixel window, the maximum pixel window, the characteristic window and the window to be changed are set.

[0035] The window to be changed determining unit is used to perform a degradation process on the grayscale value of each pixel in the grayscale image, thereby obtaining the entropy value of the feature window and the entropy value of the window to be changed, thereby obtaining the actual size of the window to be changed;

[0036] Pixel window division unit: used to obtain all the windows to be changed, extend the four sides of each window to be changed to the edge of the grayscale image, and use the minimum area surrounded by the extension lines as the pixel window, thereby obtaining all the pixel windows.

[0037] Further, the aging window merging module includes an eight-domain window acquisition unit and an aging window merging unit;

[0038] Eight-domain window acquisition unit: used to establish a two-dimensional coordinate system of the aluminum veneer image to be detected, obtain the center coordinates of each pixel window therein; randomly obtain a certain aging window and all the surrounding eight-domain windows; and mark the pixel window;

[0039] Aging window merging unit: used to obtain a vector corresponding to a certain aging window according to a certain aging window and an eight-domain window; then obtain the pixel window WIN, and make judgments until all aging windows are marked.

[0040] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention provides an early warning system and method for the aging degree of aluminum veneers for curtain walls, obtains an image of an aluminum veneer to be detected, and divides the image of the aluminum veneer to be detected into multiple pixel windows; obtains an aluminum veneer with the same specifications as the aluminum veneer to be detected, and an aluminum veneer with various aging characteristics, and establishes an influence model of each aging characteristic on the aging of the aluminum veneer; according to the influence model, the influence weight corresponding to each aging characteristic is obtained, and then the first aging window and the second aging window are screened out; each aging window is merged to obtain the first aging area and the second aging area in the aluminum veneer to be detected, and then the aluminum veneer to be detected is early-warned and processed. The present invention obtains different aging areas by performing image analysis on various aluminum veneers, improves the detection accuracy, predicts the aging trend of the aluminum veneer, and makes processing in advance. According to different aging degrees, corresponding processing methods are made to ensure the maintenance safety and overall beauty of the building curtain wall. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 It is a flow chart of an early warning method for aging degree of curtain wall aluminum veneer of the present invention;

[0042] Figure 2 The present invention is a structural diagram of an early warning system for the aging degree of curtain wall aluminum panels. DETAILED DESCRIPTION

[0043] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0044] Example: Figure 1 As shown, the present invention provides a technical solution of an early warning system and method for the aging degree of curtain wall aluminum veneer, comprising the following steps:

[0045] Step S100: Acquire the aluminum veneer image to be detected, perform preprocessing, and divide the aluminum veneer image to be detected into a plurality of pixel windows according to each pixel point in the aluminum veneer image to be detected.

[0046] Step S110: Take the image of the aluminum veneer to be inspected and convert it into a square grayscale image. Take the lower left corner of the grayscale image as the origin and establish a plane coordinate system of the grayscale image. The pixel window is composed of a number of pixels, and the minimum number of pixels in the pixel window is set to M. 1 *M 1 , the maximum number of pixels is M 2 *M 2 , set the feature window size equal to M 0 *M 0 , 1 <M 1 <M 0 <M 2 , and the lower left corner corresponds to the origin of the plane coordinate system, the upper right corner of the feature window is used as the target point, and the initial size of the window to be changed is set to M 0 *M 0 , and the lower left corner corresponds to the target point.

[0047] Step S120: Degrade the grayscale value of each pixel in the grayscale image to obtain a grayscale level of Q; based on each pixel in the feature window, the total number of pixels with a grayscale level of Q is recorded as V q , we get the probability that the gray level is level q, p(q) = V q / M 0 *M 0 , and then get the entropy value of the feature window Among them, log 2 The logarithm with base 2 is obtained by the same method. The entropy value H(X) of the window to be changed is obtained by 2 ; If H(X) 1 <H(X) 2 , the size of the window to be changed is max() is to find the maximum value, To round up, if H(X) 1 >H(X) 2 , the size to be changed to When H(X) 2 =0, M = M 2 , To round down, if H(X) 1 =H(X) 2 , M=M 0 , and then the actual size of the window to be changed is M*M.

[0048] The larger the entropy value, the higher the degree of disorder and chaos of the system. In information theory, entropy is a measure of uncertainty. The larger the amount of information, the smaller the uncertainty and the smaller the entropy; the smaller the amount of information, the greater the uncertainty and the greater the entropy. In order to better represent, the following embodiments are given: In this embodiment, the grayscale level is reduced to Q = 8 levels. Since the original grayscale value is 0-255, the original 256 grayscale values ​​are now divided into 8 intervals, that is, the grayscale value is now represented by 0-7; if there are three pixel windows a, b and c, and the window sizes are all 5*5, the number of pixels with grayscale values ​​from 0-7 on each window are a = (15, 8, 2, 0, 0, 0, 0, 0), b = (9, 8, 6, 2, 0, 0, 0, 0) and c = (13, 12, 0, 0, 0, 0, 0, 0), then according to the formula The entropy values ​​of these three windows are H(a)=1.271, H(b)=1.843, and H(c)=1, respectively. Taking a as the feature window, if b is the window to be changed, since the entropy value of a is smaller, it means that the grayscale value distribution on a is more uniform, and since the distribution of b is not as uniform as that of a, the window situation of b is more complicated. In order to make a more detailed judgment on window b, the window of b should not be set too large, so the size of b should be relatively small. Similarly, if c is the window to be changed, since the entropy value of c is smaller, it means that the grayscale value distribution on c is more uniform, and since the distribution of a is not as uniform as that of c, the window situation of c is more uniform than that of a, so the window size of window c should be set larger.

[0049] Step S130: record the upper right corner of the window to be changed as the target point P. If the number of pixels in the grayscale image above and to the right of the target point P is less than M 2 *M 2 , then all the pixels in the upper right corner are gathered into a waiting window. If the number of pixels in the upper right corner is not less than M 2 *M 2 According to step S100, another window to be changed is obtained again, and the entropy values ​​of the two windows to be changed are obtained, and the actual size of the other window to be changed is obtained. The process is repeated in sequence to obtain several windows to be changed. The four sides of each window to be changed are extended to the edge of the grayscale image, and the minimum area surrounded by the extension lines is used as the pixel window, thereby obtaining all pixel windows.

[0050] Step S200: Obtain non-aging aluminum veneers of the same specifications as the aluminum veneer to be tested, as well as aluminum veneers with various aging characteristics, and perform performance tests on them respectively; and analyze the pre-shot images of the non-aging aluminum veneers and the images of the aluminum veneers with various aging characteristics, and establish a model for the influence of various aging characteristics on the aging of the aluminum veneers.

[0051] Step S210: Perform pressure tests on the non-aging aluminum veneer and the aluminum veneers with aging characteristics, and use the part of the aluminum veneer where the pressure test is performed as the test part. The pressure test is used to determine the maximum pressure value that the aluminum veneer can withstand when it is subjected to a pressure perpendicular to its surface. The maximum pressure value of the non-aging aluminum veneer is used as the target pressure value, the maximum pressure value of the aluminum veneer with aging characteristics is used as the aging pressure value, and the difference between the target pressure value and the aging pressure value is used as the pressure difference value, which will be normalized.

[0052] Step S220: Obtain RGB images, grayscale images, and depth images of the non-aging aluminum veneer and each aging characteristic aluminum veneer, where the aging characteristics include fading, corrosion, and deformation; randomly obtain F pixels from the test part of the aluminum veneer AL with a certain aging characteristic as aging pixels, and randomly obtain a certain pixel a from the non-aging aluminum veneer. If the aging characteristic of the aluminum veneer AL is fading, according to the RGB value (R ^ ,G ^ ,B ^ ), and obtain the characteristic value of the test part Where e is the natural logarithm, R f , G f and B f are the R, G and B values ​​of the fth aged pixel respectively; if the aging characteristic is corrosion, according to the gray value V of a pixel a a , get the characteristic value V f is the gray value of the fth aged pixel; if the aging characteristic is deformation, according to the depth value d of a certain pixel a a , get the characteristic value d f is the depth value of the f-th aged pixel; and then according to the aging characteristics of the aluminum veneer AL, the aging characteristic difference of the aluminum veneer AL is obtained: n represents the type of aging characteristics, n = 1, 2 or 3, h 1 is the fading characteristic coefficient, h 2 is the corrosion characteristic coefficient, h 3 is the deformation characteristic coefficient.

[0053] In this solution, only three common types of fading, corrosion and deformation are given, and these three types can basically cover most of the aging phenomena. In actual scenarios, they may not be limited to these three types, but their basic ideas and methods are relatively similar. For example, characteristic values ​​can be obtained through variance and RGB values. For specific implementation, please refer to this solution and will not be described in detail here.

[0054] Step S230: Establish a two-dimensional coordinate system with the aging characteristic difference as the abscissa and the pressure difference as the ordinate; for all aluminum veneers with a certain aging characteristic, based on the corresponding aging characteristic differences and pressure differences, obtain the coordinates corresponding to each aluminum veneer in the two-dimensional coordinate system, and according to the least squares method, obtain the influence model for a certain aging characteristic as Y(n) = k n *T + b n , where n represents the type of aging characteristic, Y(n) is the pressure difference, T is the aging characteristic difference, k n is the slope, and b n is the intercept, and then obtain the influence model for each aging characteristic.

[0055] Step S300: According to the influence model, obtain the influence weights corresponding to each aging characteristic, acquire the pixel points in each pixel window, and according to the influence weights, obtain the aging value of each pixel window, and screen out the first aging window and the second aging window.

[0056] According to the slope in each influence model, obtain the influence weight of the nth aging characteristic where k n is the slope of the nth influence model, and N is the total number of influence models; then, based on the average RGB value, average grayscale value, and average depth value of the pixels in a certain pixel window, according to Step S200, obtain the corresponding aging characteristic difference and corresponding pressure difference for each aging characteristic of a certain pixel window, and perform normalization. Furthermore, the aging value of a certain pixel window is: Y n is the nth pressure difference; set the first aging degree threshold D 1 and the second aging degree threshold D 2 , 0 < D 1 < D 2 , if D 1 < D ≤ D 2 , determine that a certain pixel window is the first aging window; if D 2 < D, determine that a certain pixel window is the second aging window.

[0057] It should be noted that here the slope of the model is used to represent the weights of each aging characteristic, only because the slope of the model here represents the degree of influence of the change of the aging characteristic on the pressure value, and when the slope is larger, it indicates that the corresponding aging characteristic has a greater influence on the pressure value, and the pressure value is a direct manifestation of the aging degree. Therefore, in this solution, the slope of the model is used to represent the weights of each aging characteristic. In this solution, the first aging window is the pixel window with a smaller aging degree, and the second aging window is the pixel window with a larger aging degree. The type distinction of the aging window is mainly for the rationality of the following step of merging regions.

[0058] Step S400: According to the first aging window and the second aging window, the aging windows are merged to obtain the first aging area and the second aging area in the aluminum veneer to be detected, and then the aluminum veneer to be detected is given an early warning and processed.

[0059] Step S410: Establish a two-dimensional coordinate system of the aluminum veneer image to be detected, and obtain the center coordinates of each pixel window therein; randomly obtain an aging window, the aging window type is L, obtain all pixel windows in the eight fields around the aging window, and use them as eight-field windows. If the number of eight-field windows with window type L is greater than a preset number threshold, use the window type of all eight-field windows as L, and the aging value remains unchanged, and mark the aging window and all eight-field windows; if the number is not greater than the number threshold, randomly obtain another aging window, and repeat this step S410.

[0060] In this scheme, the eight-domain windows are eight adjacent pixel windows around a pixel window in image processing, which is the conventional term in the prior art, i.e., adjacent pixel windows in eight directions of up, down, left, right, upper left, lower left, upper right and lower right, and the quantity threshold is 6. Because when the quantity threshold is 6, all the up, down, left, right and any other directions of the pixel window can be taken into account. If the aging degree in any of the up, down, left, right and any other directions is L, in order to facilitate merging, it is determined that the types of all the surrounding pixel windows are L, so the types of all the pixel windows adjacent to the eight-domain windows are L.

[0061] Step S420: The center coordinate of a certain aging window is used as P 0 , the aging value is D 0 , the center coordinate of the gth eight-domain window is taken as P g , the aging value is D g , if D 0 <D g , with P 0 As the starting point, point to P g Direction, size |D g -D 0 |, get the vector If D 0 >D g , with P 0 As the starting point, point to P g The opposite direction of |D g -D 0 |, get the vector Then, according to each eight-domain window, all vectors are obtained, and the vectors are added to obtain the pointing vector V corresponding to a certain pixel window 0 ; Then get the distance pointing vector V 0The center coordinates closest to the end point are the corresponding pixel window WIN; if the pixel window WIN is not an aging window, according to step S400, a certain unmarked aging window and the surrounding eight-domain windows are randomly obtained; if the pixel window WIN is an aging window, the eight-domain windows around the pixel window WIN are obtained, and so on until all the aging windows are marked, and the final multiple adjacent aging windows of the same type are taken as an aging area.

[0062] This plan does not make an overall judgment on the curtain wall aluminum veneer, but only judges the local aging degree, and divides the aging degree into mild and severe degrees. Mild is the first aging area, and severe is the second aging area. What we hope to achieve in the future is to implement different treatments for different aging areas, such as cleaning and waxing maintenance for mildly aged areas, and coating repairs for severe areas, to ensure its adhesion and durability, extend the service life of the curtain wall aluminum veneer, and ensure the safety and aesthetics of the building.

[0063] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. 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, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A warning method for the aging degree of curtain wall aluminum veneer, characterized in that: The steps include: Step S100: Obtain the aluminum veneer image to be detected, and perform preprocessing. According to each pixel point in the aluminum veneer image to be detected, divide the aluminum veneer image to be detected into multiple pixel windows; Step S200: Obtain non-aged aluminum veneers with the same specifications as the aluminum veneer to be detected, as well as aluminum veneers with various aging characteristics, and perform performance tests respectively; and analyze the pre-shot images of non-aged aluminum veneers and aluminum veneers with various aging characteristics, and establish an influence model of various aging characteristics on the aging of aluminum veneers; Step S300: According to the influence model, obtain the influence weights corresponding to various aging characteristics, obtain the pixel points in each pixel window, and according to the influence weights, obtain the aging value of each pixel window, and screen out the first aging window and the second aging window; Step S400: According to the first aging window and the second aging window, merge the aging windows to obtain the first aging area and the second aging area in the aluminum veneer to be detected, and then perform warning prompts and processing on the aluminum veneer to be detected.

2. The early warning method for aging degree of curtain wall aluminum veneer according to claim 1 is characterized in that: Step S100 includes: Step S110: Shoot the aluminum veneer image to be detected and convert it into a square grayscale image. Taking the lower left corner of the grayscale image as the origin, establish a plane coordinate system of the grayscale image; the pixel window consists of several pixels. Set the minimum number of pixels of the pixel window to M1*M1, the maximum number of pixels to M2*M2, set the size of the feature window to be equal to M0*M0, 1<M1<M0<M2, and the lower left corner position corresponds to the origin in the plane coordinate system. Take the upper right corner of the feature window as the target point, set the initial size of the variable window to M0*M0, and the lower left corner position corresponds to the target point; Step S120: Degrade the gray value of each pixel in the grayscale image to obtain Q-level gray levels; according to each pixel in the feature window, record the total number of pixels with the gray level of q-level as V q , and obtain the probability p(q) of the gray level of q-level as p(q) = V q / M0*M0, and then obtain the entropy value of the feature window where, log2 is the logarithm with base 2. Similarly, obtain the entropy value H(X)2 of the window to be changed; if H(X)1 < H(X)2, obtain the size to be changed of the window to be changed as max() is to find the maximum value, is to round up. If H(X)1 > H(X)2, the size to be changed is When H(X)2 = 0, M = M2, is to round down. If H(X)1 = H(X)2, M = M0, and then obtain the actual size of the window to be changed as M*M; Step S130: Denote the upper right corner of the variable window as the target point P. If the number of pixels in the upper right of the target point P in the grayscale image is less than M2*M2, converge all the pixels in the upper right into a variable window. If the number of pixels in the upper right is not less than M2*M2, obtain another variable window according to this step S100, and obtain the entropy values of the two variable windows, obtain the actual size of the other variable window, and loop in turn to obtain several variable windows. Extend the four sides of each variable window to the edge of the grayscale image, and take the smallest area surrounded by the extended lines as the pixel window, and then obtain all pixel windows.

3. The early warning method for aging degree of curtain wall aluminum veneer according to claim 1 is characterized in that: Step S200 includes: Step S210: Perform pressure tests on both the non-aged aluminum veneer and the aluminum veneers with various aging characteristics. Take the part of the aluminum veneer where the pressure test is performed as the test part. The pressure test is used to determine the maximum pressure value that the aluminum veneer can withstand when承受 perpendicular pressure to its surface. Take the maximum pressure value of the non-aged aluminum veneer as the target pressure value, and the maximum pressure value of the aluminum veneer with aging characteristics as the aging pressure value. Take the difference between the target pressure value and the aging pressure value as the pressure difference and perform normalization; Step S220: Obtain RGB images, grayscale images, and depth images of the non-aging aluminum veneer and each aging characteristic aluminum veneer, where the aging characteristics include fading, corrosion, and deformation; randomly obtain F pixels from the test part of the aluminum veneer AL with a certain aging characteristic as aging pixels, and randomly obtain a certain pixel a from the non-aging aluminum veneer. If the aging characteristic of the aluminum veneer AL is fading, according to the RGB value (R ^ ,G ^ ,B ^ ), and obtain the characteristic value of the test part Where, e is the natural logarithm, R f , G f and B f are the R, G and B values ​​of the fth aged pixel respectively; if the aging characteristic is corrosion, according to the gray value V of a pixel a a , get the characteristic value V f is the gray value of the fth aged pixel; if the aging characteristic is deformation, according to the depth value d of a certain pixel a a , get the characteristic value d f is the depth value of the f-th aged pixel; and then according to the aging characteristics of the aluminum veneer AL, the aging characteristic difference of the aluminum veneer AL is obtained: n represents the type of aging characteristics, n = 1, 2 or 3, h1 is the fading characteristic coefficient, h2 is the corrosion characteristic coefficient, and h3 is the deformation characteristic coefficient; Step S230: Establish a two-dimensional coordinate system with the aging characteristic difference as the horizontal coordinate and the pressure difference as the vertical coordinate; obtain the corresponding coordinates of each aluminum single plate in the two-dimensional coordinate system according to the aging characteristic difference and pressure difference of all aluminum single plates with a certain aging characteristic, and obtain the influence model of the certain aging characteristic as Y(n)=k according to the least square method. n *T+b n , where n represents the type of aging characteristics, Y(n) is the pressure difference, T is the aging characteristic difference, and k n is the slope, b n is the intercept, and then the influence model of each aging characteristic is obtained.

4. The early warning method for the aging degree of curtain wall aluminum veneer according to claim 3 is characterized in that: Step S300 includes: obtaining the influence weight of the nth aging characteristic according to the slope in each influence model where k n is the slope of the nth influence model, and N is the total number of influence models; furthermore, according to the average RGB value, average grayscale value, and average depth value of the pixels in a certain pixel window, and based on step S200, obtain each aging characteristic difference and corresponding pressure difference corresponding to the pixel window, and perform normalization, and then obtain the aging value of the pixel window as follows: Y n is the nth pressure difference; set the first aging degree threshold D1 and the second aging degree threshold D2, where 0 < D1 < D2. If D1 < D ≤ D2, determine that a certain pixel window is a first aging window. If D2 < D, determine that a certain pixel window is a second aging window.

5. The early warning method for aging degree of curtain wall aluminum veneer according to claim 1 is characterized in that: Step S400 includes: Step S410: Establish a two-dimensional coordinate system of the aluminum veneer image to be detected, and obtain the center coordinates of each pixel window therein; randomly obtain a certain aging window, the aging window type is L, obtain all pixel windows in the eight fields around the certain aging window, and use them as eight-field windows. If the number of eight-field windows with the window type of L is greater than a preset number threshold, the window type of all eight-field windows is L, and the aging value remains unchanged, and the certain aging window and all eight-field windows are marked; if the number is not greater than the number threshold, randomly obtain another aging window, and repeat this step S410; Step S420: The center coordinate of the aging window is taken as P0, the aging value is D0, and the center coordinate of the g-th eight-domain window is taken as P g , the aging value is D g , if D0<D g , starting from P0 and pointing to P g Direction, size |D g -D0|, get the vector If D0>D g , starting from P0 and pointing to P g The opposite direction of |D g -D0|, get the vector Then, according to each eight-domain window, all vectors are obtained, and the vectors are added to obtain the pointing vector V0 corresponding to a certain pixel window; then the center coordinates closest to the end point of the pointing vector V0, the corresponding pixel window WIN, are obtained; if the pixel window WIN is not an aging window, according to step S400, an unmarked aging window and the surrounding eight-domain windows are randomly obtained; if the pixel window WIN is an aging window, the eight-domain windows around the pixel window WIN are obtained, and then all the aging windows are marked, and the final multiple adjacent aging windows of the same type are regarded as an aging area.

6. An early warning system for the aging degree of aluminum veneer for curtain wall, used to execute the early warning method for the aging degree of aluminum veneer for curtain wall as claimed in any one of claims 1 to 5, characterized in that: The system includes a pixel window division module, an impact model establishment module, an aging window screening module and an aging window merging module; Pixel window division module: used to obtain the aluminum veneer image to be detected, and perform preprocessing, and divide the aluminum veneer image to be detected into multiple pixel windows according to each pixel point in the aluminum veneer image to be detected; Impact model building module: used to obtain non-aging aluminum veneers of the same specifications as the aluminum veneers to be tested, as well as aluminum veneers with various aging characteristics, and perform performance tests on them respectively; and analyze the pre-shot images of non-aging aluminum veneers and images of aluminum veneers with various aging characteristics, and establish an impact model of various aging characteristics on the aging of aluminum veneers; Aging window screening module: used to obtain the influence weight corresponding to each aging characteristic according to the influence model, obtain the pixel points in each pixel window, obtain the aging value of each pixel window according to the influence weight, and screen out the first aging window and the second aging window; Aging window merging module: used to merge the aging windows according to the first aging window and the second aging window, obtain the first aging area and the second aging area in the aluminum veneer to be detected, and then provide early warning prompts and processing for the aluminum veneer to be detected.

7. The early warning system for aging degree of aluminum veneer for curtain wall according to claim 6 is characterized in that: The pixel window division module includes a pixel window setting unit, a window to be changed determination unit and a pixel window division unit; Pixel window setting unit: used to capture the image of the aluminum veneer to be inspected and convert it into a square grayscale image. The pixel window is composed of several pixels. The minimum pixel window, the maximum pixel window, the characteristic window and the window to be changed are set. The window to be changed determining unit is used to perform a degradation process on the grayscale value of each pixel in the grayscale image, thereby obtaining the entropy value of the feature window and the entropy value of the window to be changed, thereby obtaining the actual size of the window to be changed; Pixel window division unit: used to obtain all the windows to be changed, extend the four sides of each window to be changed to the edge of the grayscale image, and use the minimum area surrounded by the extension lines as the pixel window, thereby obtaining all the pixel windows.

8. The early warning system for aging degree of aluminum veneer for curtain wall according to claim 6 is characterized in that: The aging window merging module includes an eight-domain window acquisition unit and an aging window merging unit; Eight-domain window acquisition unit: used to establish a two-dimensional coordinate system of the aluminum veneer image to be detected, and obtain the center coordinates of each pixel window therein; randomly obtain a certain aging window and all the surrounding eight-domain windows; And mark the pixel window; Aging window merging unit: used to obtain a vector corresponding to a certain aging window according to a certain aging window and an eight-domain window; then obtain the pixel window WIN, and make judgments until all aging windows are marked.

Images